SEMI-ANNUAL
REPORT
(for
July
- December
1998),
Contract
Number
N_'3_
OCEAN OBSERVATIONS
WITH EOS/MODIS:
Algorithm
Development
and Post
Launch
Studies
Howard
R. Gordon
University
of Miami
I
Department
of Physics
Coral Gables,
FL 33124
(Submitted
January
1, 1999)
P
.¥
/'
/'_/
r
Semi-Annual
Report
(1
July
-
31
December
1998)
NAS5-31363
Preamble
As in earlier
reports,
we will
continue
to break
our effort
into six distinct
units:
•
Atmospheric
Correction
Algorithm
Development
•
Whitecap
Correction
Algorithm
• In-water
Radiance
Distribution
•
Residual
Instrument
Polarization
• Pre-launch/Post-launch
Atmospheric
Correction
Validation
• Detached
Coccolith
Algorithm
and
Post-launch
Studies
This
separation
has been
logical
thus
far; however,
as launch
of AM-1
approaches,
it must
be rec-
ognized
that
many
of these
activities
will shift emphasis
from
algorithm
development
to validation.
For example,
the
second,
third,
and
fifth
bullets
will
become
almost
totally
validation-focussed
activities
in the
post-launch
era,
providing
the core
of our experimental
validation
effort.
Work
under
the first
bullet
will
continue
into
the post-launch
time
frame,
driven
in part
by algorithm
_
deficiencies
revealed
as a result
of validation
activities.
An addition
to this
report
is a description
of our planned
activities
for FY99
(Appendix
1).
Our
next
Semi-Annual
Report
will address
the progress
made
on this plan.
Semi-Annual Report (1 July-
31 December
1998)
NAS5-31363
Abstract
Significant
accomplishments
made
during
the present
reporting
period:
•
Installed
spectral
optimization
algorithm
in the
SeaI)as
image
processing
environment
and
successfully
processed
SeaWiFS
imagery.
The
results
were superior
to the standard
SeaWiFS
algorithm
(the
MODIS
prototype).
in a turbid
atmosphere
offthe
US East
Coast,
but similar in a clear (typical)
oceanic
atmosphere.
Inverted
ACE-2
LIDAR
measurements
coupled
with sun photometer-derived
aerosol
optical
thickness
to
obtain
the
vertical
profile
of
aerosol
optical
thickness.
The
profile
was
validated
with
simultaneous
aircraft
measure-
ments.
Obtained
LIDAR
and
CIMEL
measurements
of typical
maritime
and
min-
eral dust-dominated
marine
atmosphere
in the
U.S.
Virgin
Islands.
Con-
temporaneous
SeaWiFS
imagery
were also
acquired.
£
Semi-Annual
Report
(1
July
-
31
December
1998)
NAS5-31363
L
1. Atmospheric
Correction
Algorithm
Development.
a.
Task
Objectives:
During
CY 1998
there
are seven
objectives
under
this
task.
Objectives
(i)
and (ii)
below
are
considered
to be the most
critical.
If the work planned
under
objective
(i)
is successful,
a module
that
enables
the algorithm
to distinguish
between
weakly-
and strongly-absorbing
aerosols
will be
includedin the atmospheric correctionalgorithm.
"
-
(i)We willcontinuethe study ofthe "spectraloptimization"algorithm.The initialrealization
ofthe algorithmwillbe toprovidea flagthat willsignaltheprobablepresenceofabsorbingaerosols,
and indicatethat the qualityof the derivedproducts cannot be assured. Later realizationswill
provide an atmospheric correctionin the presenceof absorbingas wellas nonabsorbing aerosols.
(ii)We
need to testthe basicimplementation of the MODIS
atmospheric correctionalgorithm
with actualocean colorimagery. We
willdo thiswith SeaWiFS
imagery.
0ii)We
must implement our strategyfor adding the cirruscloudcorrectioninto the existing
atmospheric correctionalgorithm. Specificissuesinclude (1) the phase function to be used for
the cirrusclouds,(2) the detailsof making two passes through the correctionalgorithm, and (3)
preparationof the requiredtables.These issueswillbe addressed as time permits in CY
1998.
(iv)The basic correctionalgorithm yieldsthe product of the diffusetransmittance and the
water-leavingreflectance.However, we have shown that the transmittancedepends on the angular
distributionofthe reflectanceonly when the pigment concentrationisvery low and then only in the
blue. We
need to develop a model to includethe effectsofthe subsurfaceBRDF
for,low-pigment
waters in the blue.
(v)We
need to study the efficacyofthe presentatmospheric correctionalgorithm forremoval
of the aerosoleffectfrom the measurement
ofthe fluorescenceline height.
(vi)We
need to examine
methods
for efficiently
includingearth-curvatureeffectsinto the
atmospheric correctionalgorithm.This willmost likelybe a modificationof the look-up tablesfor
the top-of-the'-atmosphere
contributionfrom Rayleigh scattering.
(vii)We
willexamine the necessityofimplementing out-of-bandcorrectionsto MODIS.
I
Semi-Annual
Report
(1
July-3i
December
1998)
NAS5'31363
b.
Work
Accomplished:
i
..
(i)
We consider
this
task
to be one
of our most
important
atmospheric
correction
activities
of
1998
[the other
is item
(fi)
above:
testing
MODIS
algorithms
with
SeaWiFS
_imagery],
and
as
such,
the
major
part
of our effort
on atmospheric
correction
will be focussed
on it.
We have
implemented
the
spectral
optimization
algorithm
described
by
Chomko
and
Gordon
lit.
Chomko
and
H.it.
Gordon,
Atmospheric
correction
of ocean
color
imagery:
Use
of the, Junge
power-law
aerosol
size distribution
with variable
refractive
index
to handle
aerosol
absorption,
Applied
Optics,
37,
5560-5572
(1998)],
in
which
power-law
size
distributions
are utilized,
in the
SEADAS
image
processing
environment.
The
power-law
distribution
allows
a straightforward
interpolation
to size
distributions
that
are not part
of the candidate
set.
We also
interpolate
on the real and
imaginary
parts
of the complex
refractive
index.
Thus,
a complete
spectrum
of models
can be generated
from
a relatively
small
candidate
set.
We then
use
standard
optimization
techniques
to find
the
best
fitting
set of parameters.
Although
incomplete,
the first test
of the spectral
optimization
algorithm
(S OA) is v'ery encour-
aging.
The figure
on the next
page
compares
the performance
of the
SOA wRh
the NASA
standard
SeaWiFS
algorithm
(NSSA)
over the Middle
Atlantic
Bight
(MAB)
for days
279
and
281 of 1997.
The
atmosphere
over the
MAB
was very
clear
on day
281 and
the NSSA
and
SOA
yielded
com-
parable
phytoplankton
pigment
concentrations.
In contrast,
on day
279 the
atmosphere
over the
MAB
was very tttrbid,
and
there were significant
differences
between
theNSSA-
and
SOA-derived
pigment
concentrations.
Since
the pigment
concentration
is unlikely
to undergo
gross
changes
in
the MAB
in two
days
at this
time of the year,
and since
tlle SOA pigment
concentrations
are
con-
sistent
for the two
days,
while
the NSSA
concentrations
are
not,
these
images
suggest
a superior
performance
for the
SOA.
We are in the process
of preparing
a detailed
report
c)f _these results
and
will included
it in the next
semi-annual
report.
We started
a major
effort
to understand
the optical
properties
of desert
dust
transported
over
the oceans.
This is being
carried
out
in conjunction
with
It. Evans.
Cyril Moulin
has
started
as a
postdoc
on the project
in August,
and is working
nearly
full time
on this problem.
The plan is to
use SeaWiFS
imagery
from the Tropical
Atlantic
acquired
this summer,
along
with the resets
from
the Virgin
Islands
field
experiment
(see
5.b.ii,
and
Appendix
2),
to delineate
the
dust properties.
In addition,
we are using
imagery
from
the Mediterranean
(closer
to the dust
source)
in conjunction
with
LIDAIt
data
acquired
in Rome.
r_
(ii)
We are acquiring
SeaWiFS
imagery
on a regular
basis
and,
with
It. Evans,
prepared
an
end-to-end
test
of the performance
of the MODIS
algorithm
in its present
state.
To effect
this
we
created
a set
of SeaWiFS-specific
LUTs,
but
in
a format
required
by
the
MODIS
code.
Evans'
Semi-Annual
Report
(1
July
-
31
December
1998)
NAS5-31363
group has reformated SeaWiFS
imagery intothe MODIS
format,and thus,we can testthe MODIS
codes usingSeaWIFS-simulated MODIS
data. Thus farthe testshave been successful,i.e.,MODIS
code running SeaWiFS
data in the MODIS
format reproduceclwellthe SeaWiFS
code processing
SeaWiFS
data.
:
(iii)None. In the lightof the successof our spectralOptimizationalgorithm,we may
have
to make
significant
modificationsin our originalstrategy.This task has been put .onhold to free
resourcesforexamination of task (i).The issueswillbe addressed during CY
1999 with the goal
of having a post-launchimplementation strategyin placeduring CY
1999.
(iv) We
continued the development of an oceanicBRDF
model
SpecificaUy,the magnitude
of the Raman
component
has been computed as a functionofthe pigment concentration.
(v) No work was CarriedOut on thistask.
(vi) No work was carriedout on thistask.
•_
(vii)Now
that we have the MODIS
relativespectralresponse(RSR) functions,We have started
to incorporatethem intothe algorithmsfollowingthe proceduresdescribedby Gordon
(1995) ["Re-
mote sensingof ocean color:a methodology for dealingwith broad spectralbands and significant
out-of-band response", Applied Optics,34 8363-8374 (1995)].
We
have computed
the band-
Table I: Band-averaged quantitiesneeded to compute the
Rayleigh reflectanceand the Ozone transmittancefor
the MODIS
bands.
A
Band
(nm)
(i)
412
8
0.3167
443
9
0.2377
488
10
0.1610
531
11
0.1135
551
12
0.0999
667
13
0.0446
678
14
0.0417
748
15
0.0286
869
16
0.0156
mW/cm'pm
sr
(xlO00)
170.37
186.50
191.82
188.57'
187.16
154.15
149.88
128.07
97.30
1.47
3.78
22.21
65.66
83.22
48.69
39.95
12.02
3.75
averaged quantitiesrequired
to compute
the Rayleigh reflectance[(r_(A))FoS,and (F0(A))s,(See
.
.
.
Semi-Annual
Report
(1
July
-
31
December
1998)
NAS5-31363
Gordon (1995) forthe notation)]and the Ozone transmittance[(hoz(A))F0S_].These are provided
in Table
1.
In addition,
we examined
the influence
of the water vapor absorption
bands
on the computation
of the Rayleigh
reflectance.
For MODIS,
the error in ignoring
water
vapor
(up
to a concentration
of 3.3 g_cm 2) is a maximum
of 0.25%
(for
Band
15)i
For the
other
spectral
bands_
the
error is
0.1%.
In contrast,
for SeaWiFS
the maximum
error is 0.55%.
c.
Data/Analysis/Interpretation:
See item
b above.
d.
Anticipated
Future
Actions:
(i)We
willcontinuetestingthe new spectraloptimizationalgorithmusing SeaWiFS
imagery.
In particular,we
need _toevaluateitwith more imagery and differentaerosoltypes. Also, we
need to understand the reason for the dramatic imProvement
of the SOA
over the NSSA
in the
figurefollowingpage 4, along with understanding why the performance ofthe NSSA
(the MODIS
prototype)isso poor in thiscase.
(ii)We
shallcontinue testingtheprototype MODIS
algorithm with SeaWiFS
imagery until
MODIS
imagery becomes available.
(iii)None. The cirruscloudissueinthe presenceofour "spectraloptimization"method needs
to be explored.We
willresolvethe "spectraloptimization"questionsfirst.
(iv) An ocean BRDF
model isbeing testedby comparison with experimental data obtained
at the MOBY
siteand during MOCE-4.
This testingwillContinueintoCY
1999. (See 5 below).
(v)
None.
(vi) None.
(vii)We
will derive the functionsneeded to incorporatethe out-of-band influenceon the
aerosolcomponent
ofthe atmospheric correctionalgorithm (Gordon, 1995),and provideLUT's for
Rayleigh reflectancethat are specificto the MODIS
band characteristics.
f.
Publications:
R. Chovako and H.R. Gordon, Atmospheric correctionof ocean colorimagery: Use of.the Junge
power-law aerosolsizedistributionwith variablerefractiveindex to handle aerosolabsorption,
Applied Optics,37, 5560-5572 (1998).
.
Semi-Annual
Report
(1 July
-
31
December
1998)
NAS5-31363
H.R.
Gordon,
Contribution
of raman
scattering
to water-leaving
radiance:
A reexamination.
(Sub-
mitred
to Applied
Optics.)
Semi-Annual
Report
(1
July-
31
December
1998).NAS5-31363
2. Whitecap
Correction
Algorithm
(with
K.J. Voss).
As the basic
objectives
of the experimental
portion
of this task have
been
realized
(acquiring
whitecap
radiometric
data
at sea), exPerimental
work is being
suspended
until the validation
phase,
except
insofar
as the radiometer
is being
operated
at sea when
it is sufHciehtly
important
to do so,
e.g.,
the SeaWiFS
Initialization
Cruise
(MOCE-4).
Our goal
is to maintain
experience
in operating
and
maintaining
the
insfrumentation
in preparation
for the validation
phase
of the
contract.
In
addition,
we need
to reanalyze
the
Tropical
Pacific
whitecap
data
because
Of the
surprisingly
low
reflectance
increase
due to whitecaps
that
we measured
there.
This
is a unique
data
set,
as it was
acquired
in the trade
winds
with moderately
high
winds
(8-12
m/s)
and
practically
unlimited
fetch
and
duration.
This
will better
bound
the limits
of oceanic
whitecap
reflectance.
a.
Near-term
Objectives:
. Operate
the radiometer
at sea to maintain
experience
in preparation
for the validation
phase.
Reanalyze
data
acquired
during
the Tropical
Pacific
cruise.
b.
Task
Progress:
A
strategy
has
been
developed
that
we
believewill
improve
the
analysis
of
the
whitecap
data;
however,
the postdoc
identified
for the reanalysis
has accepted
a position
elsewhere
(due
to
uncertainties
in MODIS
funding),
so no progress
has been
made on whitecap
data
analysis
of either
the Tropical
Pacific
or MOCE-4
cruises.
c.
Data/Analysis/Interpretation:
See item
b above.
Z.
d.
Anticipated
Future
Actions:
We will begin
the reanalysis
of the Tropical
Pacific
data
with
the goal of submitting
a revised
manuscript
on whitecap
reflectance
by the end of the summer.
e.
Publications:
None.
9
Semi-Annual Report (1 July-
31 Decemberi1998) NAS5-31363
3. In-waterRadiance
Distribution
(with
K.J. Voss).
a.
Task
Objectives:
The
main
objective
in this
task
is to obtain
upwelling
radiance
distribution
data
at sea
for
a variety
of solar
zenith
angles
to understand
how
the _water-leaving
radiance
varies
with
viewing
angle
and
sun
angle.
This
is the experimental
component
of our BtLDF
modeling.
b.
Work
accomplished:
No new measurements
were made
of oceanic
BI_DF's
during
this reporting
period.
c.
Data/AnalysisfInterpretation:
None
(but
see 1 .b.(iv)).
d.
Anticipated
future
actions:
We will be participating
in INDOEX
during
January
to April
of 1999,
as wellas
the MODIS
initalization
cruise.
Oceanic
BRDF's
will
be
included
in
our
stdte
of measurements.
We
will
concentrate
on data
analysis
after INDOEX.
e.
Problems/Corrective
actions:
None.
f.
Publications:
None.
10
.
II
_Li
-L
;
•_UOhI :suoDe:n.lqu
d
"5
_
ouo/_
:suo:_a v
OADaa.x.mD/Smalqo.xd
"a
:-
"_lq_ .I_A_
amo_aq
_aq_
uaq_
'_l_PO m uo!_a_oa
uo!_z!_iod
_._sN_
aq_ o_.
_p
_!x!_!swos-uo!_z!_tod
,T,SDIAI/S_S
aq_ _m_od.zo_m
s_ su,rema_
_q_
IIV
._aldmo_
,¢lTe_s_q
Mou
s_ _s_
s.nt,T,
• "
.),
'
•
-,,
"
.
.
:suoDav
a_a_neI
pa_edlaDuv
"p
•auolq
:uoDe_aadaa_ui/s!s,(l_uV/e_(i
"a
"auolq
:poqstldmoaa
V
_IaOA& "q
_
i'i
"
,
.auol q :saADaarq
0
_lse.T' .u
/,s_jSa
asay_
•oj _aa_oa
a,_ tn_a _oq
pwe 'uo!_aazzoa
a.t_aqdsom_
_o_ m'q_Ho_l'a aq_ jo aam_maoj_ad
aq_ ap_ap
s.rq__._
cpam
txoq
,zosuas
a_i_._o so!_s!za_o_tp
£_!_,!_!suas-uo!_z!zeIod
a_[_ _a_t_
'aou-e.rp_z _uap_.au!
aq_ jo a_s
uo!_z!z_Iod
jo a_s
aq_ o_ spuodsaz
SICIOIAI aq_ _.
:s! azaq
uoDsanb
_!s=q
aq_
•uoBez!JelOd
luotunJlsul
ICnp!soH
"lz
1_9fIl_-ggVl_I
(966I
.xaqmaaa(I
I£
-
Xlnf
I)
_;toda_I
lunuuv-.ttuaS
Semi-Annual Report (1 July - 31 December 1998) NAS5-31363
5. Pre-launclVPost-launch
Atmospheric
Correction
Validation
(with
K.J. Voss).
a.
Task
Objectives:
The long-term
objectives
of this task
are four-fold:
'_:
(i)
First,
we need to study
aerosol
optical
properties
over the oceans
to assess
the applicability
of the aerosol
models
used in the atmospheric
correction
algorithm.
Effecting
this required
obtaining
long-term
time
series
of the aerosol
optical
properties
in typical
maritime
environments.
This
was
achieved
using
a CIMEL
sun/sky
radiometer.
This
radiometer
is identical
to those
used
in the
AERONET
Network
(inwhich
we are a participant).
(ii)
Second,
we must
be able to measure
the a_rosol
optical
properties
from
aship
during
the
initialization/calibration/validation
cruises.
The
CIMEL-type
instrumentation
could
not
be used
(due
to
the motion
of the
ship)
for this
purpose.
The
required
instrumentation
consisted
of an
all-sky
camera
(which
can measure
the entire
sky radiance,
With the exception
of the solar aureole
region)
from a moving
ship,
an aureole
camera
(specifically
designed
for ship use)
and a hand-held
sun photometer.
_.....
In the case of strongly-absorbing
aerosols,
we have
shown
that
knowledge
of the aerosol
verti-
cal structure
is critical.
Thus,
we need
to be able
to measure
the vertical
distribution
of aerosols
during
validation
exercises
as well as to build a climatology
of the vertical
distribution
of absorbing
aerosols.
This
is accomplished
with
a LIDAR
system,
which
we
have
modified
for
ship
opera-
tions.
This
LIDAR
is also
needed
to
detect
the
presence
(or
ab_s_uce) of thin
cirrus
during
the
initialization/calibration/validation
cruises.
(ill)
The third
objective
was to determine
how'accu.rately
the radiance
at the top
of the atmo-
sphere
can be determined
based
on measurements
of sky radiance
and
aerosol
optical
thickness
at
the sea surface.
This required
a critical
examination
of the effect of radi.ative
transfer
on "vicarious"
calibration
exercises.
(iv)
The forth
objective
is to utilize
data from other
sensors
that
have
achieved
orbit
(OCTS,
POLDER,
SeaWiFS
...)
to validate
and
fine-tune
the correction
algorithm.
=_
12
Semi-Annual Report (1 July-
31 December 1998) NAS5-31363
b. Work Accomplished:
(i)We
have been operating the CIMEL
instrument in the Dry Tortugas continuouslyduring
most of 1998. Ithas worked well(even survivingHurricane Georges). These data are avaliableas
part of AERONET,
and have been used by othersto study the validityof aerosolretrievalswith
SeaWiFS,
e.g.,Wang
et al.,"Remote
sensingof the aerosolopticalthicknessfrom SeaWiFS
in
comparison with insitumeasurements," Submitted to _PLS99,
18-22Jan.,1999,Meribel,France.
(ii)The
Micro Pulse Lidar (MPL)
along with a CIMEL
were deployed in the U.S. Virgin
Islandsduring June and July 1998 to tryto observethe verticaldistributionand opticalproperties
of Saharan dust. Successfulobservationswere made
during both dus_tand dust-freeperiods. A
reportdescribingthe experiment isprovidedinAppendix 2. We
had tohave the MPL
unitrepaired
(itfailedafterthe Virgin Islandsexperiment) and have procured a spare power supply to ensure
againstsimilarfailuresin the winter-springINDOEX
expe_hnent. The unit isnow back in Miami
being prepared forINDOEX.
Aureole and sky camera data acquiredduring the MOCE-4
cruiseare stillbeing reduced. A
paper describingour aureolecamera has been prepared and submitted forpublicationto JAOT.
It
isincludedas Appendix
3.
, o
(iii)The theoreticalaspectsofthiswork have been coihpleted.The next phase isto use surface
measurements to predicttop-of-atmosphereradiance.
(iv) We
have prepared a duplicateversionofthe MODIS
algorithm code to use the SeaWiFS
spectralbands. This isbeing used totestthe MODIS
codewith SeaWiFS
data. (See Section1-ii.)
c.
Data/Analysls/Interpretatlon:
(i)A paper based on a long-term study of aerosolsover the ocean has been submitted to JGR
and isincluded here as Appendix
4. It confirms the notion that high aerosolopticalthicknesses
over the TropicalAtlanticv.ndCaribbean are almost always due to mineral dust.
(ii)Preliminary analysisofour MPL
data from the U.S. VirginIslandsdeployment isprovided
in Appendix 2.
Analysis of the data resultingfrom our participationin ACE-2
(June-July i997)'isnearly
complete, and severalpublicationsare being prepared. Our main
contributionwas in providing
calibratedMPL
and CIMF, L data to delineatethe verticaldistributionof the aerosoland the
opticalproperties.Appendix
5 isa draftofa :papercomp_ing
our MPL
predictionofthe vertical
13
. r_.
- •
Semi-Annual
Report
(1
July
= 31
December
1998)
NAS5-31363
distribution
of the
aerosol
optical
depth
with
direct
aircraft
measurements.
It is being
submitted
to a special
issue
of Tellu8
devoted
to ACE-2.
As far as .we know,i_this
is the first
such
successful
comparison.
It shows
thatlthe
profile of optical
depth
can be retrieved
from the MPL
given
surface
optical
depth
measurements,
e.g.,
by a sun
photometer.
d.
Anticipated
Future
Actions:
(i)
We will continue
to keep the CIMEL
oPerating
in the Dry Tortugas,
including
the monthly
maintenance
checks.
"
_
•
"
(ii)
Our
main
focus
will
be our participation
in INDOEX.
Reduction
of the
data
from
the
Virgin
Islands,
as well as MOCE-4
will be suspended
until
the end of INDOEX.
(i!i)
We attempt
to use data acquired
during MOCE-4
(once
th e analysis
is complete)
to effect
the vicariouscalibrationSeaWiFS
Band
8 (865 nm).
(iv) See Section1.d.(ii)
e. Problems/correctly
e actions:
None.
,
-
.
_
"
f.
Publications:
....
::
'
.
A. Smirnov,
B. Holben,
I. Slutsker,
E.J.
Welton,
and
P. Formenti;
"Optical
properties
of Saharan
dust
during
ACE.2,"
Jour.
Geophys.
Res.,
103D
28,079-28,092
(1998).
E.J.
Welton,
K.J.
Voss,
D.L.
Savoie,
and J.M.
Prospero,
"Measurements
of Aerosol
Optical
Depth
over
the North
Atlantic
Ocean:
Correlations
with Surface
Aerosol
Concentrations"
(Submitted
to
.]'our. Geophys.
Res.).
J.M.
Ritter
and
K.J.
Voss,
" A new
instrument
to measure
ttie
solar
aureole
from
an
unstable
,L
platform."
(Submitted
to Jour.
Atmos.
Ocean.
Tech.).
B. Schmid, P.B. Russell,J.M. Livingston,S.Gasso, D.A. Hegg, D.R. Collins,R.C. Flagah,J.H. Se-
infeld,E. Ostrom, K.J. Noone, P.A. Durkee, H.H. Jonsson,E.J.Welton, K.JI Voss,H_R_:Gordon, P.
Formenti, M.O. Andreae, V.N. Kapustin, T.S. Bates,and P.K. Quinn, "Clear column closurestud-
iesofurban-marine and mineral-dustaerosolsusingaircraft,
ship,and ground-based measurements
.inACE-2," (Submitted to ALPS99,
18-22 January 1999,Meribel,France).
B. Schmid, J.M. Livingston,P.B. Russell,P.A. Durkee, H.H. Jonsson, D.R. Collins,R.C. Flagan,
J.H. Sehtfeld,
S.A.
Gasso,D.A.
Hegg,
E. Ostrom,K.J.
Noone,
E.•J. Welton,
K.J. Voss,
H.R. Gordon,
14
_r
'
Semi-Annual
Report
(1
July
-
31
December
1998)
NAS5-31363
?
P. Formenti,
and M.O.
Andreae,
"Clear
sky closure
studies
of lower
tropospheric
aerosol
and water
vapor during ACE-2
using
airborne
sunphotometer,
airborne
in-situ,
space-borne,
and ground-based
measurements."
(To be submitted
to Tellu8
Special
Issue
on ACE-2)
E.J.
Welton,
K.J.
Voss,
H.R.
Gordon,
H. Maring,
A_ smirnov,
s. Holben,
B. Schmid,
J.M.
Liv-
ingston,
P.B. Russell,
P.A.
Durkee,
P. Formenti,
and M.O.-Andreae,
"Ground-based
Lidar Measure-
ments
of Aerosols
During
ACE-2:
Instrument
Description,
Results,
and
Comparisons
with
other
Ground-based
and
Airborne
Measurements."
(To be submitted
to Tellu8
Special
Issue
on ACE-2)
x
_-
•
L
_
_
,
•
J!
.
• Semi-Annual
R'eport
(1
July
-
31 December
1998)
NAS5-31363
6.
Detached
Coccolith
Algorithm
and
Post
Launch
Studies
(W.M. Balch).
a.
Task
Objectives:
T
Our
MODIS
work
during involvesunderstanding allaspectsof the influenceof suspended
calcium carbonate particleson inherentopticalpropertiesin the sea. Work
during thisreporting
period focused on severalareas:processingcellcount sampleS;merging atomic absorptiondata,
particulateorganiccarbon data,and scatteringdata from MODIS
pre-launch cruisesin the Gulf
of Maine; preparation,submission,or revisionof threeMODIS-related manuscripts;participating
in another MODIS
pre-launchcruiseto the Gulfof Maine to examine in situpropertiesof calcite
particle;and analysisofcalciteopticalpropertiesfrom the Ar_ibiansea.
The algorithm forretrievalofthe detached coccolithconcentrationfrom the coccolithophorid,
E. huxleyiisdescribedin detailinour ATBD.
The key isquantificationofthe backscatteringcoef-
ficientofthe detached coccoliths.Our earlierstudiesfocusedon laboratoryculturesto understand
factorsaffectingthe calcite-specific
backscatteringcoei_cient.A
thorough understanding of the
relationshipbetween calciteabmidance and lightscattering,in situ,Willprovide the basisfor a
genericsuspended calcitealgorithm.As with algorithmsforchlorophyll,and primary productivity,
the naturalvariancebetween growth relatedparameters and opticalpropert'ies
needs to be under-
stood before the accuracy of the algorithm can be determined.To
thisend, the objectivesof our
coccolithstudiesduring thisreportingperiod have been:
(1) Working
up data from our June '98MODIS
Gulf of Maine cruise.
(2) Final pre-publicationformatting of JGR
m_muscript on coccolith
scatteringproperties.
_
(3) Continued microscope cell/coccolith
counts forlatestsamples from
the Gulf of Maine.
(4) Final publicationof a paper on coccolithopticalproperties(with
Ken Voss and Katherine Kilpatrick.
(5) Completion of11 one-day pre-launchcruisesinthe Gulf OfMaine in '
which samples forcoccolithconcentration,suspended calcium car-
bonate and opticalpropertiesoftheseparticulateswere enumerated.
For perspectiveon the directionsofour work, we providean overview ofour previousactivities:
Semi-Annual
Report
(I July - 31
DeCember
1998) NAS5-31363
Jan-J_une 1995: Research focus - chemostat cultures(in whim
algal
growth rate was preciselycontrolled)and"we examiziedhow the optical
propertiesofthese calcifyingalgaechanged as a functi°n°fgrowth.
July-Dec 1995: Research focus - shipbbard measurements
of suspended
calciteand estimatesofopticalbackscatteringas validationofthe labo-
ratory.measurements.We
participated.0ntwo month-long cruisesto the
Arabian Sea;measuring coccolithophoreabundance, production,and op-
ticalproperties.
Jan-June 1996: Research focus- fieldc_cite distributions,
during two
Gulf ofMaine cruises,one in March and one in June. '
:_
July-Dec 1996: Research focus - participatedon anothercruise to the
Gulf ofMaine and processedsamples from the Gulf of Maine.
Jan-June 1997: Research focus- continued prbcessingsamples from our
previouscruises,upgraded our laserlighi scatteringpl10t0meterused in
allOfthe calcitescatteringmeasurements, performed another pre-launch
cruiseon calciteparticleopticsin the Gulf of Maine, and analyzed our
resultsfrom the MODIS-funded
flow cytometer work,
July-December 1997: Research focus
continuedbuildingour data base
on calcite-dependentscatteringwith a cruiseto the Gulf of Maine
in
November
1997. Work
was alsoperformed on
processingthe data from
the June 1997 cruise.The resultsfromlthe flow cytometer work were
submitted forpublication.
"."
Jan-June 1998: Research focus - worked-up resultsfrom our Novem-
ber '97 Cruise,participatedin a pre-launchcruisein June '98,finalized
data merging and processingforallprev_iouscruises,integrated.thedata
stream for surfaceradiance into the suiteof opticalparameters which•
we measure [which was one ofthe issuesfrom our MODIS
ATBD
review
"
that was suggested forconnectingthe inherentopticalproperties(which
we measure) to the water-leavingradiance (which the satellite
willmea- ; -,i',
sure)].Lastly,we put considerabletime intopreparationand publishing
ofearlierresults.
Semi-Annual
Report
(1
July
-
31
December
1998)NAS5-31363
• b.
July-December
1998:
We completed
work-up
of our June
1998
cruise
in tl_e Gulf
of Maine
aboard
the
tt/V
Alba-
tross.
As with
the
November
1997
cruise,
calcite-dependent
backscattering
commonly
accounted
for 10-20%
of t0tal
backscattering.
Even
with
the low light:concUtions,
coccolithophores
were
still
remarkably
abundant
in the
Gulf of Maine.
We formerly
interpreted
strong
coccolith
influence
on
particle
optics
as a "summer-only"
issue
but
it now
can be clearly
considered
a year-round
phe-
nomenon.
Specifically,
we encountered
a coccolith
patch
in 8°C
water
SW of Nova
Scotia.
This is
the last
place
we ever
expected
to see coccoHthophores,
given
the strong
mixing
and
cool
temper-
atures.
Since
we collected
AC-9
data
of absorption
and
attenuation
(which
by
difference
gave
us
scattering),
then
we have
been
processing
the data
to calculate
bb (the backscattering
probability
bb/b)
and see how this
varies
with
the standing
stock
of chlorophyll
and
calcium
carbonate.
i
.i
f',
Given
the future
plan for more Gulf of Maine
cruises,
as:weH as our SI1VIBIOS activities,
we have
written
new software
to merge
the various
data sets into
one file that
can be submitted
to NASA's
SeaBass
data
archive.
The
software
(which
performs
vicarious
calibrations,
offset
corrections,
plus
calculating
"derived
quantities"
has now been
through
3 sets_of revisions
to streamline
it and
make
it more
efficient.
This
software
will
considerably
reduce
the
time
currently
required
to
quality
control
the numbers
and
produce
hydrographic
plots
of the _inherent
optical
properties
(a,
b, c, bb,
calcite-dependent
backscattering,
bb, temperature,
s_ilnlty,:and
fluorescence).
c.
Data/Analysis/Interpretation:
See b.'
:
d.
Anticipated
Future
Actions:
Work in the next
reportingperiod
will adc_ess
several/areas:,
_•
(1)
Suspended
calcite
analyses
from the
Gulf of Maine
cruise
series.
(2)
Publication
of
JGR
manuscript
on
coccolith
scattering
properties
: (January
1999 issue).
'
(3)
Completion
of a cruise
in the
Gulf
of Mexico,
doing
a line between
Tampa
and
Progresso,
Mexico,
followed
by a trip up the entire
east-
ern U.S.,
back
to the
Gulf of Maine
(March
and
April
!999).
This
will be a one month
campaign.
(4)
Continuation
of microscope
cell/coccolith-counts
for latest
samples
from
the Gulf of Maine,
and
Gulf
of Mexico.
18
i__:_i
•
Semi-Annual
Report
(1 July-31
December
1998)
NAS5-31363
(5)
Utilization
of 20 more
days 'at
sea in the Gulf
of Maine
during
the
summer
and fall of 1999.
e.
Problems/Corrective
Actions:
None
f.
Publications:
Voss,
K., W.
M. Balch,
and K. A. Kilpatriek,
"Scattering
and
attenuation
properties
of Emiliania
huxleyi
cells
and
their
detached
eoeeoliths,"
Limnol.
Oeeanoor.
43
870-876
(1998).
Balch,
William
M., David
T. Drapeau,
Terry
L. Cucci,
and
Robert
D. Vaillancourt,
Katherine
A.
Kilpatrick,
Jennifer
J. Fritz.
"Optical
backscattering
by calcifying
algae-Separating
the contribution
by particulate
inorganic
and
organic
carbon
fractions,"
Jour.
Geoph_s.
Res.
(In press).
Milliman,
J.,
P.J.
Troy,
W.
Balch,
A.K.
Adams,'Y.-H.
Li,
and
F.T.
MacKenzie,
"Biologically-
mediated
dissolution
of calcium
carbonate
above
the
chemical
lysocline?"
Deep
Sea
Res.
(In
press).
Graziano,
L., W.
Balch,
D.
Drapeau,
B. Bowler,
and S. Dunford,
'iOrganic
and
inorganic
carbon
production
in the
Gulf of. Maine"
Cont.
Shelf
Res.
(Submitted).
.
Balch_
W.
M.,
D.
Drapeau,
J. Fritz,
and
B.
Bowler,
"Calcification
rates
in
the
Arabian
SeaP,"
Submitted
to
Deep
Sea Res.
Special
Issue
on the Arabian
Sea (Submitted).
T
--
19
Semi-Annual Report
(1
July
-
31
December
1998)
NAS5-31363
7. Other
Activities.
The PI participated
in the MODIS
Science
Team meeting
at University
of Maryland
(December
15 and
16, 1998).
He also presented
the following
papers
at scientific
meetings.
H.R.
Gordon,
K.J.
Voss,
J.W.
Brown,
P.V.F.
Banzon,
R.E.
Evans,
D.K.
Clark,
L. Kovar,
M. Yuen,
M.
Feinholz,
and
M.
Yarbrough,
SeaWiFS
Calibrat'ion
Initaliz'ation:
Pr_ary
Results.
Ocean
Optics
XIV,
Kona,
Hawaii,
November
11-13,
1998.
H.R.
Gordon,
The NIMBUS-7
Coastal
Zone Color Scanner:
A Retrospective.
Invited
paper,
Amer-
ical
Geophysical
Union
Fall Meeting,
San Francisco,
CA,
December
6-10,
1998.
;
.'
•
-
r
f-
20
-
Semi-Annual
Report
(1
July
-31
December
1998)
NAS5-31363
8. Publications
and Submissions
1998.
H.R.
Gordon,
Vicarious
calibration
of bcean
color
sensors,
Remote
Sensing
of Environment,
63,
265-278(1998).
H. Yang
and H.R, Gordon,
Retrieval
of the Columnar
Aerosol
Phase
Function
and Single
Scattering
Albedo
from
Sky Radiance
over Land:
Simulations,
Applied
Optics,
37,
978-997
(1998).
K.D.
Moore,
K.J.
Voss,.
and
H.R.
Gordon,
Spectral
reflectance
of whitecaps:
Instrumentation,
calibration,
and
performance
in coastal
waters,
Jour.
Atmos.
"'Ocean.
Tech.,
15,496-509
(1998).
D.J.
Diner,
J.C.
Beckert,
T.H.
Reilly,
C.J.
Bruegge , J.E.
Conel,
R.A.
Kahn,
J.V.
Martonchik,
T.P.
Ackerman,
R.
Davies_
G.A.W.
Gerstl,
H.R.
Gordon,.
J-P.
Muller,
R.
Myneni,
P.J.
Sellers,
B. Pinty,
and
M:M.
Vestraete,
Multi-angle
Imaging
SpectroRadiometer
(MISR)
Instrument
De-
scription
and
Experiment
Overview,
IEEE
Transactions
on
Geoscience
and
Remote
Sensing,
36,
1072-1087
(1998).
J.V.
Martonchik,
D.J.
Diner,
R.A.
Katm,
T.P. Ackerman,
M.M.
Verstrate,
B. Pinty,
and H.R.
Gordon,
Techni.'ques for the Reetrieval
of Aerosol
Properties
Over Land
and
Ocean
Using
Multiangle
Imaging_'
IEEE
Transactions
on Geoscience
and'Remote
Sensing,
36, 1212-1227
(1998).
W.E.
Esaias,M.R. Abbott, Otis B. Brown, J.W. Campbel, K.L. Carder, D.K. Clark,R.L. Evans,
F.E. Hoge, H.R. Gordon, W.M.
Balch; R. Letelier,and P. Minnett, An
overview of MODIS
capa-
bilities
forocean scienceobservations,IEEE
Transactionson Geoscience and Remote
Sensing,36,
1250-1265
(1998i.
J.V. Martonchik, D.J. Diner, B. Pinty,M.M.
Verstrate,R.B. Myneni, Y. Knyazikhim, and H.R.
Gordon, 'DeterminationofLand and Ocean Reflective,
Radiative,and BiophysicalPropertiesUsing
MultiangleImaging, IEEE
Transactionson GeoseienceandRemote
Sensing,36, 1266-1281 (1998).
R. Chomko
and Ii.R.Gordon, Atmospheric correctionof ocean :colorimageryi Use of the Junge
power-law aerosolsizedistributionwith variablerefractiveindex to handle aerosolabsorption,
Applied Optics,37, 5560-5572 (1998).
Voss,K., W. M. Balch, and K. A. Kilpatrick,"Scatteringand attenuationpropertiesofEmiliania
huxleyicellsand theirdetached coccoliths,"Limnol. Oceanogr. 43 870-876 (1998).
A. Smirnov, B. H01ben ,I.S1utsker,E.J. Welton, and P. Formenti, "Opticalpropertiesof Saharan
dust during ACE
2," Jour. Geophys. Res.,103D
28,079-28;092"(1998).
o
21
Semi-Annual
Report
(1 July
-
31
December11998)
_NAS5-31363
Balch,
William
M.,
David
T. Dral_ean,
Terry
L. Cucci,
and
Robert
D. Vaillancourt,
Katherine
A.
Kilpatrick,
Jennifer
J. Fritz.
"Optical
backscattering
by calcifying
algae-Separating
the contribution
by partic_ate
inorganic
and
organic
carbon
fractions,"
Jour.
Geophys.
Res.
(In press).
Milliman,
J.,
P.J.
Troy,
W.
Balch,
A.K.
Adams,
Y.-H.
Li,
and
F.T.
MacKenzie,
"Biologically-
mediated
dissolution
of calcium
carbonate
above
the
chemical
lysocline?"
Deep
Sea
Res.
(In
press).
-_-_
_
H.R.
Gordon,
Contribution
of raman
scattering
to water-leaving
radiance:
A reexamination.
(Sub-
mitted
to Applied
Optics.)
_
o
E.J.
Welton,
K.J.
Voss,
D:L.
Savoie,
and
J.M.
Prospero,
"Measurements
of Aerosol
Optical
Depth
over the North
Atlantic
Ocean:
,Correlations
with
Surface
Aerosol
Concentrations"
(Submitted
to
Jour.
Geophys.
Res.)_
B. Schmid,
P.B. Russell,
J.M.
Livingston,
S. Gasso,
D.A.
Hegg,
D_R. Collins,
K.C.
Flagan,
J.H.
Se-
infeld,
E. Ostrom,
K._J. Noone,
P.A.
Durkee,
H.H. Jonsson,
E.J. W_iton,
K.J.
Voss, H.R.
Gordon,
P.
Formenti,
M.O.
Andreae,
V.N.
Kapustin,
T.S.
Bates,
and P.K.
Quinn,
"Clear
column
closure
stud-
ies of urban-marine
and mineral-dust
aerosols
using aircraft,
ship,
and ground-based
measurements
in ACE-2,"
(Submitted
to ALPS99,
18-22
January
1999,
Meribel,
France).
J.M.
Bitter
and
K.J.
Voss,
" A new
instrument
to measure
the
solar
aureole
from
an
unstable
platform."
(Submitte
d to Jour:
Atmos.
Ocean.
Tech.).
Graziano,
L., W. Balch,
D. Drapeau,
B. Bowler,
and
S. Dunford,
"Organic
and
inorganic
carbon
production
in the Gulf
of Maine"
Cont.
Shelf
Res.
(Submitted)..
Balch,
W.
M.,
D.
Drapeau,
J.
Fritz,
and
B.
Bowler,
"Calcification
rates
in
the
Arabian
Sea,"
Submitted
to Deep
Sea Res." Special
Issue
on the Arabian
Sea (Submitted).
r
B.
Schmid,
J.M.
Livingston,
P.B.
Russell,
P.A.
Durkee,
H.H.
Jonsson,
D.R.
Collins,
R.C.
Flagan,
J.H.
Seinfeld,
S.A.
Gasso,
D.A.
Hegg,
E. Ostrom,
K.J. Noone,
E.J. Welton,
K.J. Voss,
H.R.
Gordon,
P. Formenti,
and M.O.
Andreae,
"Clear
sky closure
studies
of lower
tropospheric
aerosol
and water
vapor
during
ACE-2
Using airborne
sunphotometer,
airborne
in-sitU,
Space-borne,
and ground-based
measurements."
(To be submitted
to Tellus
Special
Issue
on ACE-2)
E.J.
Welton,
K.J.
Voss,
H.tt.
Gordon,
H. Mating,
A.
Smirnov,
B.
Holben,
B.
Schmid:i
J.M.
Liv-
ingston,
P.B. Russell,
P.A. Durkee,
P. Formenti,
and M.O.
Andreae,
"Ground:based
Lidar Measure-
_
22
,.."
.
L.
,i
(8"_[OV
uo anss I I_!_ads
_nlla _ o_ pa_.mlqns
aq O.T.) ,,'s_uamaaaS_al/g
au.toq .a_
pu_
p_s_q-puno_
D
_q_o
y_!_
suos!_dmoo_pm_
's_ins_i
'uo!id!_s_(I
_u_m'_su
I :g'_IOY
_m.a'n(I
SlOSO_V jo s_u_m
gO¢Ig-_SVlq
(g66I
aoqmo_o(I
Ig
-£lnr
_)
_aodo_I
l_nuuv-.maoS
Plans
for
FY
99
NASA/GSFC
NAS5-31363•
H.R.
Gordon
12/15/98
I
:
APPENDIX
1
!
NASA/GSFC
Contract
No.
NAS5-31363
OCEAN
OBSERVATIONS
WITH
EOS/MODIS
Algorithm
Development
and
Post
Launch
Studies
-
t
Howard
R. Gordon
:
University
of Miami
Department
of Physics
Coral
Gables,
FL 33124
•
k
Plans
for
FY
99
xL
•
Plans for FY 99
NASA/GSFC
NAS5-31363
H.R. Gordon
12/15/98
Preamble
The
coming
Fiscal
Year
(1999)
was to be heavily
focused
on validation
of MODIS-derived
products.
However,
the delay
of the launch
of EOS
AM1
requires
some
modification
of
the plan.
Our approach
for the coming
year
is to use SeaWiFS
for validating
algorithms
(rather
than
MODIS-derived
products)
in a manner
similar
to the way the MODIS
Land
and
Atmospheres
Groups
have
already
been
using
the MAS.
In addition,
the delay
in the
launch
will
allow
us additional
time
to enhance
the already
existing
algorithms
prior
to
launch.
We
break
our effort
for FY
99 into five distinct
(although
interrelated)
units:
•
Atmospheric
Correction
Algorithm
Development
•
Whitecap
Correction
Algorithm
•
In-water
Radiance
Distribution
(BRDF)
•
" Preqaunch/Post-launch
Atmospheric
Correction
Validation
*
Detached
Coccolith
Algorithm
In what
follows,
it will
be seen
that
addressing
the various
tasks
within
these
units
requires
fundamental
studies,
a through
examination
of SeaWiFS
imagerY,
use of SeaWiFS
imagery
as a tool,
and
a maintenance
of measurement
and data-analysis
capability
through
the prelaunch
era into the formal
validation
phase.
Developing
the required
capability
of,
and experience
in, processing
SeaWiFS
imagery
will also
facilitate
the efficient
handling
of
MODIS
imagery
(e.g.,
for QA)
in the post-launch
era.
Atmospheric
correction
Algorithm
Development
Objectives
and
Proposed
Activities
(FY
'99)
During
FY '99
there are several
objectives
under this task.
They
are focused
on ensuring
the correctness
of the basic
correction
algorithm
and its:implementation,
improving
the
basic
algorithm
to provide
for an accurate
correction
in the presence
of strongly-absorbing
aerosols,
and addressing
several
MODIS-specific
issues.
1.
ValidateImprove
the Basic
MODIS
Algorithm
using
SeaWiFS
• "
We will
begin
the algorithm
validation
and
improvement
process
using
SeaWiFS
imagery.
As we have
been
able
to successfully
process
SeaWiFS
imagery
_(reformatted
to MODIS)
using
our MODIS
code,
we can be assured
that
the implementation
is correct.
Thus,
we
will concentrate
on the science
of the correction
algorithm
by using
SeaWiFS
to validate
the
water-leaving
radiances
and indicate
any potential
problems
and potentialimprovements.
For
example,
we have
found
using
SeaWiFS
data acquired
during
the "initialization"
cruise
in January
and Febhaary
of 1998,
that using
the 670-865
nm bands
(the
"6-8"
algorithm)
2
Plans
for
FY
99
:NASA/GSFC
NAS5-31363
H.R.
Gordon
12/15/98
resulted
in a somewhat
better
atmospheric
correction
than
using
the 765-865
nm bands
(the
"7-8"
algorithm).
This
has important
implications
for MODIS,
because
it uses
an algorithm
that
is similar
to the 7-8 algorithm
in SeaWiFS.
However,
the SeaWiFS
band
at 765
nm
overlaps
the O 2 "A" absorption
band,
while
the corresponding
MODIS
band
(748
nm)
avoidsit.
Thus,
we need
to know
whether
the better
performance
of the 6-8 compared
to
7-8 algorithm
of SeaWiFS
is due to the O 2 "A" band,
or a more
fundamental
problem
that
could
have
an impact
on MODIS.
2.
Implement
and
Test
the "Spectral
Matching
Algorithm"
We will
continue
our basic
study
of the "spectral
matching"
algorithm.
Unlike
the basic
"MODIS
algorithm,
the spectral
matching
algorithm
has the potential
for
atmospheric
correction
in the presence
of strongly-absorbing
aerosols.
The
spectral
matching
algorithm
[Gordon
et al.,
"Remote
sensing
ocean
color
and aerosol
properties:
resolving
the issue
of aerosol
absorption,"
Applied
Optics,
36, 8670-8684
(1997);
Chomko
and Gordon,
"Atmospheric
correction
of 6cean
color
imagery:
Use
of
the Junge
power-law
aerosol
size distribution
with
variable
refractive
index
to handle
aerosol
absorption,"
Applied
Optics,
37, 5560-5572
(1998)]
has already
been
implemented
in an image
processing
environment.
We will test
its efficacy
and document
its
performance
(compared
to the basic
MODIS
algorithm)
using
SeaWiFS
imagery
in regions
in which
the aerosol
properties
may
be highly
Variable
ona
day-to-day
basis,
but
the water
properties
are reasonably
stable,
e.g.,
the Middle
Atlantic
Bight
(MAB)
or the Tropical
Atlantic
and Caribbean.
Our focus
will be as complete
an assessment
as possible,
so a
decision
can be made
concerning
its replacement
of, or inclusion
in, the basic
MODIS
algorithm.
At a minimum,
it will be used
to provide
a flag in the MODIS
algorithm
that
will signal
the probable
presence
of absorbing
aerosols,
and indicate
that
the quality
of the
derived
products
cannot
be assured..
However,
the long-term
goal
is that
it replace
the
basic
algOrithm.
3.
Optical
Properties
of Wind-Blown
Dust
We need
quantitative
estimates
of the optical
properties
of wind-blown
dust,
e.g., from
Africa,
to provide
a proper
model
to effect
atmospheric
correction
using
the standard
MODIS
algorithm.
The
Optical
properties
of wind-blown
dust
are required
to effect
atmospheric
correction
using
the standard
MODIS
algorithm,
and possibly
the spectral
matching
algorithm
as well.
We
have
a unique
opportunity
now
to use SeaWiFS
imagery
that
was acquired
over
the
Caribbean
this year's
particularly
intense
dust
periods
of July
and August,
simultaneous
with
our surface
measurements
of the aerosol
vertical
profile
(using
LIDAR)
and aerosol
optical
thickness
(using
a CIMEL
sun/sky
radiometer)
made;from
St. Johns,
VI,
to develop
a model
for African
dust.
4.
MODIS-Specific
Issues
;....
We needto
address
the
detection
and
removal
of thin
cirrus
clouds,
methods
for
efficiently
including
earth-curvature
effects,
out-of-band
corrections,
BRDF
effects
on the diffuse
3
Plans
for FY
99
NASA/GSFC
NAS5-31363
H.R.
Gordon
12/15/98
transmittance,
correct
for polarization
sensitivity
of MODIS,
an.d the efficacy
of
atmospheric
correction
for
removal
of the aerosol
effect from
the measurement
of the
fluorescence
line
height.
is'
Of the issues
listed
under
this heading,
the first
that
we will
pursue
is the MODIS
out-of-
band
corrections.
As we now
have
the MODIS
relative
spectral
response
(RSR)
functions
from
MCST,
we can
complete
their
incorporation
into the algorithms
following
the
procedures
described
by Gordon
(1995)
["Remote
sensing
of ocean
color:
a methodology
for dealing
with
broad
spectral
bands
and significant
out-of-band
response",
Applied
Optics,
34,
8363-8374
(1995)].
The
second
is incorporating
the SBRS/MCST
polarizati0n-sensitivity
data
into the
atmospheric
correction
module.
This
will be effected
as described
in Gordon,
et al., (1997)
["Atmospheric
Correction
of Ocean
Color
Sensors:
Analysis
of the Effects
of Residual
Instrument
Polarization
Sensitivity,"
Applied
Optics,
36,
6938-6948
(1997)].
The
third
issue
to be examined
is the BRDF
effect.
AS described
below
(In-water
Radiance
Distribution),
we propose
to use our measurements
of the BRDF
to develop
a model
that
can be applied
to MODIS
imagery.
This
model
will be used
to address
the BRDF
on the
diffuse
transmittance.
The
foRh
issue
in importance
is efficiently
including
eaRh-curvature
effects
in the MODIS
algorithm..
Following
Ding
and Gordon
["Atmospheric
correction
of ocean
color
sensors:
Effects
of earth
curvature,"
Applied
Optics,
33, 7096-7106
(1994)]
this will
most
likely
be
a modification
of the look-up
tables
for the top,of-the-atmosphere
contribution
from
Rayleigh
scattering.
, However,
before
actually
embarking
on an implementation,
we will
examine
SeaWiFSimagery
at high
latitudes
to assess
theimpact
of neglecting
earth-
curvature
in the algorithm.
Finally,
because
of Uncertainty
in the performance
of MODIS
Band
26, and because
we
need
to assess
whether
our radiative
transfer
codes
are _ufficiently
accurate
to study
removal
of the aerosol
effect
from
the measurement
of the fluorescence
line
height,
we will
examine
this issue
and thin cirrus
clouds
only
if time
permits.
An
updated
Algorithm
Theoretical
Basis
Document
(ATBD)
for
the
Normalized
Water
LeavingRadiance
algorithmwill
be
prepared
and
submitted
by
April
30,
1999.
_
Whitecap
Correction
Al_,orithm
Objectives
and
Proposed
Activities
(FY
i99)
Our basic
goal
for the rest of the project
is to maintain
experience
in operating
and
maintaining
the instrumentation
in preparation
for the validation
phase
of the contract,
and
to complete
the analysis
of the whitecap
data
acquired
thus
far.
5
Maintaining
Measurement
Capability
We need
to maintain
our ability
to make
whitecap
measurements.
Plans
for
FY99
NASA/GSFC
NAS5-31363
H.R.
Gordon
12/15/98
The basic
objectives
of the experimental
portion
of this task
has been
realized
(acquiring
whitecap
radiometric
data
at sea),
experimental
work
is_being
Stispended
until
the validation
phase,
except
insofar
as the radiometer
is being
operated
at sea when
it is sufficiently
important
to do so, e.g.,
it was
operated
during
the SeaWiFS
Initialization
Cruise
(MOCE-
4).
This
requires
personnel
capable
of both
maintaining
and operating
the instrumentation.
6.
Reduction
and
Analysis
of Existing
Data
We need
to complete
the analysis
of the data
already
acquired.
We
have
a significant
amount
of whitecap
data that
is yet to be reduced
and analyzed.
In
addition,
we'-need
to reanalyze
the Tropical
Pacific
whitecap
data because
of the
surprisingly
low
reflectance
increase
due to whitecaps
that
we measured
there.
This
is a
unique
data
set, as it was
acquired
in the trade
winds
with
moderately
high
winds
(8-12
m/s)
and practically
unlimited
fetch
and duration.
We have
now
developed
an alternative
method
of analysis
that
we believe
is more
robust
and will provide
greater
confidence
in the
results.
Earlier
this year,
the SeaWiFS
Project
informed
us that
the present
whitecap
algorithm
was causing
the atmospheric
correction
algorithm
to fail in the South
Atlantic.
We
provided
an algorithm
adjustment
(based
on our earlier
analysis
of the Tropical
Pacific
whitecap
data),
and were
informed
by them
that
the algorithm
appeared
to be working
much
better
after
the adjustment.
This
example
underscores
the importance
of a detailed
examination
of as much
SeaWiFS
imagery
as possible
prior
to the launch
of MODIS.
-_,
In-water
Radiance
Distribution
(BRDF)
"
Objectives
and
Proposed
Activities
(FY
'99)
During
FY
'99 our objectives
for this task are:
maintaining
experience
in operating
and
maintaining
the instrumentation
in preparation
for the validation
phase
of the contract;
acquiring
more
field
data;
and using
the field
data
to develop
a model
of the BRDF
as a
function
of the solar
zenith
angle
and the water's
chlorophyll
Concentration.
Z
Maintaining
Measurement
Capability
We need
to maintain
ourability
to make
BRDF
measurements.
This
requires
personnel
capable
of both
maintaining
and operating
the instrumentation,
as
well
as reducing
the data.
8.
Acquisition
of More
Field
Data
_
.
. _ .
-
To build
and
thoroughly
test a BRDF
model,
we need
to acquire
data
over
a wider
range
of
chlorophyll
concentrations.
""
We
will operate
the Radiance
Distribution
Camera
System
(RADS)
whenever
the
opportunity
tO acquire
data over
a wider
range
of chlorophyll
concentrations
presents
itself.
For
example;
we plan
to participate
in INDOEX
in the Indian
Ocean
in January-February
-.
1999.
This
Will also
provide
an independent
data
set for.validation
of the MODIS
algorithm
using
SeaWIFS.
We will,
of course,
participate
in the MODIS
initialization
cruise.
9.
Build
a BRDF
Model
5
jr
/
Plans
for
FY
99
NASA/GSFC
NAS5-31363
H.R.
Gordon
12/15/98
We need
to develop
a model
relating
the subsurface
radiance
distribution
(BRDF)
to the
chlorophyll
concentration
of the water
"
The
data
acquired
during
the SeaWiFS
initialization
cruise
showed
that,
at a solar
zenith
angle
of -37 ° (the
mean
encountered
during
the cruise)'and
a chlorophyll
concentration
of -
0.1 mg/m 3 (the approximate
mean
encountered
during
the cruise),
an error
of-5%
in the
normalized
water-leaving
radiance
[L,,(A)]u
would
be made
if the measurement
was carded
out
at nadir
rather
than
at the angle
appropriate
to the viewing
direction
of the sensor.
Since
most
investigators
are only
capable
of measuring
the in-water
Upwelling
radiance
at nadir,
direct
comparison
of their
measurements
with
MODIS
(or SeaWiFS)
data will
result
in an
error
at, or above,
the level
of error
that is acceptable
for the MODIS
product.
This
error
can be reduced
by using
a model
(based
on the chlorophyll
concentration
and solar
zenith
angle)
to either
correct
the nadir
measurement
to the appropriate
viewing
angle,
or to correct
the MODIS
product
to provide
the normalized
water-leaving
radiance
at nadir.
In either
case,
a model
is required
and we propose
to build
one that
includes
both
elastic
and.
inelastic
(Raman)
scattering,
as it is evident
that
a considerable
portion
of the radiance
in the
blue-green
(>10%)
results
from
inelastic
processes.
In addition
we will
compare
our
: measurements
with
the
f!
Q model
of Morel
[MOrel and Gentili,
"Diffuse
reflectance
of
oceanic
waters.
1TI. Implication
of bidirectionality
for the remote
sensing
problem,"
Applied
Optics,
35, 4850--4862
(1996)]
(which
is being
used
in other
models
in the
community).
Construction
of such
a model
maY require
that the in-water
light
field
be
:simulated
with
a radiative
transfer
code
that
includes
polarization.
Pre-launchfPost-launch
Atmospheric
Correction
Validation
Objectives
and
Proposed
Activities
(FY
'99)
The
original
objectives
of this task
were
fourfold:
(1) study
aerosol
optical
properties
over
the oceans
to assess
the applicability
of the aerosol
models
proposed
for use in the
atmospheric
correction
algorithm;
(2) measure
the aerosol
optical
properties
(including
their
vertical
distribution)
from
a ship
during
the initialization/calibration/validation
cruises;
(3)
determine
how
accurately
the radiance
at the top of the atmosphere
can be estimated
based
on measurements
of sky radiance
and aerosol
optical
thickness
at the sea surface
(i.e.,
vicarious
calibration);
and (4) utilize
data from
other
sensors
that
have
achieved
orbit
(OCTS,
POLDER,
SeaWiFS
...) to validate
and
fine-tune
the correction
algorithm.
We
• have
obtained
a significant
amount
of data
toward
(1),
designed;
constructed,
or
purchased,
instrumentation
to acquire
data for (2),
completed
(3),
and (4) was discussed
under
the first
activity
above
(Atmospheric
Correction
Algorithm
Development).
Objectives
for the post-launch
validation
phase
are: maintain
experience
in operating
and
maintaining
the instrumentation
in preparation
for the validation
phase
of the contr_t;
complete
analysis
of data
already
acquired;
and participate
in the validation
phase.
10.
Maintaining
Measurement
Capabili_
: ....
We need
to maintain
our ability
to make
atmospheric
measurements
at sea.
This
requires
personnel
capable
of both maintaining
and
operating
the instrumentation,
as
well
as reducing
the data.
We will continue
to keep
the CIMEL
operation
in the Dry
6
Plans for FY 99 NASA/GSFC
NAS5-31363
H.R. Gordon 12/15/98
Tortugas,includingthemonthlymaintenancechecks_We planto usetheMPL, aureole
camera,andskycameraduringINDOEX in January-Februaryof 1999,andduringthe
MODIS initializationcruise.
11.
Complete
Analysis
of Existing
Data
We need
to complete
the analysis
of the data
acquired
previously.
.
,4
We
have
been
operating
the CIMEL
instrument
in the Dry Tortugas
almost
continuously
for
several
years.
It worked
well,
largely
due to the diligence
of E.J.
Welton
in maintaining
the instrument
and the site.
However,
we have
only scratched
the surface
in the data
analysis.
We
have
only
looked
at a few •specific
time periods,
and
need
to examine
the
entire
record.
'in
particular,
there
are several
periods
when
African
dust
was known
to be
present,
and we can
study
its properties
using
the CIMEL
data.
Also,
we have
extracted
specificdays
of the data
set_ believed
to be dust,
marine
aerosol,
or nonseasalt
sulfate
aerosols.
On these
days
we are running
our inversion
method
[Wang
and Gordon,
"Retrieval
of the Columnar
Aerosol
Phase
Function
and Single
Scattering
Albedofrom
Sky
Radiance
over
the Ocean:
Simulations,"
Applied
Optics,
32, 4598-4609
(1993)]
for
recovering
the phase
function,
to compare
with
that used
by Nakajima
etaI.
["Retrieval
of
the Optical
Properties
of Aerosols
from
Aureole
and Extinction
Data,"
Applied
Optics,
22,
2951--2959
(1983)],
used
in the Aeronet
Network).
This
work
will
be continuing.
Aureole
and sky camera
data
acquired
during
the July Hawaii
criuse
are being
analyzed,
specifically
for several
locations
while
the cruise
went
near
the volcanic
plume
to look
at the
retrieved
size
distribution
of particulates
in the plume.
The aureole
and sky cameras
(including
the sky polarization)
were
also
operated
during
the MOCE-4
cruise,
and these
data
are in the process
of being
reduced
and analyzed
as part of
the SeaWiFS
initialization.
12.
Post-launch
Validation
We will participate
in MODIS
post-launch
validation.
We
will use
the MPL,
aureole
camera,
and sky camera
during
INDOEX
in January-
February
of 1999.
In this case,
SeaWiFS
will be a surrogate
for MODIS
in the algorithm
validation
program.
We
will also participate
in the MODIS
initialization
cruise;
however,
its scheduling
will be dependent
on the MODIS
launch.
Detached
Coccolith
Algorithm
Objectives
and
Proposed
Activities
(FY
'99)
The
objectives
of our FY 99 effort
are to complete
analysis
of cruise
work
done
to
date,
continue
new
field
data in the Gulf
of Maine
and Gulf
of Mexico,
use these
data to
improve
algorithm
performance
and to apply
and validate
the coccolithophore
algorithm
using
SeaWiFS
data.
13.
Processing
o fcompletedpre-launch
cruises
We have
been
collecting
data
on coccolith
concentrations
in several
pre-launch
cruises
per year
for the past
several
years.
Most
analyses
have
now
beenperformed,
and
post-cruise
calibrations
of the scattering
measurements
have
been
done
(calibration
• standards
are always
mn
before
each
cruise,
but periodic
distilled
water
checks
are always
run,
and must
be checked
against
published
values
to verify
calibration).
While
basic
7
APPEND
IX
2
Report:
U.S. Virgin
Islands
Summer
1.998 Research
Trip
Study of the optical properties
of Saharan dust over
_:
_the ocean
Period
Of Study:
7-15-98
to
8-12-98
Ellsworth
Judd Welton
Post-doctoral
Research
Associate
University
of Miami
- Physics
Department
1320
Campo
Sano
Dr.
;_
CoralGables,
F1 33146
•
(305)-284-2325
ext 3
1.
Introduction
I Conducted
a field
measurement
program
from
July
15,
1998
to August
12,
1998
at the
Virgin
Islands
Environmental
Resource
Station
(VIERS)at
Lameshur,
St.
John,
USVI.
The purpose
of the program
Was to study
the optical
propel'ties
of airborne
particles
in
the
atmosphere
_(aeroso!s).
Aerosols
directly
effect
the
climate
by
scattering
and
absorbing
sunlight
in the
atmosphere
and
indirectly
cause
climate
effects
by
serving
as
cloud
condensation
nuclei
[Russell
et al.,
1997].
Aerosols
also
interfere
with
the
remote
sensing
of the earth's
surface
from
satellite
sensors
because
the
signals
from
the
surface
must
pass
through
the aerosol
layers.
I am funded
to provide
information
on-the
optical
properties
of
aerosols
over
the
Ocean in order
to conect
satellite
images
of the ocean
color.
Ocean
color
measurements
are
made
to determine
biologically
important
ocean
parameters,
such
as
the
concentration
of
chlorophyli
a in a given
body_of
water:The
presence
of aerosol
layers
over
the water
must
be
accounted
for
when
analyzing
the satellite
data
to insure
an
accurate
analysis
is made
[Gordon,
1997].
In
particular,
I was
measuring
two
of
the
key
optical
properties
of
aerosols.
The
fLrst property
is
referred
to as
the
aerosol
optical
depth
(AOD)
and
is
a
measure
of
the
amount
of
direct
sunlight
attenuated
by
the
aerosol
layers.
The
spectral
dependence
of
the
AOD
thr0ughout.the
visible
and
near-infared
wavelengths
also
gives
information
on
the
size
distribution
of
the
aerosols
[Van
de
Hulst,
1981].
The
second
property
I was
measuring
was
the
vertical
profile
of
the
aerosol
extinction
coefficient
(AEC).
The
AF__
is the amount
of sunlight
scattered
and
absorbed
per
unit
length
by
the
aerosolsl
Knowledge
of
the. AEC
and
also
the
altitude_of
the
•aerosol
layers
is vital
for
correct
analysis
of the ocean
color
data
[Gordon,
1997].
"
The
.VIERS
site
was
chosen
because
it is on a small,
remote
isiandthat
is largely
free
of urban
influence
and
is one
of the few
easily
accessible
locations
for
the
study
of
aerosols
• over
the open
ocean_
In addition,
theprimary
aerosol
species
I was
interested
in
studying
was
Saharan
dust.
Saharan
dust
is transported
•across
the
Atlantic
Ocean
during
the summerand early fall months[Prospero,1995] and the Virgin Islands are ideally
situatedto study thedust as it moyesacrossthe Atlantic. The influx of Saharandust
perturbs
the regional
climates
of theCaribbean
and
the Southeastern
United
States.
Also,
dust
aerosols
are difficult
to incorporate
into
satellite
correction
analysis
since
there
is little
Kistoricfil
data
on theif_optical
properties.
In addition,
dust
aerosols
are
often
irregular
in
shape
and
normal
analysis
routines
that
make
use
of
spherical
particle
assumptions
for
optical
calculations
produce
inaccurate
results
[Mishchenko
et al.,
1997].
In-situ
data on the
real
optical
properties
of
dust
aerosols
over
the
ocean
is needed
for
the
validation
and
correction
analysis
of the oceancolor
measurements.
This
report
w_descfibe
the VIERS
site
and discuss
the instrumentation
I deployed
during
the measurement
program.
I will
also
present
data
that
identifies
the presence
of dust
aerosols
as
well
as
reasonabJy
clear
days
and
also
the altitude
of
the
respective
aerosol
layers.
Finally,
I.will
discuss
the measmecl
AOD
and AEC
profiles
of both
clear
and
dusty
periods
and
relate
them
tO data
taken
during
this
period
using
the NASA
SeaWiFS
ocean
color
sensor
[Hooker
et.al.,
1992]
currently
in orbit.
2.
Site
and
Instrumentation
Descriptions
VIERS
[VIERS,
www
site,
1998]
is located
at
18.32 ° N
latitude
and
64.73
° W
longitude
and includes
a marine
laboratory
located
on Great
Lameshur
Bay
and a living
area
(campgrotmd)
located
a short
distance
inland
from
the baYl The elevation
of the lab and
the
.
campground
varies
within
just
a few
meters
of
sea-level,
thus
for
the
purposes
of
this
program
all data
was
analyzed
assuming
an
altitude
of
0 km.
VIERS
is bordered
to the
north
(inland)
by tall hills
of maximum
altitude
approximately
380
m but
has
a relatively
unobstructed
view
(within
a few
meters
of sea-level)
to the
east and
southeast
(windward).
VIERS
is located
well
within
the boundaries
of
the
Virgin
Islands
National
Park
and
its
remote
location
insures
that little
to no urban
aerosols
are present_
2
I deployedthreeinstrumentsat VIERS during themeasurementprogram.The first
two instrumentsweresunphotometerswhichareusedto measuretheAOD andit's spectral
dependence.A hand-heldsutiphotometerwas used for the first two weeks, and an
automaticsky scanningsunphotometer(Cimel)wasusedfor theremainderof theprogram.
TheCimel [Holben.etal.,
1994]
was
not
functioning
during
the
first
two
weeks
of
the
program
due to problems
with
it's
motor.
_The Cimel
sunphotometer
is shown
in Figure
1
while
operating
on the VIERS
lab roof.
•
.
Figure
1
Figure
1. The
Cimel
sunphotometer
is shown
operating
on the roof
of
the VIERS
lab.
The
Cimel
is the gray
tube
pointing
skyward
from
the top oKthe
roof.-
•
3
•The
other
instrument
was
a micr0-pulse•
lidar
system
[Spinhirne
et al.,
1995]
and
is
shown
in
Figure
2
while
operating
at the
VIERS
campground.
The
lidar
was
used
to
acquire
the
ve/'tical
profiles
of ' the
AEC
and
AOD.
The
lidar
operated
throughout
the
program.
However,
a
problem
developed
during
the
first
week
and
resulted
in
condensation
forming
on
the
mirror"
surfaces
of
the
lidar
transmitter-receiver
(T-R).
.... :Massive
condensation,on
the mirrors
prevents
successful
operation
of the lidar,
therefore
a
2
lidar
data
was
only
taken
during
periods
relatively
free
of visible
mirror
condensation.
Figure
2
:i
!i::
Figure
2. The
lidar
system
is shown
operating
at 'the vIERs
campground.'.
_
'_
'
4
3.
Cimei
and
Lidar
Data
.
_
The
data
acquire
d with
the Cimel
was
processed
by
the
AERONET
[NASA-a,
www
site,
1998]
group
at NASA
Goddard
Space
Flight
Center
and
then
supplied
to me.
The Cimel
m e_asured the Sl_tral
dependence
Of the AOD
from
July
29,
1998
to August
8,
1998.
This
AOD
data
was
used
to identify
dust
dominant
periods
from
clean
(normally
only
sea-saltaerosols
) periods
using
criteria
I developed
in my
Ph.D.
dissertation
[Welton,
?::'
: ,
1998].
Analysis
of
the Cimel
AOD
data
as well
as a few
hand-held
sunphotometer
AOD
measurements
has shown.that
July
27,
August
6,
and
August
8 were
clean
days
free
of
dust,
while
J_y
29,
July
31,
and
August
1. were
days
.with
moderate
to
heavy
dust.
Consultation
of my notes
from
these
days
also
shows
both clean
and hazy
visual
conditions
onthe
appropriate
days.
_
The
lidar
data
acquired
on the clean
and
dusty
days
mentioned
above
was
analyzed
to
determine
the
vertical - prof'tle?0f
the
AEC
at
times
coincident
with
the
Cimel
measurement.
The
AOD
measured
with
the Cimel
was
used
to calculate
the AEC
prof'des
?
from
the lidar_ data
using
an algorithm
I had
developed
in my
Ph.D.
dissertation
[Welton,
1998].
The
best
coincident
Cimel
and
lidar
data
was
available
on August
1 for
the
dusty
days
and August
8 for the clean
days.
These
days
were
chosen
based
on overlapping
Cimel
and iidar
measurement
times
as well
as the absence
of cirrus
clouds
(identification
of cirrus
clouds
is possible
by _sPecti0n
of the lidar
data)
which
falsely
increase
the AOD.
The AEC
•
"
'
'2
pr0filesand
A0D
are shown
in Figure
3 for August
i and August
8.
•
-
?-
•
5
,..
Figure
3
5
4
!
I
I
I
I
0
0!05
0.1
0.15
0.2
0.25
AEC:
980801
@ 2045
[AOD:
0.317
BER:
0.214
woP(=):
2.685]
,,,,,,,,,,
AEC:
980808
@ 1345
[AOD:
0.088
BER:
0.111
woP(n):
1.395]
-,--,--,-
Rayleigh
Scattering
Cocff.
(1/kin)
,-
.
,"
Figure
3. AEC
profiles
for the August
1 dust period
(980801)
and the
August
8 clean
period
(980808).
Tile
profile
due to molecular
scattering
(Rayleigh)
is shown
for
comparison.
The
AOD
measured
on each
day is given
in the brackets.
The
AEC
values
obtained
during
the dust
period
on August
1 are much
greater
than
those
measured
during
clean
conditions
on August
8.
Also,
the August
1 AEC
values
are
the
same
magnitude
as dust
AEC
values
I obtained
with
the
lidar
during
the
Aerosol
Characterization
Experiment
2 (ACE-2)
in Tenefife,
Canary
Islands
in
1997.
The
Figure
also
shows
the clean
August
8 aerosols
to be
confined
to a narrow
layer
extending
to only
just
over
ikm
in altitude.
However,
during
the dusty
August
1 period
the aerosols
are
seen
to extend
tO 4 km in altitude.
Most
of the aerosols:are
located
in the boundary
layer
on this
day, but a clear
dust
layer
above
the boundary
layer
is evident.
The
AEC
values
(at
each
individual
altitude)
in each
profile
may
not
be
highly
accurate
based
on
the
backscatter-extinction
ratio
(BER)
calculated
for
each
profile.
The
6
'''-2
,
BER
valuesTor
both
the •clean and
dusty
profiles
are quite
high
and
this
is usually
caused
by incorrect
AOD
v_ues
input
to the
algorithm
or a BER
values
that
are not
constant
with
altitude.
The
Cimel
instrument
is highly
calibrated
at the
Goddard
facility
thus
the
AOD
values
arem0st
likely
correct.
The
best
assumption
is _at
the BER
was
not
constant
with
altitude
during
both
of these
profiles.
This
was probably
less
likely
with
the clean
profde
as
the BER
is much
closer
to values
that
Would
be
expected
for
spherical
sea-salt
particles.
Despite
possible
error
in
the
calculated
AEC
values,
the
altitude
of
the
aerosol
layers
remains
unaffected.
Thus
the altitude
range
of the clean
and
dusty
periods
is accurate
and
useful
information.
4. In-situ
Results
and
SeaWiFS
Measurements
SeaWiFS
data
was
acquired
on August
1 and August
8 for the area surrounding
and
including
the U.S.
Virgin
islands.
The SeaWiFS
data presented
here
was
provided
by the
NASA
Goddard
SeaWiFS
web
site
[NASA-b,
www
site,
1998].
Figures
4 and
5 show
composite
Visible
wavelength
satellite
images
taken
by SeaWiFS
over
the
VIERS
site
on
August
1 and
August
8 respectively.
Dust
is
visible
in
the
images
as the
brown
color,
clouds
appear
as white,
and
clear portions
of sky
are blue
due to the
ocean
surface.
Also,
sunglitter
from
the
ocean
surface
is
present
in
the
middle
portion
of
each
image
and
unfortunately
encompasses
the VIERS
site
in Figure
4.
However,
Figure
4 clearly
shows
the-VIERS-_ite
toN _.under
a large
dust
layer
on
August
1 that
extends
far
south
of
the
island
as much
ofthis
portion
of the image
is most
likely
dust
due
to its
hazy
distinctive
brown
color.
Conversely,
Figure
5 shows
_e
VIERS
Site to be under
clear
sky
save
for
clouds
to the northdast
and southwest;
dust
is visible
far'to
the south
of the site On this day.
..
Figure
4
2;e--
Q,
Figure
4.
SeaWiFS
visible
image
showing
the
VIERS
site
under
a large
dust
layer.
Dust
appears
brown,
clouds
white,
and
clear
sky
portions
are
blue
due
to
the
ocean
surface.
Stmglitter,
present
in the middle
portion
of the image,
appears
as a gray-like
color.
,L
8
+,
,:
' Figure
5
•
+
Figure
5.-SeaWiFS
visibie
image
on
August
8
showing
the
VIERS
site
under
clear
sky
3
although,
surrounded
by
clouds
to the northeast
and
southwest.
Dust
appears
brown,
clouds
white,
and
clear.
sky
pgrtions;
are
blue
due
to
the
ocean
surface.
SunglJtter,
present
in
the
middle
portion
of the image,
appears
as a gray-like
color.
..
9
5.
Conclusion
The data in-situ
data obtained
at the VIERS
site will be used
to help develop
a better
understanding
of the impact
of dust
aerosols
on the ocean
color
analysis
performed
on the
SeaWiFS
images
shown
above.
In particular,
the data
I gathered
at the VIERS
site
will
aid
in constructing
more
accurate
models
of dust
properties
for
use
in the analysis
of satellite
data.
Also,
the information
I gathered
on
the
background
sea-salt
aerosol
properties
will
add
to
my
previously
obtained
sea-saJt
data
and
aid
in
the
understanding
of
future
measurements
of aerosol
optical
properties
over
the ocean.
Finally,
this trip
has
shown
that
the VIERS
site
is weU suited
to the Study
of
Saharan
dust
aerosols
over
the
open
ocean.
The data
and experience
I acquired
during
this measurement
program
will
aid any
possible
future
aerosol
related
work
in the Virgin
Islands
area.
Acknowledgments
I would
like
to thank Randy
Brown
and Bobby
Agostini
at VIERS
for much
help
in
setting
up
my
equipment
during
the
program
and
for
making
my
stay
pleasant
and
enjoyable.
I would
also
like
to thank
Albert
Chapin
for
the design
and
construction
of the
lidar
climate
containers.
I
am
also
grateful
to
Brent
Holben
and
the
NASA-GSFC
AERONET
network
for
loaning
me the Cimel
sunphotometer
and performing
analysis
of
the Cimel
data.
I also
gratefully
acknowledge
the support
of the National
Park
Service
and
the Virgin
Islands
National
Park
for
allowing
me
toconduct
my
research.
Finally,
I am
thankful
to receive
support
from
the NASA
contract
#NA55-31363.
'.7
10
References
Gordon,
H.R.,
Atmospheric
correction
of ocean
color
imagery
in the Earth Observing
System
era,
J. Geophys.
Res.,
102,
17081-17106,
1997.
Holben,
B.N.,
T.F.
Eck,
I. Slutsker,
D.
Tanre,
J.P.
Buis,
A.
Setzer,
E.
Vermote,
J.A.
Reagan,
Y.J.
Kaufman,
T: Nakajima,
and
F. Lavenu,
Multi-band
automatic
sun
and
sky scanning
radiometer
system
for measurement
of aerosols,
Proc.
Sixth
International
Symposium
Physical
on Measurements
and
Signatures
in Remote
_Sensing,
75-83,
1994.
Mishchenko:,
M.i.,
L.D:. Travis,
R.A.
Kahn,
and R.A.
West,
Modeling
phase
functions
for duSt-like
tropospheric
aerosols
using
a shape
mixture
of randomly
oriented
polydisperse
spheroids,
J. Geophys.
Res.,
102,
16831-16847,
1997.
NASA-a,
www
site,
http:llspamer.gsfc.nasa.gov/,
1998.
NASA-b,
www
site,
http:llseawifs.gsfc.nasa.govlscriptslSEAWIFS.html,
1998.
Prospero,
J.M.,
Saharan
dust
transPort
over
the north
Atlantic
Ocean
and Mediterranean,
Chapter
inThe
Impact
of Desert
Dust
From
Northern
Africa
Across
the
Mediterranean,
Kluwer
Academic
Publishers,
Oristano,
Sardinia,
1995.
Russell,
P.B.,
S.A_ Kinne,
and R.W.
Bergstrom,
Aerosol
climate
effects:
Local
radiative
forcing
and
column
closure
experiments,
J. Geophys.
Res.,
102,
9397-9407,
1997.
Spinhirne,
J.D.,
J. Rail,
and v.s.
Scott,
Compact
eye-safe
lidar
systems,
Rev.
Laser
Eng.,
23,
26-32,
1995.
Van
de Hulst,
H. C.,
Light
Scattering
by Small
Particles,
Dover
Publications
Inc.,
New
York,
NY,
1981.
VIERS,
www
site,
http :llwWw.islands.orglvirginlviersl
, 1998.
Welton,
E. J., Measurements
of Aerosol
Optical
Properties
Over
the Ocean
Using
Sunphotornetry
and Lidar,
Ph.D.
Dissertation,
University
of Miami,
Physics
Department,
1320
Campo
Sano
Dr.,
Coral
Gables,
F1, 33146.
11
APPENDIX
3
A new
instrument
to measure
the solar
aureole
from
an unstable
platform
Joseph
M. Ritter
and Kenneth
J. Voss
Physics
Department,
University
of Miami
Coral
Gables,
F1. 33124
305
284
2323
ext 2
fax
305
284
4222
-.
vosst_physics.miami.edu
"i/
.÷.
Abstract
A new
instrument
is described
which
can
measure
the sky radiance
distribution
near
the
solar disk
even
when
operated
on an unstable
platform.
An
imaging
detector
is used
to image
the
sky
around
the sun,
while
an occulter
on a long
pole
blocks
out
the direct
solar
radiation.
The
occulter
has a Neutral
Density
filter
in the center
which
places
an
image
ofthe
solar
disk
in each
image.
This
allows
accurate
directional
information
to be
obtained,
eliminates
possible
ambiguities
about
the angular
position
of the data,
and
increases
the accuracy
of the measurement.
A special
triggering
device
aids
in making
these
measurements
on a ship.
This
instrument
will
be used
in field
studies
of
amaospheric
aerosols
and in satellite
validation
campaigns.
1.
Introduction
w
• Aerosols
"(solid
and
liquid
particles
suspended
in'the
atmosphere)
play
an
important
role
in determining
the
Earth's
radiation
budget
(and
hence
its
climate),
as
well
as
influencing
the chemical
composition
of the gaseous
atmosphere.
Aerosols
can
affect
the
Earth
radiation,
budget
both
directly
(by
scattering
and
absorbing
solar
and
terrestrial
radiation)
and
indirectly
(by
modifying
cloud
properties
through
their
role
as
cloud
condensation
nuclei).
Unfortunately
the global
impact
of the
aerosols
optical
effects
are
not
well
understood.
In fact,
the
uncertainty
in the aerosol
radiative
forcing
is larger
than
2
the uncertain!y
in climate
forcing
by all greenhouse
gases
released
over
the past
century
(Houghton,
et
al.
1995).
TO reduce
this
uncertainty,
improved
measurements
of
the
aerosol
optical
properties
are. needed
for aerosol
and climate
field
studies.
,.
in
addition,
aerosol
optical
properties
must
be
understood
to
interpret
satellite
observations
of
the
earth's
surface
since
much
of
the
signal
received
by
a satellite
is
caused
by the
intervening
aerosols.
Thus
to validate
remote
satellite
measurements
of
ocean
and
land
properties
we
need
to improve
our
understanding
of the
aerosols
optical
properties.
The
solar
aureole
is the region
of enhanced
brightness
surrounding
the
sun
caused
by
the
scattering
of light
by aerosols.
Radiometric
measurements
of the
solar
aureole,
when
combined
with
spectral
extinction
data,
can
yield
data
products
such
as the
aerosol
size
distribution
and
the
aerosol
scattering
phase
function
for
small
scattering
angles
(Nakajima
et al.
1983;
Nakajima
et al.
1996).
For
vicarious
calibration
of ocean
color
satellite
sensors
it is desirable
to make
these
measurements
at sea
(Clark
et al.
1997).
At
sea,
platform
motion
coupled
with
the
large
dynamic
range
of
the
aureole
radiance
requires
an instrument
with
a high
signal-to-noise
ratio,
accurate
pointing,
fast
triggering,
short
exposure
times,
low jitter,
as well
as sensor
stability.
An
imaging
radiometer
meeting
these
reqiJirements
has
been
developed
and
will
be
described.
The
imaging
nature
of
the
instrument
allows
measurement
of
sky
radiance
within
one
degree
of
the
center
of
the
solar
disk
even
when
deployed
on
an
unstable
platform
such _as a ship.
_
2.
Instrument
Description
Previous
successful
approaches
to
solar
aureole
radiometry
have
involved
scanning
•radiometers
on stable
platforms
(I-Iolben
et al.
1996).
A new
approach
was
required
for
shipborne
operation
where
platform
motion
interferes
with
the
scanning
motion.
Our
approach
uses
an electronic
camera
to acquire
the
circumsolar
sky
radiance
in a single
image.
Use
of
a camera
ensures
accurate
direction
registration,
while
modern
CCD
cameras
have
the dynamic
range
necessary
for
measuring
the aureole.
The
system
components
are
illustrated
in Fig.
1.
In
brief,
the
system
employs
a
cooled
5i2x512
CCDarray,:_35
mm
lens
and
filter
wheel,
internal
and
external
cooling
system,
and
sun
alignment
Sensor
with
associated
triggering
and
digitizing
electronics.
This
package
is _mounted
on a 2m
pole
and
pointed
manually
toward
the
sun
where
an
occulter
mounted
on
the
end
of the
pole
attenuates
the
direct
sunlight
and
completely
shadows
the
camera
aperture.
A
sun
alignment
sensor
on
the
system
automatically
exposes
the
CCD
when
the Camera
orientation
is correct.
Instrument
control
and
data
acquisition
is performed
by
a remote
desktop
computer.
components
wiii be described
in detail
At
this
time
the
instrument
a.
Occulter
::
The
camera
aperture
is,shadowed
by _the
solar
occulter
so
that
scattered
radiation
entering
the
aperture
is a miiaimum
of
1/2 deg_e
fromthe
instrument
optical
axis.
The
solar
disk
itselfsubtends
a half-angle
of
1/4 degree
at the instrument.
The
center
of the
occulter
is
a neutral
density
filter.
The
attenuated
image
of
the
solar
disk,
when
combined
with
ephemeris
data,
can
be usedto
determine
the
azimuth
and
zenith
coordinates
corresponding
tO each
CCD
pixel.
_
_
4
,,"
b. .Filter
Wheel
After
the occulter,
the
first
optical
element
the light
entering
the system
encounters
is
the
fil/er
wheel
housing
window.
This
window
is coated
with
an antireflection
coating,
and
has
a 1•cm clear
inside
diameter.
Inside
the filter
wheel
housing,
10 nm
interference
filters
centered
at 440,
560,
670,
and
860
nm are used
to lirnit
the spectral
bandpass
of the
instrument.
The
filter
wheel
and
housing
provides
an
effective
light
seal
with
its
neighboring
components
without
significantly
extending
the
optical
path.
Since
this
;system
is
only
looking
at a narrow
angular
range
around
the
sun,
the
variation
in
ban@ass
of the filters
with
angular
position
is negligible
(<1%).
Each
intefi'erence
filter
is paired
with
a neutral
density
filter
to optimally
utilize
the
available
d_,namic
range
of the
sensor.
Each
spectral
channel
response
is adjusted
for
near
saturation
at
1.0 degree
from
the center
of
the
solar
disk
with
the
exception
of the
440
nm
channel.
The
low
system
response
at
440
nm
would
have
required
longer
exposure
times
to
achieve
this
level
and
lead
to potential
blurring
effects
caused
by
platform
motion.
c.
Lens
After
passing
through
theinterference
and
neutral
density
filters
the
light
enters
a
camera
lens.
The
current
lens
System
is.a
standard
35mm
lens
with
a 50 mm
focal
length
• stoppedat
f/15.
With
this
lens,
the
instrument
has
an angular
resolution
of 0.05
degrees
and
a field
of view
of +/- %5 degrees
centered
on the sun.
d.
Shutter
An
electromechanical
shutter
is employed
in this
system.
A
computer
controlled
integration
and
shuttering
system
ensures
low
integration
time
jitter
(<10gs)
while
expos!ngrapidly enough(19 msec)to eliminate imageblur due to ship motion. The
uncertainty in the retrieval of absolute radiance data is directly proportional to the
uncertaintyin the time the shutteris open from one exposureto the next. Quantitative
assessmentof jitter, andothercalibrationissuesareaddressedin thenext section.
e.
Pointing
and
triggering
An essential
feature
for
shipbome
applications
is a unique
sun-sensing
trigger
device
J
that
trips
the
shutter
only
when
suitable
alignment
of the
sun
with
the
instrument
optical
axis
is achieved.
The
trigger
developed
for
this
instrument
consists
of
a long,
baffled
collimating
tube
with
a window
and
neutral
density
filter
at one
end,
and
a photodiode
amplifier
package
at the other.
The
axis
of the
tube
is aligned
parallel
to the optical
axis
of the aureolemeter.
When
the
tube
(and
thus
the
aureolemeter)
is aligned
with
the
sun,
the photodiode
is activated
and
a pulse
is generated.
The
pulse
is amplified
and
shaped,
then
used
to trigger
the Shutter.
This
scheme
allows
measurement
of the
aureole
to within
1 degree
of the
edge
of the
solar
disk
even
when
deployed
on an unstable
platform
such
as a ship.
f.
Detector
A
commercially
available
camera
(Spectrasource
MCD-1000)
is used
as the
image
detector.
The
CCD
array
used
is
a
512-by-512
array
(TK512).
Double
correlated
sampling
digitizes
individual:pixels
at
16 bit resolution.
This
system
provides
a signal
to
noise
ratio
of
103, with
dark
noise
variation
less
than
11 counts.
The
array
temperature
is
actively
servo-controlied
by Peltier
cooling.
An
external
cooling
system
was
constructed
to help
draw
additional
heat
from
the
instrument,
and
to ensure
sensor
stability
in field
deployments..
6
g.
Data
acquisition
and
control
A
Macintosh
deskto
p computer
is employed
to control
data
acquisition
as
well
as
primary
data
reduction.
Software
was
written
to provide
a menu
driven
user
interface,
and
control
image
digitization,
storage,
shuttering
as well
as perform
field
diagnostics.
Interrupt
routines
are
written
in native
assembly
language
to accommodate
strict
timing
requirements.
The
software
incorporates
a control
loop
allowing
complete
remote
system
operation
by theexperimenter,
usually
located
outdoors
with
the
sensor
50
to
100
feet
from
the
control
system.
Complementary
instruments
are
employed
to provide
optical
depth,
measurement
location,
ambient
pressure,
humidity
and
temperature
data.
3.
Instrument
Calibration
and
Characterization
There
are
several
calibration
steps
required
to
convert
the
camera
output
into
calibrated
sky radiance
data.
In addition,
the electro-optical
system
must
be characterized
to understand
the
performance
limitations
of the system.
These
steps
are now
detailed.
a. Angular
Calibration
Angular
Calibration
is required
to determine
the mapping
between
the
image
location
on
the array
focal
plane
and
the corresponding
angle
in real
space.
This
was
done
in two
ways.
In one
test
an object
was imaged
at a known
distance.
_ The
angular
subtence
of the
object
was
calculated,
and
this
was
used
to calibratetheangular
field
of the
image.
The
?
.
result
was
then
verified
by placing
the
instrument
On a precision
rotary
table,
imaging
a
point
source,
andthen
acquiring
an
image.
The
camera
was
then
rotated
by
a known
angle
and
another_image
acquired.
This
was
done
for
angles
from
0 to 7 degrees.
The
pixel
displacement.of
the light
source
in the image
Was then
determined
and
agreed
with
7
_,
thedisplacementexpectedfrom the first testto within 0.05degrees.Eachpixel coversa
0.05degreewide (full angle)field of view.
b.
Dark
Signal
There
are
several
components
of
the dark
signal
which
axe not
easily
separable.
An
experiment
was
performed
to determine
the
time
dependent
thermally
generated
dark
current
and its
impact
on signal
to noise
degradation.
A series
of dark
images
were
taken
with
exposure
times
ranging
from
19 msec
to 100
seconds.
For
each
image
the
average
value
from
i00
pixels
in the center
of the image
was
recorded.
The
slope
of the resulting
dark
signal
versus
time
indicates
that
approximately
8 thermal
counts
per
second
per
pixel
are
generated
with
the
sensor
at
the
nominal
sensor
operating
temperature
of
252
Kelvin.
The
normal
exposure
time
for a sky
image
is 19 msec,
thus
these
thermal
counts
become
significant
only
if the
CCD
is not
cleared
immediately
prior
to
acquiring
an
image.
Thus
we
chose
to continuously
Clock
out charge
from
the
chip
prior
to exposing
an
image.
There
is also
a position
dependent
dark
signal,
a result
of pixel-to-pixel
bias
variations
as well
as
local
well
fabrication
differences.
Immediately
following
each
sky
image
a dark
image
is acquired.
The
dark
image
is then
subtracted
off
of
the
sky
data
image
as the
first
step
in the data
processing.
In addition
to this
dark
noise,
readout
noise
from
the sampling
electronics
occurs
and is difficult
to independently
estimate
or separate
from
the
dark
noise_
All
of the
noise
factors
t0gether'result
ina
standard
deviation
per
pixel
of
11 counts...
c.
Linearity
Characterization
.Charge
Coupled
device
arrays
are
inherently
linear
devices
when
properly
biased.
Nevertheless_,
an
experiment
wasperformed
to determine
the
overall
system
response
$
J
i "
linearity.
The
camera
imaged
a diffusing
plaque
from
a constant
distance.
The
source
illuminating
the _diffuser
was
then
placed
at logarithmically
spaced
distances
for
each
set
of exposures
in order
to achieve
the
desired
plaque
irradiances.
An
image
data
set
was
acquired
at each
step
and
an average
value
from
the center
100 pixels
was
recorded.
The
results
are
shown
in
Fig.
2. The
linearity
of
output
signal
relative
to
input
light
was
verified
for 3 orders
of magnitude
of viewed
radiance.
d.
Absolute
Radiometric
Flux Calibration
In order
to
determine
the
absolute
spectral
response
of
each
CCD
pixel,
it
was
necessary
to first
determine
the absolute
transmission
curve
T(2)
for
each
interference
-
neutral
density
filter
pair.
A monochromator
(Optronics
740A)
configured
for
1 nm
resolution
was
used
to measure
the
filter
transmission.
The
measurement
extended
to
where
the
transmission
was
10 .4 of
the
peak
transmittance
on each
side
of
the
filter's
P
peak.
This
was
repeated
for
all 4 interference
filters.
The
interference
filters'
bandwidth
and
center
frequencies
are
summarized
in Table
1..
Following
this
the
system
was
set
to view
a nearly
lambertian
calibrated
Spectralon
plaque
at 45 degrees
off
normal.
The
plaque
was
illuminated
by a 1000
W standard
lamp
(b'EL,
spectral
irradiance
calibration
traceable
to NIST)..
This
provided
a known
source
of radiance:for
the
system.
Unfortunately,
this
laboratory
source
is much
dimmer
than
the
solar
aureole,
thus
we
Could
not
include:the
neutral
density
filters
in
this
measurement.
The
transmission
of the neutral
density
filters
were
determined
in the same
t
manner
as
the
interference
filters,
and
these
transmission
values
were
folded
into
the
calibration.
An
absolute
calibration
factor
was
then
determined
for
each
pixel,
and
a
matrix
was
constructed
such
that
when
applied
piXel-by-pixel
to image
data,
after
the
•-
9
dark
signal
removal,
it corrects
both for optical
rolloff
as well
as generating
absolute
radiance
values.
"4
,j.
10
e.
Absolute
Radiometric
Accuracy
and
Precision
Tests
were
performed
to quantify
absolute
radiometric
jitter,
i.e.
repeatability
or the
lack
the_of
between
measurements.
A plaque
was
illuminated
with
a steady
source
and
then
20 images
were
acquired.
After
applying
the
above
calibration
procedure
to each
image,
a 100
pixel
sample
near
the center
of each
image
was
selected
for
analysis.
The
results
show
that
the
absolute
jitter,
i.e.
radiometric
precision,
for
the
entire
• electromechanical
system
is on the order
of 0.1%.
The
absolute
calibration
of the
system
i_s limited
by the accuracy
of the standard
lamp
and
is assumed
to be on the order
of 3%.
f.
Optical
Beam
Spread
Function
Since
the
lens
is focussed
to infinity,
for
a perfect
lens,
all photons
incident
at one
angle
are mapped
to exactly
1 point
on the
array.
To test
our system
a helium-neon
laser
beam
was
expanded
to a diameter
larger
than
the camera
aperture
in order
to both
fill
the
aperture
and
to decrease
the
divergence
of
the
laser
beam.
This
beam
then
directly
illuminated
the camera
aperture
from
a series
of angles
(0,
1, 2, 3, 4, 5, 6 and
7 degrees).
At each
angle
an image
was
acquired.
These
measurements
showed
that
over
90%
of the
counts
were
contained
within
0.1 degrees
from
the center
of the spot.
4.
Data
Acquisition
and
Reduction
During
the
data
acquisition
sequence,
upon
alignment
with
the
sun,
the
instrument
digitizes
and
stores
a sky radiance
image,adar
k image,
and
housekeeping
data.
To
reduce
the
data,
a
series.of
automated
programs
were
written
in
Spyglass
Transform.
These
routines
act
on a
collection
of
raw
data
(image
and
dark)
files
to
remove
the
dark
signal.
The
resulting
images
areprocessed
to produce
image
arrays
where
pixel
values
are calibrated
radiance,
and
pixel
positions
correspond
to small
solid
11
angleelementsof the sky at eachpoint in the instrument'sfield of view. Following this
procedure,theuserthendecidesby visualinspectionwhich imagesmerit further analysis.
This stepis necessaryto filter for cloudcontaminationor measurementartifacts (field of
view obstruction,flaring, etc.). The nextstepin the analysisis completedby a seriesof
automatedimageanalysisprogramsdevelopedfor theMatlabdataanalysispackage.
The Matlab programsopensadatafile containingthecalibratedimage. Embeddedin
this file arecoordinatesof thepixel closestto the centerof thesun,while a separatedata
file containssolar air mass,thechannelwavelengthmapping,solarconstants,Rayleigh,
aerosolandozoneoptical depthsfor the dataset. The imageheaderis stripped and the
data transposedand stored in a format suitable for further analysis.
A similarly
dimensioned(512x512) matrixis generatedwhereeachpixel valuecorrespondsto theair
massat the correspondingcameraimagepixel. Finally a matrix is constructedwhere
each pixel value is the scattering angle at the correspondingdata image pixel.
An
examp!eradianceimageis shownin Figure3.
Due to diffraction el'fectsit is necessaryt° filter or "digitally occult" somepixels near
the imagecenter. The softwaredeterminesthe points which lie within a small user-
definable angle of the sun center and generatesa mask to zero data in the original
radianceimagearray. Theseoccultedpixelsareshownasthewhite circle in Fig. 3. This
maskis alsoappliedto thescatteringangleand airmassmatrices.
It is useful to examinethe imagesfor sky homogeneityby examining left / right
.
o
.
symmetry
of,radiance
vs.
scattering
angle.
A
typical
scatter
plot,
for
a good
data
set,
showing
thissymmetrY
is shown
in Fig.
4.
Afterdetermining
Symmetry,
it is necessary
to resample
the
flagged
almucantar
data
into
appropriately
sized
solid
angle
bins.
This
is
12
accomplishedby settinga user-definable(half-coneangle)bin size(usually0.25degrees)
and averagingover all datawithin that scatteringcone,within the appropriateair mass
limits and where the datahasnot been digitally occulted. The almucantarsampledis
shownasthe white bandinFig. 3.
An
example
plot
showing
the resampled
almucantor
radiance
for
the
solar
aureole
is
shown
in
Fig.
5.
Once
this
aureole
radiance
is
determined
it
can
be
combined
with
independent
measurements
of
the
aerosol
optical
depth
to determine
the
small
angle
scattering
phase
function,
and
a parameter
related
to
the slope
of the large
particle
size distribution.
As
an
example
derived
product,
Fig.
6 illustrates
several
aerosol
phase
functions
derived
from
the
aureole
measurements
(Ritter
1998).
These
are
derived
from
several
measurements
of
the
solar
aureole
performed
on
a single
day.
They
were
derived
assuming
single
scattering
and
that
the
aerosols
were
contained
in a single
layer
in the
atmosphere.
They
show
a small
Variation
in
the
small
angle
portion
of
the
aerosol
scattering
function
over
this period.
5.
Conclusions
A new
instrument
has
been
described
which
can
me_isure
the sky radiance
distribution
near
the solar
disk
even_when
operated
on an unstable
platform.
Use
of an imaging
detector,
and2the
presence
of an image
of the solar
disk
in each
image,
allow
accurate
directional
information
to be obtained.
This
eliminates
possible
ambiguities
about
the
angular
position
of the data,
and increases
the accuracyof
the measurement.
A special
13
• triggering
device
aids
in making
thesemeasurements
on a ship.
This
instrument
will
be
used
in field
studies
of atmospheric
aerosols
and
in satellite
validation
campaigns.
6.
Acknowledgements
The
authors
are grateful
for support
from
the NASA
Goddard
Space
Flight
Center
under
contacts
NAS5-31363.
We
wouldalso
like
to thank
Mr. Albert
Chapin
for
his
help
in machining
the
instrument.
References
Clark,
D.
K.,
H.
R.
Gordon,
K.
J.
Voss,
Y.
Ge,
W.
Broenkow,
and
C.
Trees,
1997:
Validation
of atmospheric
correction
over
the
oceans.
J. Geophys.
Res.,
102,
17209
-
17217.
Holbenl
B,, A.
Setzer,
T. F. Eck,
A. Pereira
and
I. Slutsker,
1996:
Effect
of Dry-Season
Biomass
Burning
on Amazon
Basin
Aerosol
Concentrations
and
Optical
Properties.
J.
Geophys.
Res.,
101,
1992-1994.
Houghton,
J. T.,
L.
G.
Meira
Filho,
B. A.
Callendar,
N.
Harris,
A.
Kattenberg
and
K.
Maskelt
(Edit0rsi,
1996:
Climate
Change
(1995)"
Press,
N.Y.
intergovernmental
Panel
on
Climate
Change
(IPCC),
the
Science
of
Climate
Change,
Cambridge
University
Nakajima,
T.,
M.
Tanaka
and
T. Yamauchi,1983:
Retrieval
of the Optical
Properties
of
Aerosois
from
Aureole
and Extinction
Data.
Appl.
Opt.,
22,
2951-2959.
.;,-2
14
Nakajima, T., G. Tonna,
R. Rao,
P. Boi,
Y.J.
Kaufman
and
B. Holben,
1996:
Use
of Sky
Brightness
Measurements
from
Ground
for
Remote
Sensing
of
Particulate
Polydispersions,
Appl.
Opt.,
35,
2672-2686.
Ritter,
J.
M.,
1998:
Remote
measurement
of
aerosol
scattering
properties
and
the
development
of a novel
imaging
solar
aureole
radiometer,
Ph.D.
Thesis,
University
of
Miami,
Coral
Gables,
F1. 33124.
15
FigureCaptions
1)
2)
3)
4)
5)
6)
Schematic
of aureole
camera
system.
Linearity
plot
obtained
with
aureole
camera
system.
Data
points
are
10 x
10 pixel
averages
in the
center
of the image.
System
shows
good
linearity
at over
3 orders
of
magnitude.
Standard
deviation
of the pixel
values
are displayed
as error
bars
in the
x
direction.
Sample
aureole
image
contour.
Several
important
features
in the image
are identified.
Plot
of
the
aureole,
separating
left
and
fight
sides
of
the
aureole
almucantor.
This
illustrates
the
symmetry
of
the
data
set.
This
type
of
plot
can
be
used
for
quality
control
of the data.
Sampled
aureole
almucantor
data.
Aerosol
phase
function
derived
from
aureole
data.
Phase
functions
were
derived
using
single
scattering
and
assuming
the aerosols
were
contained
in a single
layer.
.
4,.
16
i-ilter
(nm)
440
56O
670
86O
_enn'old (nm)
"
440.9
561.3
670.2
856.6
Full
width
at half
maxLrnum
(nm)
I0
I0
12
13
Tablel.
Interference
filter
bandwidth
and center
frequencies
17
B_,up
ooou,to
jc>
139.8cm
150.5cm
I
I
I
I
I
SolarAlignment
Trigger
Optics
EXternal
cooling
system
Power
Occulter
_rture
Wheel
,Solar Alignment
Trigger
Electronics
Interface
_:i l-
I0000
,
° I
•_
1000
I
_:
Camora
Counts
Occulter
pole
".-___,
- i I_,t4m
. ._:__-_.
superstructure
'L
/.
Radiance OIW cm .2 sr-1 nm-l)
Q
0
,
Q
Q
_
O
I
I
e
°
!
-"
:
;
:
•
:
--6_
........*............... _'..............."/...............; ............... "............... *..............._.........
!
"_
i
,
•
!
i
_
!
!
!
_"
I_
.......................
I,a ..........
_................ _............... -.................----;,_.-_.,._,................ :-...............+............... _.........
,_ ............ _....................
' ................
_....................
i....................
_....................
_...................
_..................
i.........
................
• ..............._................-..............., ........
I(X)-
'_
9-
u
U
7-
i
6-
5-
<
4-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
!
!
!
2x10 o
3
4
Scattering
angle
(degrees)
+
+
+
+
I
I
I
I
5
6
7
8
,%,.
Aerosol
phase function (Pa)
O
O
O
O
..a
-a.
_
.._
._,
IX_
f_
_
IX3
f_
¢.O
CO
G0
6",
APPEND
IX
4
Measurements
of Aerosol
Optical
Depth
over
the
North
Atlantic
Ocean:
CorrelatiOns
with
Surface
Aerosol
Concentrations
Ellsworth
J. Welton,
Kenneth
J. Voss,
Dennis
L. Savoie,
and Joseph
M. Prospero
Abstract:
The
optical
properties
of
atmospheric
aerosols
are
an
important
element
of
the
global
radiation
balance.and
in
applications
such
as
remote
sensing.
One
of
the
most
important
optical
properties
is
the
aerosol
optical
depth
(AOD)
and
it's
associated
wavelength
dependence,
characterized
by
the
Angstrom
exponent
or.
Long
term
measurements
of these
aerosol
features
taken
at various
locations
are necessary
to track
seasonal
patternsof
aerosol
optical
behavior
and to determine
characteristic
differences
in
the
optical
properties
of
different
sites.
An
AOD
measurement
program
was
begun
in
August
of
1993
to
determine
the
optical
properties
of
aerosols
over
existing
Atmosphere/Ocean
Chemistry
Experiment
(AEROCE)
sites
in Miami
(Florida),
Bermuda,
and
Barbados.
A
description
of
the
optical
program,
instrumentation
and
calibration
procedures,
and
the
methodology
employed
to determine
the
AOD
and
ot are
presented.
Analysis
of
the
AOD
and
Angstrom
exponents
in
terms
of
seasonal
variations
and
correlations
with surface
concentration
measurements-is
also presented.
Seasonal
variations
in the
optical
properties
of the aerosols
were
found•
to be
due
to seasonal
changes
in the
types
and concentrations
of specific I aerosol
species.
The
correlational
analysis
has
shown
that
under
certain
circumstances
aerosol
species
may
be identified
over
the ocean
based
on
analysis
of the AOD
and Angstrom
exponent.
:_
Keywords.:-
aerosols,
optical
dePth,
Angstrom
exponent,
sunphotometer,
shadowband,
sulfate,
Sea=salt,
dust
1.
Introduction
There
is relatively
little
information
on the climatology
of atmospheric
aerosols,
particularly
over
the ocean.
However,
the radiative
effects
of
marine
aerosols
directly
alter
terrestrial
optical
properties,
such
as the planetary
albedo,
and may
play
an important
role,
directly
and
indirectly,
in
the global
climate
[Charlson,
1992].
In addition
to this
climatic
impact,
marine
aerosols
also
affect
our
ability
to
extract
surface
information
from
satellites,
in
particular
for
ocean
color
remote
sensing
[Gordon,_
1997].
Knowledge
of
the aerosol
optical
properties
are
necessary
to correct
for
these
effects
in both
Global
Circulation
Models
and in satellite
correction
algorithms.
The
most
commonly
measured
aerosol
optical
property
is the
aerosol
optical
depth
(AOD),
which
determines
how
the aerosol
attenuates
the direct
solar
beam.
The
total
optical
depth,
x,
at a
particular
wavelength
is def'med by
1
-
m(0)
In
(1)
where
m(O)
is
the
alrmass
at
zenith
angle
O,
E o is
the
extra-terrestrial
solar
irradiance
(solar
constant),
and
E is the
direct,
unscattered
and
unabsorbed
solar
irradiance
at
the
surface.
For
wavelengths
that
he outside
the usual
atmospheric
gas absorption
bands,
the total
optical
depth
may
also
be written
as the sum
lr = *'R + "Co + *a
(2)
where
"cRis the Rayleigh
opticaldepth
due
to molecular
scattering,
x o is the Chappius
band
ozone
optical
depth,
anti
"cA is the
aerosol
optical
depth.
Tl_.e basic
experimental
method
of acquiring
the
AOD
from
the'_total
optical
depth
has
been
outlined
in seve-ral
papers,
most
notably
Shaw
[1979]
_.
and
King
et al. [19801.
Measurements.
of
aerosol
optical
depths
[King.
et.
al.,.
1978;
Hoppel
et.
al.,
1990;
•
_
Ka,fmanii993_;
D"tton
et: al., .t994;Smirnov
et. aL;
1994;
1995],
made
at a variety
of locations
•
).,
.
around
the globe
and
at wavelengths
throughout
the visible
spectrum,
have
shown
that
the spectral
dependence
of the AOD
often
resembles
a power
law [Angstrom,
1964]:
=/3;t-a
(3)
where
"ca is the AOD
at wavelength
k,
13 is a scale
factor,
and
ot is the Angstrom
exponent.
The
Angstrom
exponent
_x may
be related
to the exponent
of a Junge
type
size distribution
(dn/dr
= Cr
y = o_+ 2
(4)
[Van
de Hulst,
1981].
The
exponent,
or, generally
varies
from
zero
to two.
Lower
values
of o_ are
associated
with
relatively
more
larger-sized
particles
than
aerosol
populations
with
higher
oc
values.
In order
to obtain
long
term
data
sets
of the
AOD,
hand-held
sunphotometers
were
used
at
existing
AEROCE
(Atmosphere/Ocean
Chemistry
Experiment)
sites
in Miami
(Florida),
Bermuda,
and
Barbados.
AEROCE
sites
also
measure
various
aerosol
parameters
including
concentration,
and
mass-size
distributions.
The stations
are located
at coastal
sites
and,
with
the exception
of the
AOD
measurements,
samples
are collected
only
during
on-shore
winds;
thus
the
aerosol
data
should
be
representative
of
the
regional
oceanic
aerosol.
The
hand-held
sunphotometers
were
replaced
with
automated
multi-f'dter
rotating
shadowband
radiometers
by
the
end
of
1994.
The
shadowbands
create
a more
complete
data
rectrd
as they
automatically
sample
all day,
perform
a
potential
calibration
each
day
(dependen
t on
weather),
and
also
measure
diffuse
irradiance.
The
shadowbands
operated
concurrently
with
the:
sunphotometers
for
several
months
during
the
.i-"
'
z-
'"
'
-'
"
instrument
replkcem_nt
process.
"i'able
1 indicatts
the geographical
information
and
time
period
of
the
sunphotometer
and
shadowband
measurements
:;-at each
site.
The
instrument
calibration
procedure,
methods
of data
selecton,
aerosol
optical
depths,
and Angstrom
exponents
recorded
for
each
site
are presented
in this paper.
2
Recentmeasurementsof aerosolopticaldepthshaveindicateda needto alsoanalyzetheir
otherphysicalcharacteristicsin orderto fully understandtheirclimatologicalimpact[Kaufmanet.
al.,
1994;
Smirnov
et. al.,
1995].
Li et. al. [ 1996]
have
published
results
of short
term
analysis
of
the optical
properties
of aerosols
in Barbados
using
concurrent
analysis
of aerosol
composition
and
concentration
levels.
Also,
Pueschel
et. al. [1994]
have
performed
similar
analysis
of stratospheric
Pinatubo
aerosols
from
aircraft
observations.
Finally,
several
other
researchers
[Hoppel
et.
al.,
1990;
Smirnov
et.
al.,
1995]
have
used
synoptic
air
mass
analysis
to describe
aerosol
radiative
measurements.
However,
there
have
been
no long
term
studies
of the relationship
between
aerosol
optical
depth
measurements
and the concentration
of specific
aerosol
species.
The radiative
properties
of aerosols
are a function
of two
main
factors:
the composition
of
the aerosols,
and the aerosol
production
mechanisms.
The
magnitude
of the AOD
is proportional
to
the concentrations
of the aerosols
(the scale
factor,
_).
The
Angstrom
exponent
is dependent
on
the production
mechanisms
of each
aerosol
species.
Aerosol
concentrations
are available
for
each
AEROCE
site.
Correlations
between
concurrent
measurements
of
the
AOD
and
aerosol
concentrations
and resulting
seasonal
dependencies
are also
presented
in this paper.
The
goal
of the
correlational
analysis
was
to determine
if it is possible
to distinguish
between
different
aerosol
species
based
only
on
analysis
of
the
AOD
and
Angstrom
exponent
(two
parameters
readily
available
from
satellite
remote
sensing
programs).
5.1
3
2.
Instrumentation
Description
Two
different
instruments
were
deployed
at
each
AEROCE
site:
sunphotometers,
and
Multi-filter
Rotating
Shadowband
Radiometers
(shadowbands).
The
instruments
were
used
to
determine
the
total
optical
depth
from
measurements
of
the
direct
sunlight
using
Eq.
(1).
The
AOD
was calculated
from
the total
optical
depth
using
Eq. (2).
Sunphotometers
were
first
installed
at each
AEROCE
site in the Fall of 1993.
The
sunphotometers
were
replaced
with
shadowbands
by
the end
of 1994.
2.1
Sunphotometers
The
sunphotometers
[d'Almeida
et al.,
1983]
were
manufactured
by
NOLL
GMBH
in
Germany,
and were
provided
by John
DeLuisi
at CMDL-NOAA.
The
sunphotometers
used
in this
study
had
nine
interference
f'dters,
with
passbands
approximately
5
nm
wide,
centered
at
the
wavelengths
shown
inTable
2.
Identical
filters,
selected
from
the
same
lot,
were
used
in each
sunphotometer
and were
selected
to avoid
water
vapor
absorption
bands.
The
sunphotometers
are
operated
manually
by pointing
the instrument
at the sun.
The operator
then
moves
the instrument
in
small
circles
around
the
central
direction
of
the
sun.
The
instrument
retains
the
highest
signal
measured
and
displays
this
reading,
which
is the
direct
sun
signal
that
is used
to determine
the
optical
depth.
The
operator
performs
direct
sun measurements
for each
filter
in the sunphotometer.
The
signal,
filter
number,
instrument
temperature,
and
the time
in GMT
(either
from
an
accurate
watch
or
the instrument
display)
are recorded
and
then
the process
is repeated
for
the
remaining
filters.
2.2
Shadowbands-
:"
i
The
shadowbands
were
designed
by Lee
Harrison
and Joe
Michalsky
at SUNY/Albany
and
were
manufacturedby
BatteUe
Labs
in Richland,
WA.
A:complete
description
of the shadowbands
may
be found
i n Harrisi3n
et al.i[1994].
Shadowbands
measure
total
downwelling
irradiance
and
the diffuse
irradiance
by shadowing
the irradiance
collector.
The
direct
irradiance,
determined
by
4
subtractingthediffuse irradiancefrom thetotal irradiance,is thenusedto determinethe spectral
opticaldepth.The shadowbandsareautonomousinstrumentsdesignedto operateremotelyin the
field and arecontrolledvia a phoneline modemlink. The shadowbandconsistsof two main
components:acontrolunit,anddetectorplatform.Thedetectorplatformholds thebandmotor and
detectorhousing. The detectorhousing holds the irradiancecollector and seven photodiode
detectors.Sixof thesevenphotodiodedetectorshavespectralinterferencef'dtersencapsulatedwith
thedetector.Thef'dterwavelengths,with passbands10nm wide, aregiven in Table2. Filter 1, a
broadbandfalter,andfalter7, a watervaporbandf'dter,werenot usedin this study. Filters2 to 6
werechosento avoidwatervaporabsorptionbands.Thefilters arepositionedsothattheyview the
backof theirradiancecollector.Measurementsareperformedwith the7f'dters simultaneously.
2.3
Aerosol
Concentration
Measurements
Ground-based
aerosol
concentration
samples
are collected
by
drawing
air
through
20x25
cm
Whatman
41
(W41)
filters
at a flow
rate
of
about
1.1
m 3 min -_, yielding
average
sampled
volumes
of about
1500m
3. W41
filters
have
collection
efficiencies
greater
than
90%
for nss
sulfate
and
ammonium,
95%
for
nitrate
and
sea-salt
[Savoie,
1984],
and
95%
for
mineral
aerosol
[Arimoto
et al.,
1996].
At all sites,
the aerosol
sampler
is linked
to a wind
sensor
which
controls
the operation
of the sampler
so that it is activated
only
when
the wind
blows
from
the open
ocean
sector
at a speed
greater
than
1 rrds.
FilLers are reUa'ned
to Miami
were
they
are extracted
with
deionized
water
and
the extracts
analyzed
for
major
soluble
inorganic
ions:
Na ÷ by
flame
atomic
absorption
and cr,
NO3,
SO4--,by
suppressed
ion chromatography
[Savoie
et al.,
1989]
and NH4 + by
autctnated
colorimetry.
Non
sea-salt
sulfate
(nss
SO4 =) is caloMated:
[total
SO4_
- [Na ÷ * 0.2517],
where
0.2517
is the SO42-fNa _ mass
ratio in bulk sea watfr.
The
extracted
filters
were
then
placed
in
F:,..
a muffle
furnace
for about
14 hours
at 500°C;
the residue
weight
(less
falter blank)
is assumed
to
be mineral
dust:ash._/
-::....
:_:....
5
3.
Sunphotometer
and
Shadowband
Calibrations
The
instruments
described
in Section
2 must
be calibrated
before
beginning
analysis
of the
data
taken
with
them.
For
the purposes
of this
study,
the term
calibration
refers
to all procedures
that
must
be performed
upon
the instrument
in order
to determine
the aerosol
optical
depth
using
data
acquired
with
the
instrument.
The
procedures
used
to
calibrate
the
sunphotometers
and
shadowbands
are described
in this section.
3.1
Sunphotometer
Calibrations
The calibration
of the sunphotometers
involves
determination
of Eo (in instrument
counts),
also
referred
'to as the solar
constant,
by using
procedures
based
on the
Langley
calibration
method
[Shaw,
1979;
King
et
al.,
1980].
The
sunphotometers
were
operated
by
on-site
AEROCE
personnel
who
recorded
measurements
at
approximately
10:00
am
and
3:00
PM
local
time
respectively.
Therefore,
normal
instrument
operations
do
not
acquire
enough
data
to
perform
Langley
calibrations.
Specific
Langley
measurements
had
to be
performed
with
each
instrument
from
the Miami,
Bermuda,
and
Barbados
AEROCE
sites.
An additional
sunphotometer
was
used
for calibration
processes
during
and after
the measurement
period.
Initially
the Langley
method
was
used
to perform
the calibrations,
for
each
sunphotometer,
in Miami
prior
to deployment
into
the
field.
The
instruments
were
then
sent
into
the
field,
and
operations
began
as
indicated
in
Table
1.
With
the
exception
of
poor
weather
days,
the
measurements
continued
uninterrupted
for
the
remainder"
of
the
sunphotometer
program.
It was
difficult
to
perform
sea-level
Langley
calibrations
in
these
locations
because
of
atmospheric
variability
and
cloudiness.
_.Thus'
it was
not.possible
to have
the
on-site
operator
perform
routine
Langley
calibratiom
at the_Bermuda
and Barbados
locations.
The
Miami
instrument
(M114)
was
calibrated
several
times
during
_e
sunphotometer
program,
both
in Miami
and
in Brainard
Lake,
Colorado.
An
additional
sunphotometer
(Ml19),
not
tied
to
any
location,
was
extensively
calibrated
during
an oceanographic
cruise,
off of Hawaii-,
in October
and November
of 1994.
Post-
6
calibrationsfor theBermudaandBarbadosinstrumentswereperformedin Miami at theendof the
sunphotometerprogram(throughamethoddescribedbelow).
A calibrationhistoryfor eachinstrumentwascompiledin orderto accountfor timeshifts in
the solar constants,often causedby degradationof the f'tlters. The calibrationrecord for an
instrumentconsistsof a plot of thedateversusEofor the lengthof the entiredataset. For the
periodwhentheinstrumentsweredeployed,calibrationvalueswereobtainedby an interpolation
betweentheinitial calibrationsandthepost-calibrations.Prior experiencewith thedegradationof
sunphotometerinterferencefiltersledto theuseof anexponentialfunctionfor theinterpolationof
thecalibrationconstants.An errorcorrectionprocedurewasthenemployedfor bothinstrumentsto
modifytheconventionalLangleycalibrations.
It wasdifficult to performfull Langleycalibrationson the instrumentsat the end of the
sunphotometerprogramdue to poor weather.TheMiami instrumentwascalibratedseveraltimes
duringtheinitial startup,
and
throughout
the program,
and
was considered
to be
the best
calibrated
of
the
sunphotometers.
The
Ml19
sunphotometer
was
well
calibrated,
using
the
Langley
procedure,
during
October
and November
1994
and
was
used
to calibrate
the Miami
instrument
at
the end
of the sunphotometer
program
through
a cross-calibration
procedure.
A cross-calibration
assumes
that two
identical
sunphotometers
are
present;
one
is fully
calibrated
and
is referred
to as
the reference
instrument,
while
the other
is uncalibrated
and
referred
to as
the
target
instrument.
Simultaneous
direct
beam
measurements
are made
with
each
sunphotometer
at the same
location.
The
resulting
equations
for each
instrument
are
E r = EroeXp[-mr(0)_]
,
(5)
(6)
the
t subscript
denotes
the
target
E
= EtoeXp[-.m,(0)'r]
4
where
the
r subscript,
denotes
the
reference
instrument
and
instrument.AsE
_ is known,
the x is calculated
using
the CaJibrated
reference
instrument.
Once
the
TOD
is determined,
E_'can
be:__vritten
as
_.!
i_
7
i-
"
Et exp[-_'(m,(O)
- m,(O))]
E,o = E,o _-
(7)
for each
wavelength
of the uncalibrated
target
sunphotometer.
This
cross-calibration
procedure
was
useful
because
the weather
need
only
be stable
and cloud-free
for a small
window
of time.
Ideally,
the target
and reference
measurements
should
be made
at exactly
the same
lime
so
the two
airmass
values
are equal,
the exponential
term
is zero,
and
x need
not
be determined.
However,
exactly
simultaneous
measurements
are
often
not
possible,
therefore,
the
reference
and
target
airmass
values
should
be as close
as POssible
to avoid
errors
in calculating
E_,.
The
Miami
instrument,
(Ml14),
was
cross-calibrated
against
Ml19
at
the
end
of
the
program
and the cross-calibrations
were
added
to the calibration
history
for
M 114.
Sunphotometer
Ml14
was
then
used
as
a reference
instrument
during
cross-calibrations
for
the
Bermuda
and
Barbados
sunphotometers.
These
cross-calibrations
were
then
added
to the calibration
history
for
Bermuda
and Barbados.
The calibration
histories
for
each
sunphotometer
at the three
locations
are
given
in Figures
la,
lb,
and
lc.
The solid
line
is the exponential
fit to the calibrations
given
above.
An
error
correction
procedure
was
developed
to
fine
tune
these
solar
constants.
This
procedure
assumes
that
there
is some
error,
_,
present
in the solar
constant
for
each
wavelength,
and
that
the
aerosols
above
the
sites,
on
average,
obey
the
Angstrom
spectral
dependence.
Redefining
the solar
constant
in terms
of this error
and the true solar
constant
yields
Eo = X E_
,
(7)
..
t
where
Eo is the previously
deriyed,
solar
Cons .t,3nt,:_
is the error
factor,
and
Eo" is the
true
solar
•'
~
constant.
The
measui:ed
x is given
by
"
:
_-1
in(E_)1
In(z)
re(o) '
(8)
•
•
j
.
_-,
.
-'
._,
.
-! -'.
-.
using
Eq. (7). The
first
term
on the RHS
of Eq. (8) is the true TOD
(V') as follows
from
using
Eq.
:
". ,
,
.
(1) and Eq. (7),
therefore,
the following
equation
is calculated
ln(x ) = [z"- "t:']m(O)
,
(9)
relating
the error
factor,
Z, to the difference
in measured
and true total
optical
depths,
x R and
%
am
subtracted
from
both
the measured
and true total
optical
depths
since
they
are known.
The
resulting
equation
ln(x
) = [v A - _]m(8)
,
(10)
relates
the error
factor,
Z, to the difference
in measured
and true
aerosol
optical
depths.
In the calibration
process,
a sunphotometer
reading
consisted
of recording,
E,
for
each
of
the nine
wavelengths.
The
AOD
was
then
fit to Eq.
(3),
determining
fl and
_x, and
this
equation
was
then
used
to generate
x"A, producing
the Final relation
ln(x
) = ['t"A - ,B_-'z ] m(8)
(11)
Therefore,
In(z),
is the difference
between
the measured
AOD
and
the Angstrom
fitted
AOD
for
a
given
wavelength,
times
the
airmass.
This
factor
determines
the
variation
from
the
Angstrom
power
law for that
particular
measurement.
The
AOD,
for
each
location's
entire
data
set,
was
f'trst calculated
using
the
original
solar
constants
from
the calibration
histories,
and Rayleigh
optical
depths
determined
using
Hansen
and
Travis
[1974]
and
ozone
models
provided
by Klenk
et al.
[1983].
For each
day,
the deviation
of
the AOD
from
the Angstrom
power
law
was
determined
using
Eq.
(11)
and
the
error
factor,
Z,
was
calculated.
The time
history
0f Z was fit by another
exponential
function,
yielding
an equation
for Z for
each
instrumenL
This
equation
was
used
to correct
the solar
constants
according
to Eq.
(7). The error-corrected
solar
constant
histories
are plotted
in Figures
la,
lb,
and
lc
as the dotted
lines.
Channelsi2ne
(380.2
nm)
and nine
(.1025.9
nmi
were
not
processed,
and
were
not
used
in
4''"
this study.
The
380.2
nm
filter
degraded
rapidly
in all instruments
and was
considered
unusable.
Channel
nine
deviated
significantly
from
the Angstrom
power
law,
perhaps
due
to the weak
water
9
vaporabsorptionbandaround1000nm [Shaw, 1979] which wasnotconsideredin our analysis,
or theeffectsof seaspray[Villevaldeet al.,
1994].
These
error-corrected
calibration
values
are not
significantly
different
from
the original
values
but provide
a fine-tuning
adjustment.
3.2
Shadowband
Calibrations
The
shadowbands
began
operation
in the fall of 1994.
The shadowbands
ran
automatically,
eliminating
the need
for
an
on-site
operator.
The
Miami
shadowband
sampled
data
every
minute
throughout
the
day.
The
Bermuda
and
Barbados
shadowbands
sampled
every
four
minutes
to
reduce
the
number
of
data
downloads
per
week.
As
the
shadowbands
perform
measurements
continually,
each
day
offers
the potential
of a Langley
calibration.
Therefore
the calibrations
for
the
shadowbands
are more
complete
than
those
for
the sunphotometers
since
the actual
daily
data
can
be used
to perform
the Langley
calibrations.
Gaps
are present
in all
data
sets
due
to instrument
malfunctions
and the subsequent
time needed
to repair
the problems.
A gap exists
in the Miami
data
from
September
1995
to November
1995.
This
was
caused
by data
communication
problems
and
poor
weather.
Normal
operation
began
again
in December
1995.
The
Bermuda
data
gap,
also
caused
by data
communication
problems,
resulted
in the loss
of data
from
July
1995
to November
1995.
The
communication
problems
were
fixed
in December
1995 and
shadowband
operation
was
continued.
The Barbados
shadowband
data
set only
includes
data
from
May
1995
to August
1995
due
to poor
phone
line connections
for data
transfer
and unstable
electrical
power
at the site.
All of the shadowbands
collect
enough
data
each
day
to perform
two
Langley
calibrations,
one
in the
morning
and
one
in the afternoon,
weather
permitting.
Therefore
all that
remains
is to
determine
which
of the
days
has
weather
suitable
for
Langley
calibrations.
Each
shadowband's
data
set was
maly_
using
the Objective
Langley
Regression
Algorithm
(OLRA)
[Harrison
and
Michalsky,
1994]
in,order
to recover
the solar
constants
for each
instrument.
The
OI_RA
rejected
a
large
number
of the Lang!ey
calibrations
for
all three
sites
due
to the variable
tropical
weather
at
each
location.-Howbver,_the
Strict
criteria
in
the
OLRA:
assures
that
the
remaining
Langley
calibrations
are accurate.
Once
the solar
constants
for each
shadowband
were
determined
using
this
10
technique,acalibrationhistorywascompiledin thesamemannerasfor thesunphotometers.The
calibrationhistoriesfor eachshadowbandandinterpolationsaredepictedin Figures2a,2b,and2c.
Therewere few solarconstantvaluesrecoveredfor theBarbadosshadowbanddue to the
small
time
period
of the data
set. Therefore,
the solar
constants
for the Barbados
shadowband
were
obtained
by using
the mean
value
of the solar
constant
for each
channel
instead
of the interpolations
described
above.
A linear
fit
to the calibration
histories
was
performed
for
all of
the
channels
except
channel
four
(610
rim)
of the Miami
shadowband,
and channels
four
and
five
(610
and
665
nm respectively)
of the Bermuda
shadowband.
The
filters
in these
channels
were
found
to stabilize
after
a period
of time
making
a single
linear
fit unsuitable.
Instead
a linear
fit was
performed
on the
first
part
of the calibrations,
ignoring
the stabilized
portion,
and
a second
linear
fit was
performed
on the more
stable
portion.
These
Langley
calibrations
were
fine
tuned
with
another
procedure.
This
procedure
assumes
that
there
is some
error,
_t, present
in the solar
constant.
Angstrom
spectral
dependence
in
the AOD
was
not
assumed.
The
shadowband
records
more
samples
and
with
more
variation
in
solar
zenith
angle
than
the
sunphotometer,
therefore,
it is possible
to analyze
a month's
worth
of
shadowband
data
to test
for dependence
of the AOD
on solar
zenith
angle.
We
assume
that
the set
of minimum
aerosol
optical
depths
should
not depend
on solar
zenith
angle
(airmass)
over
the span
of one
month.
Then
the lowest
optical
depths
on
a plot
of
the
AOD
versus
l/m(0)
for
a given
month
will represent
a background
AOD.
A linear
fit to the lowest
aerosol
optical
depths
in the plot
described
above
should
have
ze.riJ slope
and a yintercept
equal
to the average
background
AOD
for
that
month.
A
slope
not
equal'to
zero
would
indicate
that
the
background
AOD
has
some
and
thus
an
error
in
the
solar
constant
[lgnatov,
personal
dependence
on
the
airmass,
communication,
i996]_
.
Assuming
that the_js
some
error,
_t, in the solar
constants
yields
the following
equation
7.'-.
Eo = rE;
,
(12)
11
whereEois thepreviouslyderivedsolarconstant,IXis theerrorfactor, and E'o is the true solar
constant.InsertingEq. (12)intoEq. (1)producesthefollowing relation
"r-
ln(/l)+---Lln
=ln(_)_l
+r"
,
(13)
m(0)
m(0)
k, E J
m(0)
x" is the true
optical
depth
as it contains
the true
solar
constant,
Eo'.
The
Rayleigh
[Hansen
and
Travis,
1974]
and ozone
[Klenk
et al.,
1983]
optical
depths
are not
dependent
on the calibrations
so they
may
be subtracted
from
both
sides
of Eq.
(13) to produce
the equation
1
lra = ln(/_)_
+ _a
(14)
m(0)
Equation
(14) may
only
be used
when
both
xA and
x"A represent
the background
(minimum)
AOD.
Therefore,
the slope
of the
background
AOD
versus
l/m(0)
plot
described
above
is the
natural
logarithm
of _ and
the y intercept
is the true
background
AOD
(x'a).
This
procedure
was
used
to
obtain
monthly
values
of Ix for
each
shadowband
channel.
The
error
corrected
solar
constants
are
plotted
in Figs.
2a,
2b,
and 2c.
4. Aerosol
Optical
Depths
and Concentrations
The aerosol
optical
depths
and
corresponding
Angstrom
exponents
were
calculated
for the
AEROCE
sites
in Miami,
Bermuda,
and Barbados
using
both
the sunphotometer
and
shadowband
data.
The
uncorrected
arid the,error,
corrected
calibrations
described
in sections
3.1
and
3.2,
were
applied
to the data
sets
separately
in order
io gauge
the usefulness
of each
error
correction
method.
Data
screening
procedures
were
then
used
to
remove
optical
data
affected
by
atmospheric
phenomena
other-than
aerosols,
such
as clouds.
Both_data
screening
procedures
were
similar,
but
due
to the
nature-0f
the
instruments,
a
different
scr_ning
procedure
was
employed
for
each
instrument.
The
screening
procedures
and
comparisons
between
the
uncorrected
and
error
corrected
results
are described
below.
The tolerance
criteria
in each
of the data
screening
processes
12
werea compromisebetweeneliminatingquestionabledataandretainingenoughof thedatasetto
analyze.Finally, the resulting AOD and Angstrom exponentswere correlatedwith surface
measurementsof thespecificaerosolspeciesconcentrationsfor eachsite.
4.1
Sunphotometer
Data
Screening
Procedure
The
aerosol
optical
depths
and
Angstrom
exponents
were
calculated
for
channels
two
through
eight,
for
the
respective
data
sets,
due
to the calibration
problems
with
channels
one
and
nine.
The
sunphotometer
data screen
contained
three
levels.
Level
One
determined
the deviation
of
the measured
x^ from
the power-law-fitted
't"A for
a given
measurement,
similar
to the error
factor
procedure
above.
If the magnitude
of the deviation
between
"CAand
x' A was
greater
than
0.1,
then
the
measured
AOD
at
that
wavelength
was
rejected.
This
was
done
to
screen
out
AOD
measurements
that
varied
strongly
from
an Angstrom
power
law.
The
aerosol
optical
depths
that
survived
this level
were
then
subjected
to Level
Two
of the screen.
It should
be noted
that
a given
measurement
set consists
of two
series
of readings,
E,
for
each
of
the
nine
channels.
This
was
done
to ensure
that
atmospheric
properties
were
fairly
constant
during
the measurement
set,
as the
optical
depth
should
not
change
appreciably
during
a span
of five
minutes
(the
approximate
time
for
one
measurement
set).
The
Level
Two
screen
examined
the
difference
between
the
fh'st
and
second
measured
aerosol
optical
depths.
If the magnitude
of the difference
was
greater
than
0.03,
then
that channel
was rejected.
If only
one of the two dual
readings
survived
the Level
One
screen,
then
the Level
Two
screen
was
not performed.
Finally,
the Level
Three
screen
determines
if at least
six
of the
seven
:channels
(one
and
n_e
are
excluded)remain,
and
channels
two
and
eight
are
among
them.
If this
was
true
then
this
m'easurement
Was considered
usable,
otherwise
the
entire
measurement
was
discarded.
This
screening
process
iehsures
that
the
atmospheric
aerosol
properties
are
fairly
constant
during
the
measurement,
_ that
the
AOD
roughly
resembles
an
Angstrom
power
law
formula,
and
that
there
a_e
enough
remaining
aerosol
optical
depths
to
accurately
perform
a fit to the Angstrom
power
law (for
the wavelength
range,
412.2
nm
to 861.8
13
nm only).TheAOD wasthencalculatedfrom thesurvivingmeasurementsandusedto generatethe
Angstromparameters.TheremainingAOD valuesandtheAngstromexponentwerethenstoredfor
thatsample.
4.2
Shadowband
Data
Screening
Procedure
The
shadowbands
record
data
continuously
throughout
the
day,
as
opposed
to
the
sunphotometer's
singular
morning
and
afternoon
measurements.
Therefore,
a different
screening
method
to eliminate
bad
data
was
developed.
This
method
was
based
on
the
Sliding
Window
Optical
Depth
Procedure
(SWODP)
[Schlemmer,
personal
communication,
1996]
developed
at the
Atmospheric
Sciences
Research
Center
at the State
University
of New
York,
Albany.
The
University
of Miami
SWODP
(MSWODP)
used
the
aerosol
optical
depths
for
each
sampled
measurement
by
the
shadowband.
The
MSWODP
then
analyzed
one
day
at
a time,
starting
with
the first
measurement
sample.
The
term,
"sliding
window,"
originated
because
the
MSWODP
analyzed
a twenty
minute
"window"
of data
to determine
if
the
window
contained
usable
data
(free
from
cloud
contamination).
Two
screening
levels
were
then
applied
to
the
resulting
AOD
window
by
the MSWODP.
Level
One
performed
a linear
least
squares
fit to the
AOD
(versus
time),
and
then
calculated
the individual
AOD
deviations
from
the
fit.
If
all of
the
aerosol
optical
depths
were
within
0.01
of the linear
fit, then
the MSWODP
proceeded
to the Level
Two
screen
for that
window.
If Level
One failed,
then
the MSWODP
slid
the window
ahead
one
sample
measurement
and
began
the scr_ning
process
anew.
If Level
One
was
successful
then
the
MSWODP
applied
Level
Two,
which
determined
the
mean
AOD
for
that
window.
If the
mean
AOD
was less
then
1.0 then
the_MSWODP
rei_orded
.the mean
AOD
and
corresponding
Angstrom
exponen
t for
that
window.
If the
mean
AOD
wasgreater
than
or
equal
to
1.0
then
the
entire
window
was
rejecie, d,
and
no
data
were
recorded
for
that
window
of
time
(possible
cloud
presence).
Regardless
of Me outcbme
of Level
Two,
the MSWODP
then
slid the window
ahead
by
twenty
minutes
to the corresponding
sample
measurement,
and
the process
was
started
over
again.
The
MSWODP
yields
twenty
minute
averages
of
the
AOD,
and
the
corresponding
Angstrom
14
parametersfor eachday in thedataset.This analysisprocedureresultsin aerosolopticaldepths
that do not vary erratically, that were reasonablemagnitudesfor atmosphericaerosols, and
excludedclouds.TheremainingAOD valuesandtheAngstromexponentwere thenstoredfor the
timeof dayfalling atthecenterof thetwentyminutewindow.
4.3
Comparison
of
Uncorrected
and
Error
Corrected
Results
The
spectral
variation
of
the
aerosol
optical
depths
for
each
sunphotometer
and
shadowb_.nd
were
analyzed
to compare
differences
between
the
uncorrected
and
error
corrected
(solar
constant)
results.
The AOD
values
were
plotted
versus
the wavelengths
and the Angstrom
fit
was
applied
to the data.
The
comparison
focused
on examining
any
changes
to the
Angstrom
fit
caused
by the application
of the error
correction
procedure.
The
uncorrected
sunphotometer
results
are
not
much
different
from
the
error
corrected
results,
indicating
that
the data
were
not
changed
significantly
by the
error
correction
procedure.
However,
a small
improvement
to the
Angstrom
power
law
fit
was
obtained
using
the
error
corrected
sunphotometer
results.
The
average
chi-squared
data
fitting
parameter
for
each
data
set
indicated
a better
power
law
fit. The
Miami
chi-squared
data
fitting
parameter
was
0.086
for
the
uncorrected
results
and 0.034
for
the error-corrected
results.
The
Bermuda
uncorrected
and
error-
corrected
chi-squared
parameters
were
0.065
and
0.012
respectively.
Finally,
the
Barbados
uncorrected
and error-corrected
chi-squared
parameters
were
0.744
and 0.058.
The
sunphotometer
error
correction
procedure Iassumes
that the true
AOD follows
the Angstrom
power
law.
Furthermore,
the sunphotometer
data screen
explicitly
screened
out
days
(for
both
uncorrected
and
error
corrected
results)
that did not accurately
fit the
Angstrom
power
law.
For
these
two
reasons,
an improvement
in the fits
to the Angstrom
power
law
between
uncorrected
and
error
corrected
results
was expected.
There
were
significant
differences
between
the
uncorrected
and
error-corrected
..-
.-
,
.
-,
results
for certain
channel
s of each
shadowbaiad.
Howeverl
all of
the error-corrected
changes
also
resulted
in a better
average
Angstrom
power
law fit. In particular,
a clear
bias
in channel
five
of the
15
Miami shadowbandwas removedafter using the shadowbanderror correctionprocedure.The
Miami chi-squareddatafitting parameterwas0.521 for theuncorrectedresultsand0.122 for the
errorcorrectedresults.TheBermudauncorrectedanderrorcorrectedchi-squaredparameterswere
0.300 and0.131 respectively.Finally, theBarbadosuncorrectedanderror correctedchi-squared
parameterswere 0.071 and 0.059. The shadowbanderror correction procedure and the
shadowbanddatascreendid notassumeanyparticularspectralform of theAOD.However,results
obtainedusingtheerror correctedresultsmoreaccuratelyportrayeda powerlaw fit comparedto
theuncorrectedresults.
LevelOneof thesunphotometerdatascreendeterminedthedeviationof themeasuredAOD
from the Angstrom
power
law.
Turning
off
Levels
Two
and
Three
of
the
sunphotometer
data
screen
allowed
the percentage
of measurements
rejected
by only
Level
One
to be determined.
The
Level
One
sunphotometer
data
screen
rejected
3%
of the Barbados
measurements
and
7%
of
the
Miami
and
Bermuda
measurements.
Therefore,
at
all
three
locations,
over
90%
of
the
sunphotometer
AOD
measurements
resembled
an
Angstrom
power
law.
Also,
there
were
significant
improvements
in the Angstrom
power
law
fits
using
the
error-corrected
shadowband
results
compared
to using
the uncorrected
shadowband
data.
This
improvement
indicates
that
the
majority
of shadowband
AOD
measurements
also
resembled
an Angstrom
power
law,
particularly
since
no spectral
dependence
of the AOD
was assumed
but the error
corrected
data
more
accurately
fit the Angstrom
power
law.
As a result of this analysis,
the majority
of the aerosol
optical
depths
measured
over
Miami,
Bermuda,
and
Barbados
were
able
to be
fit to an Angstrom
power
law
in
the wavelength
range 400
nm to 860 urn.
4.4
AOD
and
Concentration
Correlation
Procedure
The
optical
data
were:analyzed
to determine
seasonal
patterns
and
to identify
correlations
with
specific
aerosol
types
and concentrationsi
The
basic
procedure
was
to search
both
the optical
data
and
the aerosol
concentration
data
for
periods
that
overlapped.
The
aerosol
sampling
period
was
usually
one
day
long,
however
in some
cases
the
concentration
measurement
period
was
16
severaldays long. As a result, theoptical datawere averagedover the length of the aerosol
measurementperiodduringthecorrelationprocess.For eachcorrelationperiod,thefirst stepwas
to determinewhichaerosoltypeswerepresentduringthecomparisonperiodandif oneappearedto
bedominant.Theaerosolsof interestduringthisstudywerenon-sea-saltsulfate(nssSO:), nitrate
(NO3),sea-salt,andmineraldust.
The determinationof a dominantaerosolspecieswas basedon analysisof the mass
concentrations.It must benotedthatsea-saltaerosolshavealow mass-scatteringefficiencyMSE
[Li et al.,
1996]
and
they
are
confined
to
the
marine
boundary
layer.
Therefore,
the
overall
contribution
of sea
salt
to the AOD,
is generally
expected
to be much
less
than
the
contributions
from
the other
aerosol
species.
However,
the sea-salt
concentrations
at each
location
were
usually
greater
than
the other
aerosol
concentrations;
therefore,
a direct
comparison
of
the
concentration
levels
alone
would
exaggerate
the importance
of sea-salt
on the optical
behavior.
To de-emphasize
the role
of sea-salt
aerosols,
the other
aerosol's
concentrations
were
fin'st compared
to that
of
sea
salt.
First,
the mean
mass
concentration
of each
aerosol
at the measurement
sites
was
calculated,
the results
are displayed
in Table
3. If the concentration
of sea
salt
was
within
30%
of it's
mean
value
for the site in question,
and the concentration
of the other
aerosol's
was
50%
less
than
that
of
the sea-salt,
then
sea-salt
was
considered
to be dominant.
If this
initial
criteria
was
not
met,
then
sea-salt
was
excluded
from
the
correlation
and
the
dominant
aerosol
was
chosen
from
the
remaining
aerosol
species
by
assigning
dominance
to the aerosol
with
a concentration
at least
50%
greater than
that
of the other
aerosols.
If both
the sea-salt
comparison
and
the remaining
aerosols
comparison
failed
to determine
a dominant
aerosol,
then
that
period
was
discarded.
If a particular
aerosol
was
found
to be
dominant
then
the
optical
effects
recorded
during
that
period
were
attributed
to this aerOsol
SPecie s:i=i_
":
Analysis
of flae correlated
data
sets
was
undertaken
to study
how
the AOD
and
Angstrom
exponents
vary
dtiringperiods_:of
dominance
by
different
aerosol
species.
Relative
increases
or
decreases
in dominant
aerosol_60ncenirations
from
one i.period
to the
next
should
correlate
with
corresponding
increases
or
decreases
in the
AOD.
Also,
long
term
measurements
of
the
near-
17
surface concentrationsof aerosolshave shown that individual aerosol types often have
characteristicsizedistributions[Savoieet al.,
1982;
Kaufman
et al.,
1994].
Non-sea-salt
sulfate
particles
are primarily
found
in the submicron
range
with
mass-median
diameters
generally
below
0.5
microns.
Dust
particles
over
the ocean
(that
is,
1000
km
or
more
from
the sources)
typically
have
a mass
median
diameter
of
several
microns
or
larger,
although
there
is a substantial
sub-
micron
component
[Li et al.,
1996].
Sea-salt
aerosols
also
are predominantly
supramicron,
with
a
mass
median
diameter
in 4-6 microns.
Therefore,
periods
of non-sea-salt
sulfate
dominance
should
contain
particles
that
are primarily
submicron
and
thus
should
yield
a higher
Angstrom
exponent
relative
to a period
when
sea-salt
or dust
is the dominant
aerosol;
as in the latter
case
the particles
would
primarily
be
at
least
several
microns
in
size.
It
should
be
noted
that
another
primary
anthropogenic
aerosol
is nitrate.
While
nitrate
aerosols
are derived
from
gaseous
precursors,
as are
non-sea-salt
sulfates,
in the marine
environment
the nitrate
size distribution
follows
the surface
area
of sea-salt
aerosols
[Savoie
et al.,
1982]
due
to a tendency
of the nitrate
precursors
to adsorb
into
existing
sea-salt
particles.
Also,
nitrate
concentrations
were
relatively
low
at all sites
except
Miami
(Table
3). The
majority
of Miami
periods
of high
nitrate
concentrations
correlated
with
moderate
to
high
sea-salt
concentrations.
Therefore,
nitrate
aerosols
were
not
analyzed
in
the
correlation
process.
4.5
AOD
and
Concentration
Correlation
Results
The
monthly
averaged
AOD
at 500
nm,
the
Angstrom
scale
factor
13, and
the
Angstrom
.÷.
Exponent
_,
along
with
their
respective
standard
deviations
(_i)
are
presented
for
Miami,
Bermuda,
and
Barbados.
in Tables
4a,
4b,
and
4c.
The
AOD
data
from
Table
4 are displayed
for
Miamil
Bermuda,
and Barbadosin
Figures
3a,
4a,
and
5a,
respectively.
Tiffs
data
is the result
of
combining
.the datasets
from
the' sunphotometers
and
the shadowbands
for each
site.
Figures
3a,:4a,
and 5a show
ch_acteristic
increases,
in the AOD
during
the summer
months
with
minimum
AOD
values
during
the winter
months.
The
relative
changes
in the AOD
values
for
18
eachsitearetheresultof changesin theconcentrationsof particularaerosolspecies.Figures3b,
4b, and5b displaythemonthlymeanaerosolconcentrationsat Miami, Bermuda,and Barbados.
Thehighsea-saltconcentrationsmeasuredateachsiteareevident.However,as mentionedabove,
therelativeimportanceof thesea-salto theAOD islessdueto it's low MSE.
The Miami and Barbadosconcentrationdatashowsthewell known influenceof Saharan
dusttransportduringthe summermonths[Prospero,1995]. The summerdustconcentrationsat
eachsite aremuch higherthan aerosolconcentrationsfrom otherseasonsand thus createhigh
summertimeAOD values.Thedecreasein dustconcentrationsduringthewinter, early spring, and
latefall, resultsin sea-saltbeingthedominantaerosolmostof the time.
During
these
periods
the
AOD
is lower
due to the low
MSE
of sea-salt.
Miami
occasionally
experiences
non-sea-salt
sulfate
dominant
periods,
particularly
during
the spring
and summer,
and relatively
high
AOD
values
may
occur.
These
sulfate
dominant
periods
often
occur
during
the dust
season
and
therefore
analysis
of
the
Angstrom
exponent
correlations
(see
below)
is
necessary
to
determine
which
aerosol
is
primarily
responsible
for the increase
in AOD.
Barbados
was
not
significantly
impacted
by sulfate
and
it's
AOD
patterns
may
be
understood
solely
on
the
analysis
of
the
sea-salt
and
dust
concentrations.
Bermuda
does
not
show
as
much
seasonal
dependence
in
both
AOD
and
aerosol
concentration
(except
for
sea-salt)
as
the
other
two
sites.
Dust
concentrations
at Bermuda
were
much
lower
than
at Miami
and Barbados,
even
during
the summer
months.
The
relative
absence
of
much
dust
at Bermuda
is the
primary
reason
for
the
lack
of
strong
seasonal
dependence
in
the
AOD.
However,
Bermuda
does
have
a strong
seasonal
dependence
in the sea-salt
concentrations.
Sea-salt
dominance
during
the winter
and fall months
creates
low
AOD
values
relative
to the spring
• months.
During
thespring
months,
non-sea.salt
sulfates
from
North
America
are
present.
The
Bermuda
sulfate
concentrations
are lower
than
those
seen
at
Miami,
however,
the
high
MSE
of
sulfate
[Li
et al.,
1996]
creates
AOD
values
that
are
high
relative
to the
sea-salt
induced
AOD
values
at Bermuda.
This
i_sults
'in the small
seasonal
AOI_
dependence
observed
for Bermuda.
19
The Angstromexponentswere correlatedwith the aerosolconcentrationsfor each site
basedon the proceduredescribedabove.Table 5 displaysthe meanAngstromexponentand
correspondingstandarddeviation(c_) for eachdominantaerosol.TherewerenegativeAngstrom
exponentscalculatedat eachsite.ThenegativeAngstromexponentswereattributedto lessaccurate
Angstromfitting atlow AOD valuessince74%of the negativeAngstromexponentscorrespondto
AOD (at 500 nm) below 0.1 in value. The majority of the negativeAngstromexponent,higher
AOD (> 0.1)datahadexponentsonly slightlynegative(- 0.1or higher).
The non-sea-saltsulfate dominantperiods at both Miami and Bermuda have mean
Angstromexponentsof 1.216and 1.012respectively.Thesea-saltanddust dominantperiodsat
eachsite havemeanAngstromexponentsrangingfrom 0.205 to 0.585 in magnitude.The high
sulfateexponentsrelativeto the sea-saltand dust exponentsshow the predominanceof smaller
particlesduring the sulfate
periods.
Furthermore,
using
the
standard
deviations
in Table
5
and
combining
sea-salt
and
dust
into
a "natural"
aerosol
type
it is possible
to show
the
difference
between
sulfate
and natural
(sea-salt
and dust)
aerosol
exponents.
Based
on the analysis
above,
the
Angstrom
exponents
for sulfate
measurements
over
the ocean
range
from
0.863
to 1.365
while
the
Angstrom
exponents
for natural
aerosols
over
the ocean
range
from
0.184
to 0.614
in value.
A goal
of this research
was to determine
if it was
possible
to distinguish
between
different
aerosol
species
over
the ocean
based
only
on analysis
of the AOD
and
Angstrom
exponent.
As can
be
seen
from
analysis
of the
seasonal
dependence
of
the
AOD
and
aerosol
concentrations
and
correlations
with
the Angstrom
exponent,
at the moment
it is not
generally
possible
to distinguish
between
specific
aerosol
types
based
solely
on
analysis
of
the
AOD
and
Angstrom
exponent.
However,
it is generally
possible
to distinguish
between
natural
(sea-salt
and
dust)
and
sulfate
aerosols
over
the ocean,
and
in specific
cases
of
high
dust
concentrations
it is also
possible
to
distinguish
between
sea-salt
and-dust
species
over
the ocean.
Figure
s 6a,
6b, and 6c display
plots
of
theAOD
versus
the Angstrom
exponent
for
Miami,
Bermuda,
and Barbados.
The
AOD
and exponent
values
are daily
averages
of the entire
data
set
for
each
site.
Two
characteristic
patterns
are evident
in the figures.
In both
the
Miami
and
Bermuda
20
plots, theexponentsarelow at low AOD values.Theexponentsgraduallyincreasein magnitude
with correspondingincreasesin the AOD
until
an
asymptote
in exponent
value
is reached.
The
Barbados
plot
shows
what
appears
to be an opposite
pattern;
decreasing
exponents
with
increasing
AOD
values.
In addition,
there
is a small
portion
of the Miami
plot that
shows
relatively
high
AOD
values
with
correspondingly
low
exponents,
similar
to the Barbados
plot.
The
plots
in Figures
6
are redisplayed
in Figures
7a,
7b,
and
7c after
applying
the correlational
procedure.
The
data
in
each
plot is segmented
according
to dominant
aerosol
type.
The
Miami
plot
shows
that
the majority
of sulfate
dominant
periods
have
high
AOD
values
(above
0.2)
and high
Angstrom
exponents
(generally
greater
than
1.0).
There
is one
low
exponent
sulfate
period
(below
0.5),
but
the AOD
is much
lower.
The
Miami
plot
also
shows
that
the sea-
salt and dust
dominant
periods
have
Angstrom
exponents
that
are mostly
below
0.5,
as mentioned
above.
However,
it is also
evident
that
several
of the dust
dominant
periods
also
have
high
AOD
values
(approximately
0.2),
while
there
is only
one sea-salt
dominant
period
with
a high
AOD.
The
Bermuda
plot
shows
a similar
trend,
as for
Miami,
for
the sulfate
dominant
periods.
The high
AOD
sulfate
periods
coincide
with
high
Angstrom
exponents
(approximately
1.0).
There
is also
one
sulfate
period
with
a low exponent
(below
0.5),
but this
period
also
corresponds
with
a
lower
AOD
value,
as
found
in
the
Miami
plot.
For
the
purposes
of
this
discussion,
we
are
concerned
only
with
strong
dominant
aerosol
occurrences
and
the
low
sulfate
AOD
periods
are
discarded.
The
Bermuda
sea-salt
dominant
periods
are also
similar
to the Miami
sea-salt
periods,
with
lower
AOD
values
(below
0.2)
and
low exponents
(below
1.0).
The dust
dominant
periods
in
Bermuda
correspond
with
higher
Angstrom
exponents
(mostly
between
0.5
and
1.0)
than
those
seen
for
Miami.
However,
the dust
dominant
AOD
values
for
Bermuda
are
much
lower
on
the
average
(mostly
below
0.!)
than
for Miami
dust dominant
periods.
The
Barbados
plot
shows
very
similar
sea-salt
dominant
period
behavior
as
that seen
for
Miami.
Sea-salt
periods
have
low
AOD
values
(mean
around
0.1)
and
low
exponents
(below
0.5).
The
dust
dominant_periods
for Barbados
show
a clear
pattem
of decreasing
Angstrom
exponents
with
increasing
AOD.
As in Miami,
high
dust
period
AOD
values
(above
0.2)
correspond
to low
21
exponents(below 0.5). The lower dust periodAOD values(near 0.1) have higher Angstrom
exponents
(above
0.5),
similar
to the dust
dominant
trend
shown
for Bermuda.
General
statements
may
be made
by combining
the results
from
each
site.
Figure
8 shows
the AOD
plotted
versus
the Angstrom
exponent
for each
dominant
aerosol
period
from
all the sites.
In general,
strong
non-sea-salt
sulfate
dominant
periods
are characterized
by AOD
values
from
0.2
to 0.3,
and Angstrom
exponents
from
1.0
to
1.5
in magnitude.
In contrast
to the sulfate
periods,
nearly
all sea-salt
dominant
periods
have
AOD
values
from
0.1
to 0.2,
and
Angstrom
exponents
from
0.0
to 1.0
in magnitude.
The
dust
dominant
periods
range
in AOD
from
below
0.1
to just
under
0.4 in value,
and
the majority
of Angstrom
exponents
range
from
0.0
to
1.0 in magnitude.
The
sea-salt
and
dust
dominant
periods
overlap
in both
AOD
and
Angstrom
exponent
values
for
most
ranges
in
question,
and
therefore
are
not
easily
distinguished.
Both
sea-salt
and
dust
dominant
periods
do not
show
the same
characterization
in AOD
and
Angstrom
exponent
as
do
sulfate
dominant
periods,
therefore
it is possible
to distinguish
strong
non-sea-salt
sulfate
dominant
periods
from
sea-salt
and
dust
periods
based
only
on
the
analysis
of
the
AOD
and
Angstrom
exponents.
The
dust dominant
periods
show
a characteristic
trend
in the AOD
and Angstrom
exponent.
Dust
dominant
periods
show
that
the Angstrom
exponents
are higher
(from
0.5
to 1.0)
for
lower
AOD
values
from
0.1
to 0.2,
and
that
the exponents
decrease
in magnitude
(to below
0.5)
as the
AOD
increases
to above
0.2
in value.
This
is equivalent
to saying
that
on
the
average,
the
dust
Angstrom
exponents
decrease
rapidly
as the dust
concentration
increases.
This
same
trend
does
not
occur
for
sea-salt
dominant
periods.
Therefore,
it is possible
to distinguish
strong
dust
periods
from
sea-salt
periods
because
the
sea-salt
AOD
values
do
not
normally
reach
above
0.2
in
magnitude.
AOD
values
above
0.2
and
corresponding
Angstrom
exponents
below
0.5
in
magnitude
are a strong
indication
of dust
dominance.
The
ability
to identify
strong
non-sea-salt
_sulfate
dominant
periods
over
the
ocean
by
analysis
of
only
the
AOD
and
Angstrom
exponent
has
been
demonstrated.
Also,
the
ability
to
distinguish
strong
dust
dominant
periods
from
normal
sea-salt
dominant
periods
has
also
been
22
shown. The characteristicAOD and Angstrom exponentvaluesthat may be used to identify
particularaerosolspeciesovertheoceanaresummarizedin Table6.
5.
Conclusions
The
results
of AOD
analysis
for Miami,
Bermuda,
and Barbados
over
approximately
a two
and a half year
period
of time
have
been
presented.
Together,
the three
sites
provide
good
spatial
coverage
for
the western
half
of the tropical
North
Atlantic
Ocean.
Also,
the time period
of
over
two
years
allows
for
good
temporal
coverage
which
is essential
for
proper
seasonal
studies.
The
results
obtained
from
this work
agree
well
with
other
AOD
data
sets
taken
in the
North
Atlantic
region.
For instance,
the seasonality
of average
AOD
values
exhibited
in the AVI-I
AOD
values
[Husar
et aL,
1997]
over
the Caribbean
and
along
the coast
of the Southeast
United
States
agrees
with
our data
set.
The
results
obtained
from
correlations
of
the
AOD
with
the
aerosol
concentrations
also
agree
well
with
work
done
by other
researchers
that involved
correlating
AOD
measurements
over
the ocean
with
air mass
trajectories.
Reddy
et al. [1990]
and Smirnov
et al. [1995]
have
shown
that
air mass
types
from
continental
northern
regions
(such
as the United
States)
produce
high
AOD
(approximately
0.2)
and
high
Angstrom
exponents
(> 1.0)
when
they
move
over
the ocean.
This
result
agrees
well
with
the AOD
and Angstrom
exponents
obtained
in this work
during
non-sea-salt
sulfate
dominant
periods.
Also,
AOD
and Angstrom
exponents
presented
by Smimov
et al.
[1995]
for
a tropical
maritime
air mass
(AOD
~ 0.15,
o_ ~ 0.42)
are similar
to those
found
in this
work
during
sea-salt
dominant
periods.
Finally,
AOD
and Angstrom
exponents
from
Reddy
et al.
[1990]
for air masses
from,
or influenced
by,
the Saharan
region
(AOD
- 0.39,
ot ~ 0.37;
AOD
~ 0.13,
ct - 0.77),
produce
AOD
andlAngstrom
exponents
similar
to those
found
in this
work
for
dust
dominant
periods.
23
Therearethreeprimaryconclusionsdrawn from this analysis.The first is thatthespectral
variationof theAOD overeachsitemayberepresentedbytheAngstrompowerlaw.Thesecondis
thatcharacteristicseasonalpatternsintheAOD ateachsitearepresentandarecausedprimarilyby
seasOnalchangesin thetypesandconcentrationsof theaerosolspecies.Finally, it is possibleto
distinguishbetweendifferentaerosolspeciesovertheoceanby analyzingthespectralvariationof
theAOD, particularlyduringstrongperiodsof dominancebyasingleaerosolspecies.
The recentlaunchof oceanviewing satellitesensorssuchasSeaWiFSand the upcoming
launchof theMODIS sensorontheEOS-AMsatelliteareusedto determinethe oceancolor by
measuringupweUingsunlightfrom theocean.However,thesun'sradiancemust passthroughthe
atmospherebeforereachingthesensor.Themeasuredradiancesmustbecorrectedfor atmospheric
effectsin orderto calculatetheoceancolor.An overviewof atmosphericorrectionalgorithmsmay
befoundin Gordon[1997].Thebasicprocedurefor bothsingleandmultiplescatteringalgorithms
involvesthe determinationof thecorrectionfactor e [Gordon, 1997] at blue wavelengthsafter
calculatingein the nearinfared(NIR). Thehigh degreeto which theAOD measurementsateach
siteresembledanAngstrompowerlaw in thevisible andNIR meansit is possiblethat,for ocean
color correction,the spectralvariationof e canbe modeledusingpower-law size distributions
[Chomkoand
Gordon,
1998].
The seasonal
variability
of the AOD
must
be understood
to properly
analyze
the
optical
data
and
when
using
AOD
for
other
secondary
purposes,
such
as
input
to
climate
models
and
atmospheric
correction
algorithms.
For
instance,
use
Of a typical
AOD
measured
during
a mid-
summer
Saharan
dust
p_sage,
would
not
be representative
of normal
year-round
conditions
over
the
tropical
North
Atlantic
Ocean."Thes.e
types
of
problems
may
be
avoided
by
proper
understanding
of the seasonal
AOD
changes.
For
instance,
it would
be best
to use
a typical
AOD
measured
duringmid-wihter
for climate
studies
and
correction
algorithms
used
to model
normal
open
ocean
conditions,
and to use a typical
AOD
from
mid-summer
to study
the pertubative
effects
of Saharan
dust
over
the North
Atlantic
Ocean.
Also,
the use of correct
AOD
values,
and
hence
the
24
correctaerosolmodel,basedon theunderstandingof seasonalchangesin theaerosolspecieswill
improvethedeterminationof theoceancolorcorrectionfactorE.
Themeasurementof spectralAOD from the groundis arelativelywell known procedure,
andsurfacebasedmeasurementsof the AOD over the oceanareincreasingin numberas more
attentionisfocusedonunderstandingtheclimaticeffectsof aerosolsonaglobalscale.Many of the
sensorson upcomingsatellitesarecapableof producingAOD valuesat, or near, the spectral
varietyof ground-basedsensors.Also,satelliteAOD measurementsareperformedonly for cloud-
freeareas,andtherefore,like surfaceAOD measurements,theresultingdatashouldnot beeffected
by cloudprocessingof theaerosol.Combinationof bothground-basedandsatellitespectralAOD
measurementswill producemuch neededglobal data of the AOD and Angstrom exponent,
particularlyovertheoceans.Theidentificationof specificaerosolspeciesbasedon theanalysisof
theAOD andAngstromexponenthasbeenshown in this study. The ability to identify aerosol
species
from
analysis
of other
ground-based
and
satellite
derived
spectral
AOD
and
the criteria
in
Table
6 will complement
other
current
satellite
aerosol
identification
programs,
such
as the TOMS
aerosol
index
product
[Herman
et al.,
1997].
Future
work
will
involve
examining
the
transport
patterns
and
source
locations
for
the
aerosol
species
investigated
in this
study.
Also,
the criteria
developed
in this
work
for
identifying
aerosol
species
based
on
measurements
of
the
spectral
AOD
(Table
6)
will
be
used
to
aid
in
analyzing
data
taken
at the AERONET
Cimel
sunphotometer
[Holben
et al.,
1994]
site
in the
Dry
Tortugas,
Florida.
25
Acknowledgements
The
authors
would
like to thank
the
following
people:
Albert
Chapin
and
Derrick
Snowden
for
help
in performing
sunphotometer
measurements
and
calibrations;
Miguel
Izaguirre
and
Tom
Snowden
for help
in an'anging
site
communications
and
logistics;
Hal
Mating,
and
Xu
Li for
providing
useful
information
on
aerosol
measurement
techniques
and various
aerosol
properties;
Joe Michalsky
and Lee
Harrison
for providing
much
needed
information
and suggestions
on performing
shadowband
measurements;
Alexander
Ignatov
for
providing
information
on the procedure
used
to correct
the
shadowband
calibrations;
Jim
Schlemmer
for
providing
the algorithm
to the
SUNY
Sliding
Window
Optical
Depth
Procedure;
and
AEROCE
site
operators
Anne
Glasspool,
Cornealius
Shea,
Matthew
Krasowski,
Megan
McKay,
H. B. Gooding,
and
Dr.
F. Henriques
for
help
in maintaining
the on-site
insmaments.
The
optical
measurements
were
funded
by NASA
contract
#NA55-
31363,
and
NOAA
grant
#NA16RCO471-01.
The
aerosol
chemistry
measurements
were
made
under
National
Science
Foundation
grants
ATM9013125
and ATM9414846.
References
d'Almeida,
G.A.,
P. Koepke,
and
E.P.
Shettle,
Atmospheric
Aerosols:
Global
Climatology
and Radiative
Characteristics,
A. Deepak
Pub.,
Hampton,
Virginia,
1991.
Angstrom,
A., The
parameters
of atmospheric
turbidity,
Tellus,
16, 64-75,
1964.
Arimoto,
R, R.A.
Duce,
D.C.
Savoie,
J.M.
Prospero,
R. Talbot,
J.D.
Cullen,
U. Tomza,
N.F.
Lewis,
and
B.J.
Ray,
Relationship
among
aerosol
constituents
from
Asia
and
the North
Pacific
during
PEM-West
A, J. Geophys.
Res.,
101,2011-2023,
1996.
Charlson,
R.J.,
S.E.
Schwartz,
J.M.
Hales,
R.D.
Cess,
J.A.
Coakley
Jr.,
J.E.
Hansen,
and
D.J.
Hofmann,
Climate
forcing
by anthropogenic
aerosols,
Science,
255,
423-
430,
1992.
Chomko,
R., and H. Gordon,
Atmospheric
correction
of ocean
color
imagery:
Use
of the
Junge
power-law
aerosol
size distribution
with
variable
refractive
index
to handle
aerosol
absorption,
(In Press)
Appl.
Optics,
1998.
Dutton,
E.G.,
P. Reddy,
S. Ryan,
and J. DeLuisi,
Features
and effects
of aerosol
optical
depth
observed
at Mauna
Loa,
Hawaii:
1982-1992,
J. Geophys.
Res.,
99,
8295-
8306,
1994.
Gordon,
H.R.,
Atmospheric
correction
of ocean
color
imagery
in the Earth
Observing
System
era,
J. Geophys.
Res.,
102,
17081-17106,
1997.
Hansen,
J.E.,
and L.D.
Travis,
Light
scattering
in planetary
atmospheres,
Space
Sci.
Reviews,
16, 527-610,
1974.
Harrison,
L., J. Michalsky,
and J. Bemdt,
The
Automated
multi-filter
rotating
shadow-
band
radiometer:
An instrument
for optical
depth
and radiation
measurements,
Appl.
Optics,
33,
5118-5125,
1994.
Harrison,
L., and J. Michalsky,
Objective
algorithms
for the retrieval
of optical
depths
from
ground-based
measurements,
Appl.
Optics,
33, 5126-5132,
1994.
Herman,
J.R.,
P. Bhartia,
O. Torres,
C. Hsu,
C. Seftor,
and
E. Celarier,
Global
distribution
of UV-absorbing
aerosols
from
Nimbus-7/TOMS
data,
J. Geophys.
Res.,
102,
16911-16922,
1997.
Holben,
B.N.,
T.F.
Eck,
I. Slutsker,
D. Tanre,
J.P.
Buis,
A.
Setzer,
E. Vermote,
J.A.
Reagan,
_Y.j. Kaufman,
T. Nakajima,
and
F. Lavenu,
Multi-band
automatic
sun
and sky scanning
radiometer
system
for measurement
of aerosols,
Proc.
Sixth
InternatiOnal
Symposium
Physical
on Measurements
and Signatures
in Remote
Sensing;.75-83,
1994
j
".
Hoppel,
W.A.,
J. Fitzgerald,
G.
Frick,
R. Larson,
and
E. Mack,
Aerosol
Size
Distributions
and:.Optical
Properties
Found
in the Marine
Boundary
Layer
Over
the
AtlanticOcean,J.
Geophys.
Res.,
95;
3659-3686,
1990.
Husar,R.B., J.M. Prospero,L.L. Stowe,Characterizationof troposphericaerosolsover
theoceanswith the NOAA
advanced
very high
resolution
radiometer
optical
thickness
operational
product,
J. Geophys.
Res.,
102,
16889-16909,
1997.
Ignatov,
A, Personal
Communication,
NOAA
NESDIS
ORA,
Washington
D.C.,
20233,
1996.
Kasten,
F., and A.T.
Young,
Revised
optical
air mass
tables
and approximation
formula,
Appl.
Optics,
28, 4735-4738,
1989.
Kaufman,
Y.J.,
Aerosol
optical
thickness
and
atmospheric
path
radiance,
J. Geophys.
Res.,
98,
2677-2692,
1993.
Kaufman,
Y.J.,
A. Gitelson,
A. Karnieli,
E. Ganor,
R.S.
Fraser,
T. Nakajima,
S.
Mattoo,
and
B.N.
Holben,
Size
distribution
and scattering
phase
function
of
aerosol
particles
retrieved
from
sky brightness
measurements,
J. Geophys.
Res.,
99,
10341-10356,
1994.
King,
M.D.,
D. Byrne,
B. Herman,
and J. Reagan,
Aerosol
size
distributions
obtained
by inversion
of spectral
optical
depth
measurements,
J. Atmos.
Sci.,
35,
2153-
2167,
1978.
King,
M.,
D. Byrne,
J. Reagan,
and B. Herman,
Spectral
variation
of optical
depth
at
Tucson
Arizona
between
August
1975
and December
1977, J. Appl.
Meteorol.,
19,
16,639-16,650,
1980.
Klenk,
K., P. Bhartia,
E. Hilsenrath,
and A. Fleig,
Standard
ozone
profiles
from
balloon
and
satellite
data
sets,
J. Clim.
Appl.
Meteorol.,
22, 2012-2022,
1983.
Li, X., H. Maring,
D.
Savoie,
K. Voss,
and J. Prospero,
Dominance
of Mineral
Dust
in
Aerosol
Light-Scattering
in the North
Atlantic
Trade
Winds,
Nature,
380,
416-419,
1996.
Prospero,
J.M.,
Saharan
dust
transport
over
the north
Atlantic
Ocean
and Mediterranean,
Chapter
inThe
Impact
of Desert
Dust
From
Northern
Africa
Across
the
Mediterranean,
Kluwer
Academic
Publishers,
Oristano,
Sardinia,
1995.
Reddy,
P.J.,
F.W.,
Kreiner,
J.J.
DeLuisi,
and
Y. Kim,
Aerosol
Optical
Depths
Over
the
Atlantic
Derived
From
Shipboard
Sunphotometer
Observations
During
the
1988
Global
Change
Expedition,
Global
Biogeochem.
Cycles,
4, 225-240,
1990.
Savoie,
D., and J. Prospero,
Particle
Size
Distributions
of Nitrate
and Sulfate
in the
Marine
Atmosphere,
Geophys.
Res.
Lett.,
9,
1207-1210,
1982.
Savoie,
D., Nitrate
and:nomsea-salt
sulfate
aerosols
over
major
regions
of the world
•oceans:
concentrations,
sources,
and fluxes,
Ph.D.
Dissertation,
Rosenstiel
School
of Marine
and
Atmospheric
Science,
University
of Miami,
4600
Rickenbacker
Causeway,
Miami,
F1, 33149,
1984.
Savoie,
D., J.M.
Prospero,
and E.S.
Saltzman,
Non-sea-salt
sulfate
and nitrate
in trade
wind
aerosols
at Barbados:
evidence
for long-range
transport,
J. Geophys.
Res.,
94,
5069-5080,
1989.
Schlemmer,
J., Personal
Communication,
Atmospheric
Sciences
Research
Center,
SUNY
Albany,
NY,
12205,
1996.
Shaw,
G.E.,
Aerosols
at Mauna
Loa:
Optical
properties,
J. Atmos.
Sci.,
36,
862-869,
1979.
Smimov,
A., A. Royer,
N.
O'Neill,
and
A. Tarussov,
A study
of the link
between
synoptic
air mass
type
and atmospheric
optical
parameters,
J. Geophys.
Res.,
99,
20967-20982,
1994.
Smimov,
A., Y. Villevalde,
N. O'Neill,
A. Royer,
and A. Tarussov,
Aerosol
optical
depth
over
the oceans:
Analysis
in terms
of synoptic
air mass
types,
J.
Geophys.
Res.,
100,
16639-16650,
1995.
Van
de Hulst,
H. C., Light
Scattering
by Small
Particles,
Dover
Publications
Inc.,
New
York,
NY,
1981.
Villevalde,
Y.V.,
A.V.
Smirnov,
N.T.
O'Neill,
S.P.
Smyshlyaev,
and
V.V.
Yakovlev,
Measurement
of aerosol
optical
depth
in the Pacific
Ocean
and the North
Atlantic,
J.
Geophys.
Res.,
99,
20983-20988,
1994.
Table1
Location
Miami, Florida
Lat: 25.75N
Lon: 80.2W
Altitude:SeaLevel
Bermuda
Lat: 32.38N
Lon: 64.7W
Altitude:SeaLevel
Barbados
Lat: 13.18N
Lon: 59.43W
Altitude:SeaLevel
TimePeriod
August1993
to
December1995
August1993
to
December1995
August1993
to
December1995
Channel
Number
1
2
3
4
5
6
7
8
9
Table2
Sunphotometer
Filter Wavelength(nm)
380.2
412.2
440.5
501.8
551.2
675.2
777.9
861.8
1025.9
Shadowband
FilterWavelength(nm)
Unfiltered(Broadband)
410
500
610
665
860
940
Location
Miami
Bermuda
Barbados
nssSulfate
(l.tg/m3)
2.202
1.783
0.765
Table3
Nitrate
(btg/m 3)
2.090
0.979
0.599
Sea Salt
(lag/m 3)
8.892
11.412
19.997
Dust
(btg/m 3)
5.633
2.056
14.864
Month-
Year
Aug-93
Sep-93
Oct-93
Nov-93
Dec-93
Jan-94
Feb-94
Mar-94
Apr-94
May-94
Jun-94
Jul-94
Aug-94
Sep-94
Oct-94
Nov-94
Dec-94
Jan-95
Feb-95
Mar-95
Apr-95
May-95
Jun-95
Jul-95
Aug-95
Sep-95
Oct-95
Nov-95
Dec-95
Total
Table4a MonthlyMiamiAOD Results
m
0.114
0.024
0.142
0.068
0.142
0.020
0.121
0.034
0.121
0.030
0.120
0.035
0.151
0.073
0.166
0.038
0.243
0.091
0.163
0.058
0.234
0.055
0.075
0.013
0.127
0.058
0.119
0.054
0.093
0.039
0.097
0.041
0.091
0.032
0.115
0.049
0.118
0.020
0.143
0.037
0.288
0.078
0.165
0.053
0.148
0.077
0.099
0.032
0.1i41
. _:0.050
0.070
0.069
0.080
0.050
0.071
0.083
0.069
0.104
0.111
0.128
0.209
0.059
0.079
0.056
0.049
0.093
0.059
0.062
0.065
0.094
0.108
0.103
0.100
o o45
0.084
0.022
0.020
0.011
0.013
0.005
0.035
0.030
0.035
0.033
0.057
0.034
0.011
0.026
0.027
0.014
0.019
0.017
0.024
0.010
0.019
0.023
0.030
0.043
,
0.016
0:034
tZ
0.734
0.976
0.833
1.269
0.740
0.566
1.063
0.706
1.098
0.217
0.148
0.347
0.524
0.987
0.762
-0.246
0.443
0.846
0.768
0.480
1.294
0.494
0.347
1.110
0.684
0.228
0.374
0.001
0.056
0.398
0.375
0.240
0.553
0.319
0.457
0.100
0.319
0.773
0.509
0.713
0.820
0.579
0.501
0.284
0.311
0.199
0.565
0.557
0.565
0.367
Month-
Year
Aug-93
Sep-93
Oct-93
Nov-93
Dec-93
Jan-94
Feb-94
Mar-94
Apr-94
May-94
Jun-94
Jul-94
Aug-94
Sep-94
Oct-94
Nov-94
Dec-94
Jan-95
Feb-95
Mar-95
Apr-95
May-95
Jun-95
Jul-95
Aug-95
Sep-95
Oct-95
Nov-95
Dec-95
Total
Table4b MonthlyBermudaAOD Results
0.161
0.161
0.117
0.160
0.O99
0.102
0.157
0.169
0.134
0.270
0.116
0.126
0.100
0.075
0.068
0.064
0.081
0.104
0.091
0.172
0.168
0.134
"Lr.
0.129
0",.¢
0.043
0.041
0.025
0.037
0.012
0.023
0.041
0.050
0.047
0.125
0.072
0.131
0.061
0.041
0.021
0.028
0.035
0.027
0.050
0.077
0.057
0.069
0.047
0.104
0.015
0.103
0.020
0.081
0.018
0.102
0.009
0.069
0.011
0.079
0.005
0.099
0.026
0.105
0.022
0.085
0.030
0.170
0.072
0.081
0.045
0.097
0.125
0.049
0.035
0.060
0.015
0.077
0.016
0.085
0.033
0.075
0.023
0.059
0.025
0.068
0.023
0.095
0.034
0.088
0.030
0.072
0.024
0.086
0.025
C_
_tx
0.603
0.290
0.622
0.232
0.540
0.214
0.614
0.304
0.531
0.298
0.322
0.372
0.666
0.268
0.667
0.415
0.655
0.259
0.599
0.233
0.474
0.179
0.424
0.334
1.058
0.746
0.288
0.663
-0.209
0.351
-0.920
0.973
-0.236
0.817
0.697
0.616
0.108
0.416
0.818
0.187
0.851
0.318
0.558
0.659
0.442
0.433
Month-
Year
Aug-93
Sep-93
Oct-93
Nov-93
Dec-93
Jan-94
Feb-94
Mar-94
Apr-94
May-94
Jun-94
Jul-94
Aug-94
Sep-94
Oct-94
Nov-94
Dec-94
Jan-95
Feb-95
Mar-95
Apr-95
May-95
Jun-95
Jul-95
Aug-95
Sep-95
Oct_95
Nov-95
Dec-95
Total
:.
Table
4c
Monthly
Barbados
AOD
Results
m
0.078
0.033
0.072
0.019
0.064
0.006
0.084
0.029
0.129
0.057
0.078
0.058
0.184
0.157
0.201
0.151
0.247
0.095
0.362
0.128
0.210
0.067
0.169
0.090
0.257
O. 107
0.144
0.060
0.200
-
+0.165
0.085
0.059
0.029
0.055
0.021
0.051
0.015
0.071
0.022
0.097
0.055
0.064
0.056
0.167
0.156
0.167
0.102
0.224
0.084
0.267
0.111
0.132
0.055
0.118
0.076
--
°
0.260
O. 108
0.147
0.063
0.176
0.137
0.073
wt_
0.442
0.483
0.360
0.262
0.508
0.605
0.275
0.256
0.148
0.504
0.731
0.640
0.014
0.047
0.138
0.361
0.166
0.256
0.322
0.180
0.446
0.674
0.245
0.245
0.139
0.225
0.211
0.287
0.059
0.187
0.220
Location- Dominate
Aerosol
Table5
o_
Miami - nssSulfate
Bermuda- nssSulfate
1.216
1.012
0.209
0.292
Miami - SeaSalt
Bermuda- SeaSalt
Barbados- SeaSalt
0.205
0.580
0.406
0.202
0.376
0.080
Miami - Dust
Bermuda- Dust
Barbados- Dust
0.280
0.585
0.339
0.235
0.127
0.268
AOD (500rim)
0.0
-
0.2
>
0.2
Table
6
Angstrom
Exponent
0.0
-
0.5
>
0.5
0.0
-
0.5
0.5-
1.0
>
1.0
Primary
Aerosol
Species
sea-salt
or
dust
indeterminate
dust
indeterminate
non-sea-salt
sulfate
Figure
la
Channel
Wavelength
(nm)
I"1
380.2
•
412.2
[]
440.5
501.8
[]
551.2
5.5
O
v
4.5_
4
w
3.5
--
'
I
'
I
January
1993
to January
1995
Channel
Wavelength
(nm)
[]
675.2
•
777.9
[]
861.8
1025.9
Figure
la.
Calibration
history
for the
Miami
sunphotometer.
The
solid
line
is the
fit
to the
Langley
and cross-calibrations.
The dotted
line
is the fit to the error
corrected
calibrations.
Figure lb
o
v
,-I
o
gtl
v
,-.1
O'
-1-
-2
-------_.•_____
m
w
v
'
I
'
I
--
ED
--
--
--
"
'
I_
'
I
January
1993
to January
1995
Channel
Wavelength
(nm)
[]
380.2
•
412.2
I_
440.5
,a,
501.8
i
551.2
Channel
Wavelength
(nm)
[]
675.2
•
777.9
[]
861.8
'IF
1025.9
Figure
lb.
Calibration
history
for the Bermuda
sunphotometer.
The
solid
line
is the fit to the
Langley
and cross-calibrations.
Thedotted
line
is the fit to the error
corrected
calibrations.
Figure
lc
1
v
,-1
0
-1
-2
'
I
'
I
Channel
Wavelength
(nm)
[]
380.2
•
412.2
[]
440.5
41,
501.8
[]
551.2
o
v
/
¢ Jall.Ual'y
1993 to January
1995
Channel
Wavelength
(nm)
[]
675.2
•
777.9
[]
861.8
6
1025.9
Figure
lc.
Calibf_itiod
history
for the Barbados
sunph0tometer.
The solid
line
is the
fit to the
Langley
and cross-caiibrations.
The d°tted
lihe
is the fit to the error
corrected
calibrations.
u
u
Figure
2a
6000
5500_
ii 11
0
4000
3500
3000
2500
2000
1500
1000
34400
34500
34600
34700
34800
Day SmceJanl,
1900
34900
35000
35100
n
In
O
e.
O
6000
5000 1
4000
.
[]
,.o
2000
•
1°°i
t
[]
[]
I
I
I
I
I
I
34400
34500
34600
34700
34800
34900
35000
35100
Uncorrected
Solar Constant
(Ch2, 410
nm)
Error Corrected
Solar Constant
(Ch2,
410 nm)
Uncorre.eted
SolarConstant(Ch3,500 rim)
ErrorCorrectedSolarConstant(Ch3,500 nrn)
UncorrectedSolarConstant(Ch4,610 nan)
ErrorCorrectedSolarConstant(Ch4,610 nm)
Uncorrected
Solar Constant
(ChS, 665
rim)
Error Cor_.cted
Solar Constant
(Ch5, 665 nm)
Uncorrected
Solar Constant
(Ch6, 860 am)
ErrorCorrectedSolarConstant(Ch6, 860 am)
Figure
2a. Calibration
history
for the Miami
shadowband
(May
1994 to December
1995).
The solid
line is the fit to the OLRA
Langley
calibrations.
a
5500
Figure
2b
O
2500-_
rt
2000-
[]
1500-
1000
34400
34500
34600
34700
34800
34900
35000
35100
6000
5500_
5000 1
4500-
I
Day Since Jan
1, 1900
[]
4000
°
3500-
m
3000-
O
2500
-
a
•
2000-
1500-
1000
34400
34500
34600
34700
34800
34900
35000
35100
Uncorrected
Solar Con*tam
(Ch2,
410 nm)
Error Corrected
Solar Constant
(Ch2, 410
nm)
Uncorrected
Solar Constant
(Ch3,
500 nm)
Error Corrected
Solar
Constant
(Ch3, 500
nm)
Uncorre_'ted
Solar
Constant(Ch6, 860 rim)
ErrorCorrectedSolarConstant(Ch6,860 nm)
Uncorrected
Solar Constant (Oh4,
610 urn)
Errcz Corrected Solar Commmt
(Ch4, 610
tun)
Uncorr_t_d SolarCo_t_t
(ChS,665 rim)
ErrorCorrectedSolarConstant(OaS,665 nm)
Figure
2b. Calibration
history
for the Bermuda
shadowband
(May
1994 to December
1995).
The solid
line
is the fit to the OLRA
Langley
calibrations.
a
@
6000
5000
4000
3000
2000
I000,
0
m
'I
Figure
2c
A
v
0
U
U
34850
34865
34880
e,t
v
Av
6000
I
34895
13
m
o
m
It
O
"_
w
34910
34925
34940
Day Since Jan 1, 1900
Uncorrected
Solar Constant
(Ch2,410
am)
Error Corrected
Solar Constant
(Ch2, 410
nm)
Uncorrected
Solar
Constant
(Ch3,500
am)
Error Corrected
Solar Constant
(Ch3,
500 nm)
Uncorrected
Solar Constant
(Ch4,
610 nm)
Error Corrected
Solar Constant
(Ch4,
610 nm)
.q
5000-
4000-
3000-
2000-
I000-
0
0
v
v
_I.
¥
o
I
I
i
I
I
34850
34865
34880
34895
34910
34925
34940
13
[]
0
Uncorrected
Solar Constant
(Ch5,665
rim)
Error Corrected
Solar Constant
(Ch5, 665
rim)
Uncorrected
Solar Constant (Ch6.
860 am)
En'or Corrected
Solar Commxat (Ch6, 860
rim)
Figure
2c
Calibration
history
for the Barbados
shadowband
(June
1995 to August
1995)
The solid
line is the fit to the OLRA
Langley
calibrations.
b-,.
,,.I _'<
%0
>
0
8
oN
N
g.
e_
o
o
Aug93
Sep93
0c193
Nov93
Dec93
J_94
Feb94
M_a94
A_r94
May94
Jim94
&_94
Aug94
-
Sep94 =
0ct94 -
Nov94.
Dec94 -
Jang5 -
Feb9S -
Mon_Aly
Mew
AOD
(500
nm)
I
!
I
.........................
....
-1=04- ................
- - - -b....--@ ..---.,4 ............
.......
1"@4...............
....
t--0"'1- ...............
.......
b-O-4
............
..........
I-------_
' ....
....
mm
m ...........
..........
!
;
I.......
---O
....................
- - - _"_0'_"4-
.............
- - - _O--.-I
..............
--I-'O--I.................
--4--O'--I-................
---I-O-4-.................
---I'--_'---4
...............
N_S
......
_._-
................
ApzS5-
-_-0--1
..............
May95
Jun95
............
I
_,
:--
.........................
......
_O--,,,-I
...........
- - - -II_-.,O-_
...........
.........................
- - - I-g--I
.................
J_95
Aug95
-
Sep95
0ct95
Nov95
Dec95
"rl
i..=h
t_
25
Figure
3b
D
@
m
Monthly Mean nss Sulfalz Conc. (ILtg/m3)
Monthly Mean Sea Salt Conc. (lag/m3)
Monthly Mean Dust Conc. (lag/m3)
Figure
3b.
Miami
monthly
mean
aerosol
concentrations
from
August
1993
to
December
1995.
m
O
4:,
%O
>
0
o
Q
Monthly Mean AOD
(500 nm)
O
_
O
O
]
I
I
Aug93 ...........................
Sep93 .- ......
_q_-4
............
0c_93 .........
I---O--"4............
Nov93
-1...... _
................
Dec93 .........
_"-0-4- ............
Jan94.......
•
..................
Feb94 ...... I-O-4.................
Ma,94
Apf94 •
May94
_94
Jui94.
A 94
Sep94.
0c_94-
Nov94 .- - -_
...................
Dec94. -- I-O-I- ...................
Maygs
Jun95
Ju195
Aug95
Sep95
Oct95
Nov95
Dec95
.......
-I.,-O-,4-............
.......
-4"-0"-4
...........
..... 4---0--4 ..............
---I
_
|
.............
;
I.........
--
I
_IL
| ...............
--I'-O"4
..................
- -I'"O-4 ..................
"11
_.._°
4_
Figure
4b
5
i
i
i
i
i
i
i
i
"_20
,
,
,
I
O
I
I
I
I
I
I
O_
I
I
I
15.
1
'
'
I
I
I
0,0.,i.r
t
t
I
r
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
:
........
I
I
I
I
I
I
I
I
I
I
........
,_0t_
',
,,
,
n
'
0
L
InP
mill
him
n_P
lIP
wlP"
_1_
Iqlm
'wP
_
I_lW
mlpW
I
I
|
I
I
J
Date
0
6
[]
Monthly
Mean
nss Sulfate
Conc.
(_tg/3)
Monthly
Mean
Sea Salt Conc.
(lag/m3)
Monthly
Mean
Dust Conc.
(laghn3)
Figure
4b.
Bermuda
monthly
mean
aerosol
concentrations
from
August
1993
to December
1995.
w=*
I_
o
L_
8
o
Aug93
Sep93
0ct93
Nov93
Dec93
J'_a94
Feb94
Max94
ApI94
May04
Jtta94
Monthly Mean AOD
(500 nm)
O
O
O
I
I
I
........................
........................
--I-O--I.................
--I-O-I-
..................
- - -I-O'I .................
_ _ _ _b--Q.===l_ ............
:
_
i................
I
_
|
--
I -
........
,
_.''--'=I - -
O
Ju194
...............
!
Aug94 _- ........
:
;
! .......
[
i
!
s_p94"1-.... :
;
I.........
I
oo,94
f .........................
I
""'_"t-.........................
I
De¢94
J- .........................
i
Jan95 ...........................
Feb95 ...........................
M_95
...........................
Ap_5-
Ma795
..........................
JuR95
.........
I
I
I--
JR195
..... I""--_ _"'I ............
Aug95 .............
Sep95 ...........................
0c195.
Nov95.
Dec95.
C_
Ln
Figure
5b
0
....
,
,
,
....
,
....
,
.........
,
,
i
r
[
[
i
i
i
i
i
i
i
[
i
i
i
I
i
i
i
_
J
i
i
i
i
i
5
i
i
i
30-1:
I : :
I ', ,_ : :
: :
: :
: :
: : : ', : :
', : :
:
', : ', :
/'
.........
E
...........
/.,
......
J
i
i
_
i
I
I
I
J
t
I
I
_1,
f
i
_
_
I
I
I
I
I
i
I
E
i
I
i
I
I
|1
I
I
I
!
I
I
I
I
[]
,11.
W
I
I
i
I
i
t
I
I
i
I
_
t
i
i
J
t
I
20-1,
,
,
_
_
,
,
,
o
,
_
,
,
_
........
-r
......
5__
l
D
,
,
i
,
_
,
,
,
,
,
,
_.-
_
_
,
,
,
,
,
,
_
,
i
t
,
,
,
,
i
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
t
I
lO4 ....
,
, U!a
...........
, ,
,
,
,
,
....
i
i
I
i
I
I
I
i
I
i
I
I
I
I
i
i
i
i
1
i
I
I
i
i
I
I
i
I
i
I
I
I
I
I
I
i
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
l
I
I
I
I
I
I
I
I
I
I
I
I
I
i
I
I
I
I
i
I
I
I
I
I
I
I
i
i
e2
0
Monthly
Mean
ms Sulfate
Conc.
(lag/m3)
Monthly
Mean
Sea Salt Conc.
(lag/m3)
Monthly
Mean
Dust Conc.
(_tg/m3)
Figure
5b.
Barbados
monthly
mean
aerosol
concentrations
from
August
1993
to December
1995.
2
1.5
0.5
-0.5
-1
Figure6a
i
rr-I
!
0.1
I
I
I
0
0.2
0.3
0.4
0.5
0.6
AOD
(500
nm)
Figure
6a. Miami
daily
averaged
AOD
(500
rim)
versus
the Angstrom
exponent.
Figure
6b
0.5
....
o--P ..... ° .................
0
0.1
0.2
0.3
0.4
0.5
0.6
AOD
(500
nm)
Figure
6b. Bermuda
daily
averaged
AoD
(500
nm)
versus
the Angstrom
exponent.
Figure6c
21.,4 .........................................
1
i 1°°o
..............
-0.5]
-1
.... _O__til__
, ..........................................
,
,
,
,
__[
0
0.1
0.2
0.3
0.4
0.5
0.6
AOD
(500
nm)
Figure
6c. Barbados
daily
averaged
AOD
(500
nm)
versus
the Angstrom
exponent.
Figure
7a
2
1.5_
.........
ti ..........
I_ ...................
--- .....
-]
1 -1 ...................
10-i::i_- .........
O
Angstrom
Exponent
(laSS Sulfate)
_
0.5J
.... _1_-
_
...................
_,
Angstrom
Exponent
(Sea Salt)
"
0 tliiii:i!i_-_--ii!iiiiiiiiiiiiiiii!i
!
"Angstr°m
Exp°nent
(Dust)
-0.5
............
-1
0
0.1
0.2
0.3
0.4
0.5
0.6
AOD (500 nm)
Figure
7a. Dominant
aerosol
correlation
results:
Miami
daily
averaged
AOD
(500
nm)
versus
the
Angstrom
exponent.
1.5
0.5
-0.5
-1
Figure7b
-
[
- ii__i__!iii_iiiiiiiii
i i .....
!- ......
AngstromAngstr°m
Exponent
(nSSExponent(Sea Sulfate)salt)
........
'_ ..........................
[]
Angstrom
Exponent
(Dust)
.....
_@_...................................
_ ......
.-
@
I
I
!
I
I
0
0.1
0.2
0.3
0.4
0.5
0.6
AOD
(500 rim)
Figure
7b. Dominant
aerosol
correlation
results:
Bermuda
daily
averaged
AOD
(500
nm)
versus
the Angstrom
exponent.
Figure
7c
2
_,
1
.... t&_.--l_
......................
[
@
Angstrom
Exponent
(Sea
Salt)
[]
0.5
-_,_ .... -_---_
.......
-I_ .......
[
[]
Angstrom
Exponent
(Dust)
0- -----_...._----_--_---_
..... _t..... _........
-0.5
.......................................................
-1
I
I
I
I
I
0
0.1
0.2
0.3
0.4
0.5
0.6
AOD
(500 nm)
Figure
7¢. Dominant
aerosol
correlation
results:
Barbados
daily
averaged
AOD
(500
rim)
versus
the
Angstrom
exponent.
Figure8
2
u
°5
u
................................................................
El
[]
1 - ----Ei-iD ............ __
_____ .........
Angstrom
Exponent
(ms
Sulfate)
0.5 q .... 8_i
--_--
Angstrom
Exponent
(Sea Sal0
6
#
0,
_
[]
(Dust)
•./
t_I_
_i!,_
_
roll_
_
--
Angstrom
Exponent
°1......_-°_ ....._w---_.....,_.........................
/
1M
-0.5 _ ...... 9_........................................................
/
6'['
-1/
,
_
.
_
J
0
0.1
0.2
0.3
0.4
0.5
0.6
AOD (500 rim)
Figure
8. Dominant
aerosol
correlation
results:
Combined
Miami,
Bermuda,
and Barbados
daily
averaged
AOD
(500
nm)
versus
the Angstrom
exponent.
Draft:
for
Tellus
ACE-2
special
issue
APPENDIX
5
Ground-based
Lidar
Measurements
of Aerosols
During
ACE-2:
Instrument
Description,
Results,
and
Comparisons
with
other
Ground-based
and
Airborne
Measurements
EUsworth
J. Welton
l, Kenneth
J. Voss
l, Howard
R. Gordon
1, Hal
Mating
2,
Alexander
Smimov
3,Brent
Holben
4, Beat
Schmid
5, John
M. Livingston
6,
Philip
B. Russell
7, Philip
A. Durkee
s, Paola
Formenti
9, Meinrat
O. Andreae
9
1.
University
of Miami,
Physics
Department,
Miami,
FI
2.
University
of Miami,
Rosenstiel
School
of Marine
and Atmospheric
Science,
Miami,
F1
3.
Science
Systems
and Applications,
Inc.
- NASA
/ GSFC,
Greenbelt,
MD
,.
4.
NASA
/ GSFC,
Greenbelt,
MD
5. Bay
Area
Environmental
Research
Institute,
San Francisco,
CA
6.
SKI International,
Menlo
Park,
CA
7.
NASA
Ames
Research
Center,
Moffett
Field,
CA
8.. Naval
Postgraduate
School,
Monterey,
CA
9.
Max
Planck
Institute
for Chemistry,
Biogeochemistry
Dept.,
PO 3060,
55020
Mainz,
Germany
Draft
version
12-09-98
1
Draft:
for
Tellus
ACE-2
special
issue
Abstract:
A micro-pulse
lidar
system
(MPL)
was
used
to measure
the vertical
and horizontal
distribution
of aerosols
during
the Aerosol
Characterization
Experiment
2 (ACE-2)
in June
and
July
of
1997.
The
MPL
measurements
were
made
at the
Izafia
observatory
(IZO),
a
weather
station
located
on a mountain
ridge
(28018 , N,
16030 , W,
2367
m
asl)
near
the
center
of the island
of
Tenerife,
Canary
Islands.
The
MPL
was
used
to acquire
aerosol
backscatter,
extinction,
and
optical
depth
profdes
at
IZO.
System
tests
and
calibration
procedures
are discussed,
and
analysis
of aerosol
optical
prof'des
acquired
during
ACE-2
are presented.
The
optical
characterization
of
upslope
aerosol
layers
normally
present
at
IZO
(the
background
aerosol),
and
of a Saharan
dust episode
during
the
middle
of ACE-2,
are presented.
Comparisons
of the
MPL
data
with
data
from
other
co-located
instruments
are
also
presented.
The
comparisons
showed
good
during
both
the clean,
upslope
period
and
the dust
episode.
Finally,
results
show
the possible
perturbation
of the bottom
of the
dust
layer
by upslope
winds
from
the mountain.
Draft
version
12-09-98
2
Draft: for
Tellus
A CE-2
special
issue
1.
Introduction
The
Aerosol
Characterization
Experiment
2 (ACE-2)
ran
from
June
16,
1997
to
July
25,
1997.
The
purpose
of ACE-2
was
to study
the radiative
properties
and
physical
characteristics
of anthropogenic
aerosols
from
Europe,
and dust
aerosols
from
Africa,
as
they
are transported
across
the North
Atlantic
Ocean.
An
overview
of
ACE-2
operations
and specific
activities
can be found
in Raes
et al. (this
issue).
One
of the ACE-2
activities
was
the
"clear
sky
column
closure
experiment"
(CLEARCOLUMN)
and
was
aimed
at
simultaneous
measurements
of aerosol
properties
using
a variety
of different
platforms
in
order
to assess
the aerosol
direct
radiative
forcing
(Heintzenberg
and
Russell,
this issue).
The
work presented
in this paper was
part of the CLEARCOLUMN
effort
during
ACE-2.
This
paper will
focus
on lidar measurements
of the vertical
and horizontal
structure
of aerosols
surrounding
the Izafia observatory
(IZO)
on
Tenerife,
Canary
Islands
during
ACE-2.
IZO
is located
on a mountain
ridge
(28018 ' N,
16030 , W,
2367
m asl)
near
the
center
of
the
island
and
has
often
been
used
as
a site
for
the
study
of
various
aerosol
properties
(Prospero
et al.,
1995;
Raes
et al.,
1997;
Smirnov
et al.,
1998).
However,
lidar
measurements
at IZO have
not been
made prior to this
study.
Thus,
accurate
knowledge
of
the spatial
distribution
of aerosols
in the atmosphere
surrounding
IZO
is needed
in order
to
correctly
apply
the
in-situ
IZO
data to studies
of
aerosols
in the
Eastern
North
Atlantic
region.
The
lidar
data
are
used
to
assess
the
vertical
distribution
and
horizontal
homogeneity
of the aerosols
in the atmosphere
around
IZO (10-30
km radius).
In
addition
to
the
lidar
observations,
in-situ
aerosol
mass
concentration
and
extinction
coefficient
measurements
were
made
at IZO.
These
in-situ
measurements
were
used
to aid
in
the
calibration
of
the
lidar
system
(as
described
in
Section
3),
and
in
comparisons
with
the near-IZO-range
lidar data.
Sunphotometer
measurements
were
also
made
at IZO
in order
to supply
spectral
aerosol
optical
depth
(AOD)
measurements
for
Draft
version
12-09-98
3
Draft:for
Tellus
ACE-2
special
issue
CLEARCOLUMN
efforts
and
for
use
in
a lidar
inversion
algorithm
(Appendix).
The
algorithm
uses
the sunphotometer
AOD
along
with
the lidar
data
to produce
profiles
of
the
aerosol
extinction
coefficient
and
AOD,
and
the columnar
backscatter-extinction
ratio.
The
lidar
derived
aerosol
optical
data
are used
to show
normal
IZO
site
conditions,
as well
as
conditions
seen
during
Saharan
dust
passages.
Finally,
comparisons
between
the lidar
data
and
data
from
other
ACE-2
CLEARCOLUMN
instruments
are
presented.
In
addition
to
daily
comparisons
with
the other
IZO
instruments,
joint
measurements
of the AOD
on the
afternoon
of
the
July
17,
1997
dust
episode
were
performed
with
the
lidar,
a
sunphotometer
on
board
an
ACE-2
aircraft,
and
a radiometer
installed
on
the
nearby
volcano
of
Tenerife
(Teide).
The
comparisons
demonstrate
the
success
of
the
lidar
calibration
techniques
and
the lidar
inversion
algorithm,
and
show
that
lidar
analysis
can
produce
accurate
profdes
of ambient
aerosol
optical
properties.
2.
Instrumentation
2.1
Micro-pulse
Lidar
System
(MPL)
The
lidar
used
in
this
study
is
a
micro-pulse
lidar
system
(MPL)
and
is
manufactured
by
Science
& Engineering
Services
Inc.,
USA.
Basic
MPL
design
and
background
is described
in Spinhime
(1993)
and Spinhime
et al. (1995).
The
MPL
system
is revolutionary
in that
it uses
rapidly
pulsed
low
intensity
laser
light
instead
of
slowly
pulsed
very
high
intensity
laser
light
that
has
been
used
in previous
lidar
systems.
The
MPL
system
has
laJ output
energies
and
the
beam
is expanded
to 20.32
cm
in diameter
which
achieves
ANSI
eye-safe
standards.
A pulse
repetition
frequency
(PRF)
of 2500
Hz
allows
the system
to average
many
low
energy
pulses
in a short
time
to
achieve
a good
signal-to-noise
ratio.
In
practice,
an averaging
time
of
1 minute
is used
for
dam
collection
but
the
stored
signals
can
be
averaged
over
longer
periods
if
necessary
during
post-
Draft
version
12-09-98
4
Draft:
for
Tellus
ACE-2
special
issue
analysis.
The
MPL
system
also has a high
vertical
spatial
resolution
(75
m is used
with
the
MPL
system
in this
study).
Finally,
the MPL
system
is small
compared
to previous
lidar
systems
and
is therefore
much
more
portable
than
its predecessors.
The
small
size
of
the
MPL
system
allows
the
operator
to perform
lidar
measurements
at any
zenith
angle
by
carefully
tilting
the
instrument
to the
desired
angle.
It
is
therefore
possible
to
perform
horizontal
and
slant
path
measurements
with
the
MPL
as
well
as
the
normal
vertical
measurements.
Care
must
be taken
when
operating
the MPL
during
sunny
days
as
direct
sunlight
entering
the MPL
can
cause
serious
damage
to the
detector.
The
MPL
must
be
tilted
away
from
the sun or turned
off and covered
in such
conditions.
The
MPL
is pictured
schematically
in Figure
1. The
MPL
transmitter-receiver
(T-R)
is located
inside
the climate
housing
and
consists
of a black
20.32
cm diameter
cassegrain
telescope
with
optics
and
electronics
mounted
directly
below
the
telescope.
The
laser
supply
and scalar
(data
binning
unit)
are connected
to the T-R,
and
along
with
the control
computer,
they
must
be located
inside
a separate
climate
controlled
environment.
The
laser
supply
contains
a diode
pumped
Nd:YLF
laser
with
a fundamental
pulse
output
wavelength
of
1046
nm
that
is converted
to 523
nm
for
lidar
use
after
passage
through
a frequency
doubling
crystal.
The MPL
system
used
in this
study
was
operated
at the full
laser
power
supply
setting
of
1 W.
The
pulse
duration
is
10 ns
with
a PRF
of
2500
Hz
and
output
energies
ranging
from
1 to 6 IxJ depending
upon
system
performance.
Signals
are received
using
the same
telescope
and are recorded
with
a Geiger
mode
avalanche
photodiode.
The
signals
are
stored
as
photons/sec
(ph/sec).
Since
the
receiver
is a
telescope
focused
at infinity,
the T-R
has
difficulty
accurately
imaging
near-range
signals
onto
the
detector.
This
proble
m is referred
to as overlap
error
and causes
the near-range
signals
(0 to
approximately
2 km)!o
rapidly
fall off in intensity
the closer
they
are to the T-R.
Since
the
majority
of aerosols
are contained
in the first
several
kilometers
of the
atmosphere
(or
as at
IZO,
the
flu'st
several
kilometers
in
range
from
IZO),
the
overlap
problem
must
be
Draft
version
12-09-98
5
Draft:for
Tellus
ACE-2
special
issue
overcome.
The
procedures
used
in this study
to correct
for
the MPL
overlap
are discussed
in Section
3.
The
signals
are
binned
in the
scalar
according
to their
time-of-flight
from
transmission
to signal
reception
and
correspond
to steps
of
75
m
in
range.
Our
MPL
system
has a pause
of approximately
1 Its from
activation
of the detector
to emission
of the
laser
pulse.
Thus,
we disregard
the first
two
signal
bins
and re-zero
the range
such
that
the
third
signal
bin
represents
the signal
return
from
75 m.
A control
computer
is connected
to the scalar
and
is used
to control
lidar
operation,
to visualize
real-time
lidar
output,
and to store
the resulting
lidar
data.
The
data
are stored
in
1 hour
binary
f'des
with each
record
containing
a header
followed
by the
signal
in ph/sec
at
the successive
75 m increments
up to a preset
range
(30 km).
The
maximum
MPL
range
having
usable
data
typically
varies
from
approximately
30 km
at night
to
about
10
km
during
reasonably
clear
daytime
conditions.
The
lidar
signals
stored
on
the
control
computer
contain
background
noise
that is present
from
laser-detector
crosstalk
(afterpulse)
and
sunlight
at
523
nm.
Afterpulse
noise
must
be
corrected
in
post-analysis
and
the
procedure
is discussed
in Section
3. Background
sunlight
noise
is measured
by the MPL
in
real
time
by
measuring
the detector
signal
after
the maximum
altitude
signal
(30
km)
has
arrived
and
before
the next
pulse
is fired.
This
background
signal
is stored
and
used
to
correct
the final
signal
by subtracting
its value
from
each
binned
signal
in post-analysis.
The
header
information
contains
the
time,
output
pulse
energy,
instrument
temperatures,
background
sunlight
energies,
and
information
necessary
for the operator
to determine
the
altitude
resolution
used
for each
record
in the data
f.de.
2.2
Other
IZO
Instruments
ln-situ
aerosol
measurements
were
made
at
IZO.
Aerosol
mass
concentration
measurements
were
made
with
a Rupprecht
& Patashnick
Model
1400a
tapered
element
oscillating
microbalance
and extinction
coefficients
were
obtained
with
a Radiance
Research
particle/soot
absorption
photometer,
a TSI
Model
3563
integrating
nephelometer,
and
a
Draft
version
12-09-98
6
Draft:for
Tellus
ACE-2
special
issue
Radiance
Research
Model
M903
integrating
nephelometer.
The
IZO
in-situ
measurements
were
used
for
MPL
calibrations
and
in comparisons
with
the MPL.
A NASA
AERONET
cimel
sunphotometer
(Holben
et al.,
1998)
was
also
operated
at IZO
for
the
duration
of
ACE-2.
The
cimel
was
used
to acquire
independent
measurements
of the
AOD
(Smirnov
et
al.,
1998)
for
input
to the lidar
inversion
algorithm
(Appendix)
and
to
perform
aerosol
measurements
specific
to AERONET
and
CLEARCOLUMN
operations.
2.3
Airborne
and
Teide
Instruments
The
MPL
was
used
to
supply
morning
lidar
signal
profiles
via
fax
to ACE-2
headquarters
to assist
in the pre-flight
briefing
for the Center
for Interdisciplinary
Remotely
Piloted
Aircraft
Studies
Pelican
aircraft,
which
was
one
of
the
aircraft
participating
in
ACE-2.
AOD
measurements
were
made
on-board
the
Pelican
with
the
NASA
Ames
Airborne
Tracking
14-channel
Sunphotometer,
AATS-14,
(Schmid
et al.,
this
issue).
AOD
measurements
were
also
made
on the island's
volcano,
Teide
(28016 , N,
16036 , W,
3570
m asl),
during
ACE-2
using
a multi-f
"alter rotating
shadowband
radiometer
(Formenti
et al.,
1998;
Formenti
et
al.,
this
issue).
AOD
data
acquired
with
the
AATS-14
and
the
shadowband
are used
in the comparison
on July
17, 1997.
3.
The
Lidar
Equation
and
MPL
Calibrations
The
basic
lidar
equation
for returned
signals
is given
by
Is;
1
S_(z)
=
(z - z,) 2 [fiR(z)
+ fla(z)]
exp
-2
L (trR(z')
+ trA(Z'))dz"
(1)
where
SL(Z) is the
lidar
signal
at altitude
z
(m),
C is the
system
constant
(principally
a
function
of the
optics),
E is the output
energy
in ktJ,
ZL is the lidar
altitude
(m),
[_(z)
and
t_(z)
are
the
backscatter
(m 1
sr 1)
and
extinction
(m 1)
coefficients
respectively,
the
R
Draft
version
12-09-98
7
Draft:for
TeUus
ACE-2
special
issue
subscript
denotes
a Rayleigh
quantity
(due
to molecular
scattering),
and
the
A subscript
denotes
an aerosol
quantity.
The
backscatter
coefficient
is related
to extinction
by
fl(z)
= R(z)tr(z)
,
(2)
where
R(z) is the backscatter-extinction
ratio
(BER)
with
units
of sr -1. The
aerosol
BER
is
considered
to be
constant
for
each
profile
in
this
study
and
thus
is
referred
to
as
the
columnar
backscatter-extinction
ratio,
R A. The
BER
is related
to the aerosol
phase
function,
PA(O)
(normalized
to 4r0,
and
the single
scattering
albedo,
too, by
4x*R A =
tooPA(n)-
It
is useful
to rewrite
Eq.
(1)
by
multiplying
by
the
range
squared,
(Z-EL)
2,
to
remove
the range
dependent
fall off in the
signal
returns
and
to use
the BER
to rewrite
the
equation
in terms of only
the backscatter
coefficient,
St(z)
= CE[flR(z)+
fla(z)]exp[-2[.
R,
I_,_R(z
)dz
exp
-'_-A
I_L fla(z')dz"
(3)
where
S_(z)
is referred
to as the range-corrected
lidar
signal,
R_ is the
BER
for
Rayleigh
scattering,
and R A is the BER
for
the
aerosols.
The
lidar
equation
must
be solved
for
the
unknown
aerosol
quantities,
_A(Z),
CA(Z),
and
R A. The
Rayleigh
optical
functions
are
constructed
using
data
from
Hansen
and
Travis
(1974).
The
BER
values
used
above
are
assumed
to
be
constant
with
altitude.
While
R R is
constant,
R A may
actually
vary.
Algorithms
exist
for
lidar
analysis
using
altitude
dependent
R A values
(Klett,
1985;
Kovalev,
1993)
but
require
additional
assumptions
or
measurements
of
the
vertical
structure
of the
aerosol
optical
properties
that
were
not
possible
for
this
work.
The
lidar
inversion
algorithm
in this
study
uses
an independent
AOD
measurement
to iterate
a basic
lidar
inversion
(Fernaldet
al.,
1984)
to
produce
the
13A(Z) and
_A(Z)
profiles,
and
to
calculate
the value
of R A. The
lidar
inversion
algorithm
used
in this
study
is discussed
in
Draft
version
12-09-98
8
Draft:
for
Tellus
ACE-2
special
issue
the Appendix.
Errors
related
to assuming
constant
BER
values
are
also
addressed
in the
Appendix.
The
above
discussion
relates
to
vertical
lidar
measurements.
Horizontal
lidar
measurements
are
used
to
assess
the
horizontal
homogeneity
of
the
atmosphere
at
a
particular
altitude.
The
backscatter
and
extinction
coefficients
for
a
horizontal
lidar
measurement
during
conditions
of
horizontal
homogeneity
are
constant
by
deffmition.
Therefore,
a horizontal
lidar
signal
simplifies
Eq.
(3) to
SH(x)
= CE[flt_(z
L) + fla(ZL)]exp[-2(crR(z
L) + O'a(ZL))X ] ,
(4)
where
S,(x)
is the horizontal
lidar
signal,
x is the horizontal
range
in m,
and
the
values
of
I_i(zt) and ai(z r) are
constants
with
respect
to x.
Furthermore,
taking
the natural
logarithm
of both
sides
of Eq.
(4) gives
ln[SH(X)
] =--2[CTR(ZL)+tTA(ZL)]X
+ ln[CE(flR(ZL)+
flA(ZL))]
.
(5)
Thus,
the
slope
of
ln[Sn(x)]
versus
the
range
x
yields
-2t_,o,.,
and
the
y-intercept
is
ln[CEI3totJ
for
the
horizontal
homogeneous
case.
If
the
atmosphere
is
not
horizontally
homogeneous
at the lidar
altitude,
then
a horizontal
lidar
plot of ln[Sr(z)]
versus
the range
x
will not produce
a straight
line.
MPL
Calibration
Procedure
Equation
(3)
is
an
ideal
lidar
signal.
Actual
lidar
signals
are
effected
by
both
afterpulse
and
overlap
problems
as mentioned
in Section
2.
Thus,
an
actual
MPL
range-
corrected
signal
is given
by
S,(z)=
CO(z)E[fl.(z)+
[3a(z)]exp
-=
['
[3R(z')dz"
exp-
J"
[3A(z')dz"
L
,
(6)
+ A(Z)
where
O(z) and A(z)
represent
the overlap
and afterpulse
functions.
Draft
version
12-09-98
9
Draft:
for
TeUus
ACE-2
special
issue
Calibration
of the MPL
system
involves
correcting
for
the
afterpulse
and
overlap
functions.
Additionally,
lidar
calibration
can
also
involve
determination
of the lidar
system
constant,
C.
However,
the determination
of the system
constant
was
not
a focus
of
this
study
since
it is not
needed
to produce
the
optical
coefficient
profiles
when
the
AOD
is
measured
by an independent
instrument
(Appendix).
The
calibration
procedures
applied
to the MPL
during
ACE-2
differ
from
the normal
MPL
calibration
techniques
(Welton,
1998).
The
laser
frequency
doubling
crystal
in the
MPL
system
burned
midway
through
ACE-2.
The
cause
of
the
burned
crystal
was
attributed
to a poor
ground
connection
between
the laser
temperature
controller
on the laser
supply
and the laser
itself,
located
inside
the T-R.
Data
continued
to be taken
with
the MPL
system
because
the problem
was not noticed
until
the end
of the experiment.
The
data
acquired
after
the crystal
burn
had
noticeable
effects
caused
by signal
loss
and diffraction
from
the burn
pattern.
Signal
loss
resulted
from
light
scattered
off
axis,
by
the hole,
that
was
lost
before
reaching
the T-R.
Diffraction
effects
were
believed
to be the
cause
of distortions
in the outgoing
laser
pulses.
These
diffraction
effects
caused
distortion
of the MPL
overlap
characteristics
and altered
the afterpulse
signal.
These
problems
became
worse
as
the
experiment
continued.
Thus
it was
not
possible
to use
the
pre-experiment
calibrations
or post-calibrations
to correct
the entire
data
set taken
during
the
experiment.
Therefore,
a new
lidar
calibration
procedure
was developed
to handle
the MPL
data
during
ACE-2.
The
procedure
is based
on normalization
of the MPL
signals
to those
signals
from
a molecular
(Rayleigh)
only
atmosphere
and is described
below.
Due
to its
unique
location,
IZO
is in the
free
troposphere
at night
(Raes
et
al.,
1997).
The
MPL
performed
vertical
profile
measurements
during
normal
ACE-2
nighttime
lidar
operations.
Several
nights
Were
very
clean
and
the
lidar
returns
were
assumed
to
represent
pure
Rayleigh
scattering
with
the exception
of
the afterpulse
and
overlap
effects.
This
assumption
was
based
on normal
nighttime
conditions
and
inspection
of both
aerosol
Draft
version
12-09-98
10
Draft:
for
Tellus
A CE-2
special
issue
mass
concentration
and
aerosol
extinction
coefficient
measurements
made
at
the
observatory
during
the night.
The
early
mornings
(00:00
GMT
to 03:00
GMT)
of
June
29
and
July
15,
1997
were
chosen
as having
clean
(negligible
aerosol
concentrations)
based
on the low
aerosol
concentrations
and extinction
coefficients
shown
in Figures
2a and 2b.
At 00:00,
01:00,
02:00,
and
03:00
GMT,
a
15 minute
average
Rayleigh
lidar
signal,
SR(Z), was
calculated
using
Eq.
(3) with
13k(Z) = 0,
E obtained
from
the
actual
time
corresponding
measured
lidar
signal,
and with
C set equal
to 100,
The
actual
measured
lidar
signals
are expressed
using
the following
equation,
St.(z)
= CE O(z)flR(z)exp[-
2-_-L
RR
_"
Sz flR(z')dz']
+ A(z),
(7)
(8)
where
[_A(Z) is still
assumed
to be zero.
Eq.
(8)
can
be rewritten
in terms
of the Rayleigh-
only
signal
as
St.(z)
= O(z)
SR(Z) + A(z)
(9)
The
term
SR(z ) is calculated
and
the term
SL(Z) is measured
with
the MPL
system,
thus
the
only
unknowns
in Eq.
(9) are O(z)
and A(z).
A linear
regression
was
performed
using
Eq.
(9),
the calculated
Rayleigh
signal,
and
the measured
lidar
signal
for
each
altitude
bin
in
each
period
(4 each
night)
from
the chosen
nights.
The
y intercepts
where
used
to construct
the afterpulse
function
and
the
slope
was
used
to
determine
the
overlap
function.
The
resulting
overlap
and afterpulse
functions
are shown
in Figures
3a and
3b.
The
overlap
function
for June
29 approaches
an asymptote
of almost
10,
instead
of
the usual
value
of 1, as the range
increases
beyond
2 km.
This
is due
to setting
C equal
to
100
for
the calibration
procedure.
The
actual
value
of C for
the June
29 period
was
most
likely
much
higher
than
100.
This
is also
the
reason
for
the
negative
afterpulse
values
Draft
version
12-09-98
11
Draft:
for
Tellus
ACE-2
special
issue
calculated
for
June
29.
The
overlap
and
afterpulse
functions
for
June
29 do
not
represent
the physical
overlap
and
afterpulse
values
for
this
period
due
to the arbitrary
choice
of
C.
However,
the overlap
and
afterpulse
functions
still
produce
the
correct
lidar
calibration.
Also,
the MPL
crystal
problems
increased
in magnitude
as the experiment
progressed,
and
the value
of C decreased
significantly.
The
value
of C was very
close
to 100 by July
15,
as
evidenced
by
the
overlap
asymptotic
limit
of
approximately
1 for
this
day.
Also,
the
afterpulse
values
for
July
15 are similar
to those
obtained
using
the MPL
with
no
crystal
problem
(Welton,
1998).
The
overlap
and afterpulse
functions
were
used
to correct
MPL
signals
only
during
the days
immediately
after
the calibration
night.
As an example
of applying
the calibration
functions
to the MPL
data,
Figure
4 shows
the original
lidar
signal
measured
at 00:00
GMT
on June
30,
the calculated
Rayleigh
signal,
and
the corresponding
overlap
and
afterpulse
corrected
signal.
The
signal
now
resembles
a free
troposphere
Rayleigh-only
lidar
signal
and demonstrates
the success
of the calibration
procedure.
4.
Analysis
of ACE-2
MPL
Data
The
MPL
was
operated
on a daily
schedule
that
involved
vertical,
horizontal,
and
slant
path
(T-R
tilted
to 60 ° zenith
angle)
measurements
at specific
times
of the day.
Vertical
measurements
were
typically
performed
from
00:00
GMT
to 10:30
GMT
and
again
from
16:30
GMTto
23:59
GMT
each
day.
Horizontal
measurements
were
usually
performed
from
10:30
GMT
to
11:00
GMT
and
from
16:00
GMT
to
16:30
GMT,
and
slant
path
measurements
were
made
each
day
from
11:00
GMT
to 16:00
GMT.
Slant
path
rather
than
vertical
orientation
was
necessary
during
mid-day
to prevent
direct
sunlight
from
entering
the T-R
and damaging
the MPL
detector
and
optics.
The
schedule
was
occasionally
altered
to accommodate
Pelican
overflights
and
special
ACE-2
directed
activities.
For
this
study,
only
vertical
and selected
horizontal
measurements
are discussed.
Draft
version
12-09-98
12
Draft:
for
Telh_s ACE-2
special
issue
MPL
installation
and
testing
was
performed
during
the
first
weeks
and
normal
operation
began
on
June
28.
The
insla-ument
problems
with
the
MPL
system
became
substantial
after
July
20,
1997
and
the resulting
data
from
the
correction
procedure
were
not
considered
reliable.
Thus,
only
MPL
data
from
June
28 to July
20
were
analyzed
for
this
study.
The
MPL
signals
were
calibrated
using
the procedure
discussed
in
Section
3.
The
signals
were
then
divided
by
the
lidar
constant
C
(set
equal
to
100)
and
the
corresponding
output
energy
E.
The
resulting
profile
is
referred
to
as
an
attenuated
backscatter
signal
(ABS,
units
of
m l
sr _)
because
it is a profile
of
the
total
backscatter
coefficient
and is attenuated
by the exponential
transmission
function.
The ABS
profiles
for
all the vertical
measurements
made
from
June
28 to July
20 showed
that
no
aerosols
were
present
above
an
altitude
of approximately
6 km
during
ACE-2.
Also,
most
days
during
ACE-2
produced
similar
ABS
profiles
and
were
identified
as
the
normal
site
condition
influenced
by upslope
aerosols.
The
periods,
July
7 to July
9,
and
July
16 to
July
18,
showed
much
higher
ABS
values
relative
to the
normal
site
profiles.
These
two
periods
correspond
to the fLrSt and
second
Saharan
dust
passages
observed
during
ACE-2.
This
study
will focus
on the normal
upslope
aerosol
conditions
at IZO
during
June
29 to July
1,
and
the Saharan
dust
episode
from
July
16 to July
18.
4.1
Analysis
of
Upsiope
Aerosols
During
the
day,
local
heating
near
IZO
(along
the
mountain
ridge)
creates
an
upslope
wind.
This
local
wind
carries
aerosols
from
within
the
marine
boundary
layer
(MBL)
below
IZO,
to the level
of
17.O
and
beyond.
The
upslope
aerosols
appear
in the
early
morning
as the sun
rises
and subside
by the late
afternoon
as the sun
sets
and
the air
temperature
stabilizes.
The
presence
of upslope
aerosols
during
the daytime
is characteristic
of
normal
conditions
at the
IZO
site
(Raes
et
al.,
1997),
therefore,
it
is
necessary
to
understand
the upslope
aerosol's
spatial
distribution
and
optical
prof'de
before
analysis
of
the Saharan
dust layers
can be attempted.
Draft
version
12-09-98
13
Draft:for
TeUus
ACE-2
special
issue
Time
series
ABS
prof'des
are shown
for June
29
to July
1 in Figure
5.
Also,
ABS
profiles
from
early
morning
to late
afternoon
on June
29
are shown
in Figures
6a
and
6b.
The ABS
profde
at 06:15
GMT,
approximately
45 minutes
before
sunrise
(-
07:00
GMT)
is representative
of a Rayleigh-only
profile,
no
aerosol
layers
are
present.
However,
the
proffie
at 07:15
GMT
shows
a weak
aerosol
layer
extending
to under
6 km
in altitude.
The
prof'des
at 10:15
GMT
and
17:15
GMT
also
show
aerosol
layers
extending
to under
6 km
in altitude
but
with
much
higher
ABS
values
just
over
IZO.
These
mid-day
ABS
profdes
are significantly
less
than
the earlier
ABS
prof'des
at higher
altitudes
due
to the attenuation
of the upslope
layer.
The
ABS
prof'de
at 19:15
GMT
shows
the
aerosol
layer
subsiding,
with
ABS
values
similar
to
the
07:15
GMT
prof'fle,
and
lower
than
during
mid-day.
Finally,
the ABS
prof'de
at 22:15
GMT
shows
no indication
of aerosol
layers,
and
instead
resembles
the Rayleigh-only
ABS
profde
at 06:15
GMT.
The
ABS
prof'des
in Figures
6a
and
6b clearly
show
the
presence
of
the
upslope
aerosols
and
this
pattem
is typical
for
normal
upslope
conditions
at IZO during
ACE-2.
Upslope
Aerosol
Extinction
Coefficient
and
Optical
Depth
Profiles
The
AOD
measured
with the IZO cimel
was
used
to calculate
the aerosol
extinction
coefficient
(AEC),
the columnar
backscatter-extinction
ratio
(BER),
and
AOD
prof'des
for
the upslope
aerosol
layers
on June
29,
1997
using
the
inversion
algorithm
discussed
in the
Appendix.
The
prof'des
shown
for
June
29
are
representative
of
normal
upslope
aerosol
conditions
during
ACE-2.
The
lidar
AEC
profdes
are
shown
in Figure
7a
and
the
AOD
proftles
are shown
in Figure
7b.
The
daytime
average
AEC
measured
at IZO
using
the
nephelometer
is also
shown
in Figure
7a, and the AOD
measured
by the Teide
shadowband
is displayed
in Figure
7b. The Rayleigh
extinction
coefficient
profile
is shown
in Figure
7a
for comparison.
Significant
AEC
values
(greater
than
the
Rayleigh
extinction
coefficient)
extend
from just
under
3 km
to approximately
3.5
km
in altitude
for
the 08:15,
17:15,
and
18:15
Draft
version
12-09-98
14
Draft:
for
TeUus ACE-2
special
issue
GMT
profiles.
The
bottom
of the upslope
layer
is lower
during
the late morning
profile
at
10:15
GMT
and
at this
time
the
layer
extends
from
approximately
2.5
km
to
3.5
km
in
altitude.
The upslope
layers
at IZO typically
begin
forming
at approximately
07:00
to 08:00
GMT
and begin
dissipating
just
prior
to 18:00
GMT.
The
formation
of the upslope
aerosols
corresponds
with
a gradual
lowering
of the bottom
of the layer
through
to the late morning,
and
the dissipation
of
the upslope
aerosols
corresponds
with
a gradual
increase
in
the
altitude
of the bottom
of the layer.
The
peak
upslope
aerosol
AEC
values
are approximately
1.5e-5
m -_. These
are not
large
compared
with
boundary
layer
values
(Schmid
et al.,
this
issue)
but
are higher
than
free-troposphere
values
(Rayleigh
values).
The
AEC
values
peak
at approximately
3 km
in
altitude,
but are much
lower
near
IZO
throughout
the day.
In order
to check
that
this
was
not
due to a poor
calibration
of the MPL,
the nearest
range
lidar
AEC
value
(2.442
km)
was
compared
to
the
daytime
(07:00
to
19:00
GMT)
AEC
measured
at
IZO
using
a
nephelometer.
The
average
lidar
AEC
value
at 2.442
km
for
the
10:15
and
17:15
GMT
profiles
is 4.4e-6
m 1, which
agrees
well
with
the IZO
nephelometer
AEC
value
of -5e-6
m
1. However,
the 08:15
and
18:15
GMT
lidar
AEC
values
(at
2.442
km)
are
much
lower
than
the
IZO
value.
This
is most
likely
due
to the
rapidly
changing
aerosol
conditions
during
the formation
and dissipation
of the upslope
layer.
The
changing
aerosol
conditions
(thus,
varying
AOD
values
from
the cimel)
and
overall
low
AEC
values,
results
in poorer
inversion
results.
The AOD
measured
with
the Teide
shadowband
is compared
to the lidar
AOD
value
at 3.57
km in Figure
7b. The
low AOD
values
measured
by all instruments
demonstrate
the
absence
of
Saharan
dust
during
this
period.
The
lidar
AOD
value
at
3.57
km
is
approximately
half
of the Teide
AOD
value.
The
AOD
values
from
both
instruments
are
small
and the difference
between
the two are within
the bounds
of
measurement
error.
The
Draft
version
12-09-98
15
Draft:for
TeUus ACE-2
special
issue
difference
in AOD
may
also
be due
to the presence
of upslope
aerosols
specific
to Teide
which
are not
present
one kilometer
over
IZO.
Upslope
Aerosol
Baekseatter-Extinetion
Ratios
The
AEC
and
AOD
profiles
shown
in Figures
7a
and
7b
each
correspond
to
a
calculated
columnar
BER
value
from
the lidar
inversion.
The
BER
values
for
the
08:15,
10:15,
17:15,
and
18:15
GMT
profiles
on June
29 are 0.018,
0.038,
0.031,
and
0.017
sr -1
respectively.
Low
BER
values
(-0.020)
during
the early
morning
and
late
afternoon,
and
higher
BER
values
(-0.035)
during
mid-day
were
characteristic
of
upslope
aerosol
conditions
at
IZO
during
ACE-2.
The
low
BER
value
of
-0.020
occurs
during
the
formation
and
dissipation
of the
upslope
aerosol
layer,
while
the
higher
BER
value
of
-0.035
occurs
during
the stable
mid-day
period
of the upslope
layer.
The
lidar
t0oPA(_)
value
for
the early
morning
and
late afternoon
periods
averages
-0.25
sr -1, and
averages
-0.44
sr _ for
the
mid-day
period.
Phase
functions
derived
from
the
cimel
during
ACE-2
were
generated
using
Mie
theory
(O.
Dubovik,
personal
communication)
and
were
used
in conjunction
with
the
lidar
t0oPA(r_)
values
in
order
to
assess
the
impact
of
particle
non-sphericity.
The
lidar
BER
values,
and
hence
the
lidar
tOoPA(_)
values,
calculated
using
the
lidar
inversion
alogrithm
in
the
Appendix
are
not
dependent
upon
Mie
theory
and
are
direct
calculations.
Mishchenko
et
al.
(1997)
have
shown
that
there
are unique
differences
in the aerosol
phase
function
for
the ratio
of
non-
spherical
to spherical
measurements.
They
have
shown
that,
in
general,
there
is
tittle
difference
at forward
scattering
angles
(0 ° to 10°),
but
that
when
particle
non-sphericity
is
taken
into account
increased
side-scattering
(near
120 °) and
decreased
backscatter
(at
180 °)
occurs
relative
to spherical
particle
calculations.
The
data
shown
in Plate
5 of Mishchenko
et al.
(1997)
depicts
the
behavior
of
both
Mie
and
non-spherical
phase
functions
for
different
size
parameters,
x,n, using
a refractive
index
of 1.53
+ 0.008i.
Draft
version
12-09-98
16
Draft:
for
TeUus ACE-2
special
issue
The
cimel
(henceforth
referred
to as Mie)
phase
functions
at 0 ° and
180 ° and
the
lidar
O_oPA(g) values
(no
Mie
dependence)
were
used
to determine
the
degree
of
particle
non-sphericity
and
a value
ofxotr by comparison
with
the Plate
5 Mishchenko
et al.
(1997)
data.
The
Mie
phase
functions
at 0 ° are just over
30 sr _ in the early
morning
(07:50
GMT)
and
from
70 to 90 sr -_ at mid-day
(measurements
at 11:00
and
17:30
GMT).
The
Mie
phase
functions
at
180 °
are
approximately
0.3
sr -_
in
the
early
morning,
and
average
approximately
0.4
sr _ during
mid-day.
The
forward
and
backward
Mie
phase
function
values
given
above
do
not
exactly
match
the
behavior
of
Mie
functions
in
Plate
5
of
Mishchenko
et al.
(1997)
most
likely
due
to differences
in the
refractive
indices
used
to
generate
them.
However,
the
ratio
of
the
lidar
to Mie
phase
function
values
still
yields
useful
information.
Comparing
the forward
Mie
phase
function
values
from
the cimel
with
those
given
in Plate
5 of Mishchenko
et al.
(1997)
give
x,r r ranges
from
6 to
10,
which
correspond
to r, rr values
from
0.5 to 0.8 microns
using
the lidar
wavelength.
The
Mie
phase
functions
at 180 ° are not
much
different
from
the
lidar
tOoP^(_)
values
for
morning
and
mid-day.
The
lidar
PA(m) and Mie
P^(r0
values
would
then
agree
very
well
assuming
coo is
not significantly
less
than
1, indicating
that the particles
are fairly
spherical
in shape.
Upslope
Horizontal
Lidar
Signal
Results
Analysis
of horizontally
aimed
lidar
measurements
(approximately
due
East)
during
upslope
conditions
was
performed.
The
natural
logarithm
of the horizontal
ABS
at
10:45
GMT
on June
29 is shown
in Figure
8 along
with
a calculated
Rayleigh
horizontal
plot.
The
presence
of
the
upslope
aerosols
along
the
side
of
the
mountain
is
shown
by
the
elevated
and
non-linear
ABS
values
out
to approximately
1.5
km
from
the
side
of
the
mountain.
The
plot
becomes
linear
after
1.5
kin,
thus,
the atmosphere
does
appear
to be
horizontally
homogeneous
from
1.5
km
out
to 6 km
(the
maximum
daytime
range
of
the
Draft
version
12-09-98
17
Draft:for
Tellus
ACE-2
special
issue
MPL
during
most
of
ACE-2).
However,
near
the side
of
the
mountain,
and
near
IZO,
horizontal
homogeneity
does
not exist
due
to the presence
of the upslope
aerosols.
4.2
Analysis
of
Saharan
Dust
Episode
Three
Saharan
dust
episodes
occurred
during
ACE-2.
Each
episode
was
characterized
by
the
presence
of
dust
layers
at
and
above
the
IZO
site.
The
first
dust
episode
started
at mid-day
on
July
7
and
continued
until
the
afternoon
of
July
9.
The
second
dust
episode
started
late in the evening
on July
16 and
continued
until
the morning
of July
18. The
last
dust episode
started
on the morning
of July
25
and
continued
into
July
26,
past
the end
of ACE-2.
During
much
of the first
dust
episode,
the MPL
was
orientated
in
the
slant
path
position.
During
this
episode,
inspection
of
the
dust
layer
lidar
returns
and
IZO
aerosol
concentration
and
nephelometer
data
showed
that
very
little
of
the
dust
was
at the
IZO
altitude.
The
decision
was
made
to orient
the MPL
on a slant
path
in
order
to attempt
to
measure
dust
data
lower
than
the MPL's
lowest
vertical
measurement
range
(75
m).
As
a
result,
there
is little
vertical
MPL
data
during
the
In'st
dust
episode.
Also,
the
last
dust
episode
occurred
after
the
period
when
the
MPL
data
could
be
accurately
corrected.
Therefore,
no usable
MPL
data
exists
for the last dust episode.
The
results
presented
below
for
Saharan
dust
layers
are derived
from
analysis
performed
on data
acquired
during
the
second
dust
episode,
from
July
16 to July
18.
It is
important
to
note
that
sulfates
and
other
aerosol
species
have
often
been
correlated
with
dust
episodes
at IZO
and
elsewhere
over
the North
Atlantic
Ocean
(Welton
et al., unpublished
data;
Maring,
personal
communication).
Therefore,
the results
presented
in this
section
for
dust
conditions
at IZO
are likely
to include
some
effects
from
aerosols
other
than
dust,
and may
in fact
underestimate
the effects
of the dust
aerosols
alone.
A time
series
of ABS
profiles
from
July
16 through
July
18 is shown
in Figure
9.
The calibration
of the MPL
was more
difficult
for
this
period,
relative
to June
29,
because
Draft
version
12-09-98
18
Draft." for
Tellus
ACE-2
special
issue
the MPL
crystal
problems
were
much
worse
than
in late
June.
The
fmal
July
16 to July
18
calibrated
lidar
data contains
more
noise
than
the June
29 to July
1 data.
The
straight
lines
running
across
the
figure
are due
to noise
in the calibrations
and
not
aerosol
effects.
The
problem
is only
significant
during
mid-day
at
altitudes
over
6 km
(which
exceeds
the
maximum
aerosol
altitude
observed
during
ACE-2
using
the MPL).
This
was
due
to a low
signal-noise
ratio
caused
by
background
sunlight
and
very
high
dust
concentrations,
corresponding
to high
signal
attenuation,
during
mid-day.
The
temporal
extent
of the dust
layer
is clearly
evident.
The
dust
layer
appeared
at
approximately
22:00
GMT
on July
16 at an
altitude
of
approximately
3.5
km.
The
layer
dropped
in altitude
by the morning
of July
17 with
the majority
of the dust
at altitudes
from
about
2.5
km
to about
4 km
until
the
late
afternoon.
The
layer
thickness
narrowed
in
altitude
considerably
after
18:00
GMT
on July
17.
Most
of the dust
remained
at altitudes
from
about
2.75
km to 3.5 km
for the duration
of the episode,
which
ended
the morning
of
July
18.
Dust
Aerosol
Extinction
Coefficient
and
Optical
Depth
Profiles
The
AOD
measured
with
the
IZO
cimel
was
used
to
calculate
the
AEC,
the
columnar
BER,
and
the
AOD
profiles
for
the
dust
layers
on
July
17,
1997
using
the
inversion
algorithm
discussed
in
the
Appendix.
The
lidar
AEC
prof'fles
calculated
throughout
the day
(08:15,
10:15,
17:15,
and
18:45
GMT)on
July
17 are
shown
in Figure
10 along
with
the Rayleigh
extinction
coefficient
profde
for
comparison.
The
AEC
values
for July
17 are orders
of magnitude
higher
than
the upslope
values
on June
29 and
thus
the
presence
of the
dust
is obvious.
The
peak
AEC
values
were
between
1.5e-4
and
2e-4
m -1
and
were
located
just
above
3 km
in altitude.
Significant
AEC
values
(greater
than
the
Rayleigh
extinction
coefficient)
were
present
from
the IZO
altitude
to just
under
5 km.
The
inversion
results
for
10:15
GMT
and
17:15
GMT
are slightly
effected
by
the
calibration
noise
problem
discussed
above.
The
signals
are too
noisy
to start
the inversion
Draft
version
12-09-98
19
Draft:
for
Tellus
ACE-2
special
issue
at 5 km
or higher
for these
two
times.
Thus,
the inversions
were
started
at 4.8
and
4.5
km
for the two
profiles
respectively.
The
end
result
is a profile
which
has
fairly
accurate
AEC
and AOD
values
due to the high
concentrations
of aerosol,
but may
have
less
accurate
BER
values.
The
results
from
the
08:15
and
18:45
GMT
profdes
were
not
effected
by
this
problem.
Figure
1 la shows
the July
17 lidar
AEC
prof'des
at 10:15
and
18:45
GMT.
Figure
l la
also
shows
the
average
AEC
measured
by
the
IZO
nephelometer
at
both
mid-day
(Daytime)
and after
18:00
GMT.
The lidar
AEC
values
at 2.442
km
agree
well
with
the IZO
nephelometer
AEC
values.
Figure
lib
shows
the
lidar
AOD
profile
at
18:45
GMT,
the
AATS-14
(onboard
the Pelican
aircraft)
AOD
prof'de
from
18:30
to
18:45
GMT,
and
the
AOD
measured
by
the
Teide
shadowband
and
the
IZO
cimel
for
this
time
period.
The
AATS-14
AOD
values
immediately
above
the
IZO
altitude,
average
0.218
+
0.05
AOD
units.
This
portion
of the Pelican
flight
corresponds
to horizontal
flight
tracks
across
the
mountain
ridge,
approximately
50 meters
over
IZO.
The
spread
in AOD
_
0.05)
for
these
tracks
is
evidence
of
slight
changes
in
the
horizontal
homogeneity
of
the
dust
layer
overhead.
The
AOD
values
from
all the instruments
agree
within
approximately
0.02
AOD
units
for
most
of
the
profile
and
they
agree
better
than
the
+
0.05
AOD
spread
from
horizontal
inhomogeneity
for
the entire
profile.
The
excellent
agreement
between
the
hdar
18:45
GMT
data
and the data
from
the other
instruments
for
this
time
shows
that
the MPL
calibrations
and
inversion
algorithm
worked
successfully
and
that
the BER
calculated
for
this prof'de
was
accurate
for this dust
episode.
Dust
Aerosol
Backscatter-Extinction
Ratios
The
calculated
lidar
BEg
values
for
the
08:15,
10:15,
17:15,
and
18:45
GMT
profiles
were
0.026,
0.048,
0.073,
and
0.027
sr "_ respectively.
The
same
diurnal
increase
in the BER,
as was
present
for
the
upslope
aerosols,
may
exist
during
the
dust
episode
because
slightly
higher
BER
values
(average
-0.06
sr _)
during
mid-day
relative
to BER
Draft
version
12-09-98
20
Draft:
for
Tellus
ACE-2
special
issue
values
for
morning
and
late afternoon
(average
-0.025
sr t)
were
calculated
on
July
17.
However,
the mid-day
BER
values
may
be inaccurate
due
to the calibration
noise
problem.
Elevated
BER
values
during
mid-day
could
also
be
caused
by
the
presence
of
a small
amount
of small
spherical
upslope
aerosols.
However,
analysis
of
this
possible
effect
is
beyond
the scope
of this paper.
Thus,
attention
was focused
on the moming
and late
afternoon
iidar
profiles
(08:15
and
18:45
GMT
respectively)
because
they
were
not
influenced
by
significant
noise
problems.
The
lidar
tooPA(n)
value
for
these
periods
is approximately
0.3
sr -_. Mie
phase
functions
were
calculated
(O.
Dubovik,
personal
communication)
using
the cimel
data
on
July
17. The
Mie
phase
function
values
at 0 ° were
near
200 sr -zat 08:00
GMT
and near
300
sr z at 18:30
GMT.
The
Mie phase
function
values
at 180 ° for
08:00
and
18:45
GMT
were
nearly
identical
and averaged
approximately
0.6 sr _. The
ratio
of the lidar
PA(n)
value
with
the
Mie
PA(_)
value
from
the
cimel
is
-0.5
assuming
too =
1,
and
the
ratio
is
-0.625
assuming
too = 0.8.
The
ratio
of lidar
to Mie
PA(g)
values,
calculated
using
ranges
of
too
similar
to those
determined
during
this dust
episode
(Formenti
et al.,
this
issue;
Schmid
et
al., this
issue),
agree
well
with
the results
presented
in Mishchenko
et al.
(1997)
for
non-
spherical
particles.
Therefore,
the
aerosols
present
during
this
dust
episode
were
non-
spherical.
Also,
comparison
of the measured
cimel
and lidar
phase
function
values
with
the
phase
functions
in Plate
5 of Mishchenko
et
al.
(1997)
gives
xet r ranges
from
16 to 20,
corresponding
to r_
ranges
from
1.3
to
1.7
microns
(assuming
the index
of
refraction
1.53+0.0080.
Dust
Horizontal
Lidar
Signal
Results
Horizontal
profiles
(lidar
aimed
approximately
due East)
of the natural
logarithm
of
the
ABS
on
July
17 at
11:15
and
18:50
GMT
are
shown
in
Figure
12,
along
with
a
Draft
version
12-09-98
21
Draft:
for
Tellus
ACE-2
special
issue
calculated
Rayleigh
profile
for
comparison.
Both
measured
lidar
prof'des
are
non-linear
within
2 km of IZO,
indicating
that horizontal
homogeneity
did
not
exist
near
the mountain
ridge.
The
11:15
GMT
prof'de
shows
an increasing
and non-linear
plot out
to about
1.5
km
in range,
followed
by a fairly
linear
(but
noisy)
plot
from
1.5
to 4 km.
The
18:50
GMT
profile
also
shows
a non-linear
plot
out to about
2.5 kin,
followed
by a linear
plot
from
2.5
to 4 km.
Therefore,
the
atmosphere
does
appear
to be
horizontally
homogeneous
from
approximately
2 km
to 4 km (the maximum
horizontal
range
with
dust
present)
away
from
the mountain
ridge
during
this dust episode.
Both
horizontal
prof'des
show
that
the aerosols
are not
horizontally
homogeneous
close
to the side
of
the ridge
(out
to
about
2 km).
However,
the
sharply
increasing
plot
within
the first
kilometer
of range
for the 11:15
GMT
profde
shows
that
a large
amount
of
aerosol
was present
within
1 km from
the ridge
relative
to the
situation
at
18:50
GMT.
The
increase
in aerosol
within
1 km
of the ridge
during
daytime
(11:15
GMT)
corresponds
to
the upslope
period.
The
18:50
GMT
profile
shows
that
less aerosol
was located
close
to the
mountain
side
(within
1 kin),
and
corresponds
to the period
after
the upslope
has
subsided
(-18:00
GMT).
The
upslope
wind
motion
may
have
changed
the
dust
layer
near
the
mountain
and
indicates
the
importance
of
considering
upslope
effects
on
the
horizontal
homogeneity
of the region
around
IZO.
5.
Conclusions
The
operation
of
the
MPL
system
during
ACE-2
has
shown
that
this
new
lidar
technology
can
be used
successfully
in the field.
ACE-2
closure
comparisons
between
the
MPL
system
and
other
independently
operated
instruments
have
shown
that
the
MPL
calibration
procedures
and
inversion
algorithm
succeed
in
producing
accurate
optical
profiles
throughout
the entire
range
of the profile.
This
is significant
because
it shows
that
Draft
version
12-09-98
22
Draft:
for
Tellus
ACE-2
special
issue
the overlap
and
afterpulse
problems
can
be overcome,
even
when
the
MPL
has
suffered
an
instrument
problem.
The
results
of
lidar
analysis
during
ACE-2
have
shown
several
interesting
characteristics
of the upslope
aerosols
and
the Saharan
dust
episode
during
the
middle
of
the experiment.
The
upslope
aerosols
were
seen
to form
a layer
several
hundred
meters
above
and to the sides
of
IZO during
the day,
and to subside
by nightfall.
The
peak
AEC
values
obtained
in these
upslope
aerosol
layers
were
low
compared
to MBL
values,
but
were
approximately
25%
greater
than
Rayleigh
values.
The dust layer
on July
17 was
seen
to reside
mostly
above
and to the sides
of IZO,
possibly
held
off
by the motion
of upslope
winds
despite
f'mdings
that
show
upslope
winds
have
been
known
to be
weaker
during
dust
episodes
(Raes
et
al.,
1997).
The
possible
perturbation
of
the
dust
layer
by
the
upslope
effect
is significant
due to the fact
that the IZO
site
is used
during
summer
months
to study
Saharan
dust
layers.
Peak
AEC
values
obtained
during
the dust
episode
were
an
order
of
magnitude
higher
than
Rayleigh
values.
Finally,
the dust
layer
altitude
ranges
observed
during
ACE-2
(from
just
over
2 km
to under
5 km)
correspond
well
with
other
investigations
of the vertical
structure
of the Saharan
air layer
(SAL)
over
the North
Atlantic
Ocean
(Carlson
and Prospero,
1972;
Karyampudi
and Carlson,
1988).
Results
of this study
also
show
that
upslope
aerosols
(rar from
0.5
to 0.8
microns)
are smaller
relative
to aerosols
during
the dust
episode
(re_r from
1.3
to
1.7
microns).
The
differences
in
rat
values
between
upslope
and
dust
conditions
agree
well
Angstrom
exponents
measured
with
the
IZO
cimel
during
ACE-2
(Smirnov
et
al.,
!998).
The
Angstrom
exponent
during
upslope
conditions
on
June
29
was
approximately
1.0,
indicating
may
small
particles,
and
the Angstrom
exponent
during
the afternoon
of the July
17 dust
episode
averaged
0.16,
indicating
a greater
portion
of
larger
particles.
Another
important
result
was
the determination
of
an accurate
BER
value
(0.025
sr _) for
the
dust
episode
on July
17.
Knowledge
of an accurate
BER
during
dust
episodes
will
aid
in the
Draft
version
12-09-98
23
Draft:
for
Tellus
ACE-2
special
issue
analysis
of future
lidar
measurements
in regions
influenced
by dust
aerosols.
Finally,
the
degree
of non-sphericity
of the upslope
aerosols
was
found
to be small,
but
the degree
of
non-sphericity
of
the
aerosols
during
the
dust
episode
was
large
and
found
to
be
in
agreement
with
previous
theoretical
studies
of dust-like
aerosols.
Several
areas
requiring
further
study
have
also
been
identified.
Using
the
initial
results
from
work
presented
in
this
study
and
others
in
this
issue,
more
in-depth
comparisons
and
analysis
of
data
from
the
different
measurement
platforms
will
be
addressed
in future
work.
In addition
to more
accurate
results
of the type
discussed
in this
study,
attention
will
be
given
to correct
determination
of the
altitude
dependence
of
the
index
of
refraction
and
the single
scattering
albedo.
This
will
in turn
produce
better
lidar
inversions
and
help
to determine
the correct
BER
for
both
the
upslope
and
dust
aerosols.
Lastly,
further
studies
of the horizontal
homogeneity
of the region
surrounding
17,0
will
be
attempted.
Appendix:
The
Lidar
Inversion
Algorithm
The
lidar
inversion
algorithm
used
for
this
study
is presented
in this
section.
Also,
errors
inherent
to the algorithm
are discussed.
The
primary
error
is due
to the assumption
of a constant BER value, RA.
The
Solution
to
the
Lidar
Equation
The
basic
I3A(z) solution
to lidar
data
taken
according
to Eq.
(3) is referred
to as the
backward
Fernald
two-component
solution
(Fernald
et al.,
1972)
as it uses
the value
of
the
backscattering
coefficient
at some
maximum
altitude,
zm, as
a boundary
value
and
then
successive
values
of
l_A(z)
are
calculated
as
the
altitude
is
decreased
toward
the
lidar
altitude,
zL. The
solution
is given
below
in a format
for
algorithm
development
(Fernald,
1984),
Draft
version
12-09-98
24
Draft:
for
Tellus
ACE-2
special
issue
/_A(x - 1) =
where
S,(x-
1)W(x
- 1,x)
S,(x)
+
1
+m
[S,.(x)
+ Sr(X
-- 1)W(X -- 1, X)]AZ.
LiRA
-fiR(x-
1),
(A.1)
and x is the
altitude
bin
one
step
above
x- 1,
and
Az
is the lidar
range
interval
(75
m).
In
order
to obtain
the extinction
coefficient
proffie,
each
value
of the backscattering
coefficient
need
only
be divided
by R A.
The
basic
lidar
algorithm
that
uses
Eq.
(A. 1) to solve
for
the aerosol
profiles
must
assume
that
R^
and
the backscattering
coefficient
at some
maximum
altitude,
13^(zm),
are
known.
R^
is not
usually
known,
but
the latter
constraint
is usually
valid
as
aerosols
are
normally
confined
to the marine
boundary
layer
(MBL),
or
at least
at low
altitudes
above
the lidar
(such
as over
IZO),
therefore,
z,, can
be chosen
at an altitude
where
13^(z_) = 0. An
algorithm
was
developed
for
this
study
that
uses
an independently
measured
AOD,
x^,
as
input
and produces
a _,
profile
that
integrates
to the measured
AOD
and
will
also
calculate
the
value
of R^.
The
Lidar
Inversion
Algorithm
This
algorithm
is
based
on
procedures
described
in
Femald
et
al.
(1972)
and
Marenco
et al. (1997).
The
alogrithm
produces
extinction
coefficient
and AOD
profiles,
and
also
calculates
the
BER.
The
algorithm
is described
below
and
presented
schematically
in
Figure
13.
The
first
step
in the
algorithm
requires
determination
of
I_^ at
some
maximum
altitude,
z,.,
and
is done
by inspection
of
the calibrated
signals.
Inspection
of
the
signals
Draft
version
12-09-98
25
Draft:
for
TeUus ACE-2
special
issue
during
ACE-2
showed
that
there
was
a maximum
altitude,
above
which,
no
aerosol
appeared
to be present
(_A ~ 0). The
inversion
algorithm
value
of Zm was
chosen
to lie just
above
this altitude,
with
_A set equal
to 0.
The
second
step
in the algorithm
is the calculation
of 13A one
altitude
step,
75
m,
below
Zm. This
is done
by solving
Eq.
(A.1)
with
13A(X=--Zm)= 0,
and
the Rayleigh
profile
quantities;
I_R(z), Ca(Z ), and R a, from
Hansen
and
Travis
(1974).
In
order
to
solve
Eq.
(A.1),
R A is required.
For
this
step
in the
algorithm,
R A is set
equal
to
1 and
[SA(X-1) is
calculated.
This
process
is repeated
downward
through
the atmosphere,
with
R A =
1 and
lSA(X+l)
obtained
from
the previous
step,
until
the value
of
_k is calculated
at the
lowest
altitude
bin
(75 m above
the MPL
system
altitude).
The
next
step
in the algorithm
is to improve
the
estimate
of
RA
(determination
of
RA ,e,,)-
The
new
value
for
R A is determined
using
the
backscattering
coefficient
profile
calculated
in the previous
step
(with
R A =
1) and
the
independently
measured
AOD,
"cA.
R A..
is calculated
using
the following
equation,
_:m
•
p
)dz
Ra ,_,
=
(A.2)
va
with
"CAfrom
the
independent
measurement.
The
backscattering
coefficient
profile
is now
recalculated,
using
_A(X=Zm)
=
0,
but
with
RA
= R^
,e,,.
This
step
is
continued
until
successive
values
of R A and R A newdiffer
negligibly
(percent
difference
between
g A and
R A
,_,
less
than
0.5).
The
final
backscattering
coefficient
profile
and
R A are
then
used
to
calculate
the extinction
coefficient
profile,
era(Z).
The
extinction
coefficient
profde
is then
Draft
version
12-09-98
26
Draft: for
TeUus A CE-2
special
issue
numerically
integrated
from
ZL to Zm, and
then
subtracted
from
"_Aat each
altitude
step,
to
produce
an AOD
profile,
xA(z).
Thus
the
final
data
products
from
the
algorithm
are
the
extinction
coefficient
and AOD
profiles
and R A.
Errors
in
the
Results
from
the
Lidar
Inversion
Algorithm
This
algorithm
was
tested
with
artificial
lidar
data
to study
the
effects
of
errors
caused
by the algorithm
and the assumption
of a constant
R A (Welton,
1998).
Both
a single
and a two
aerosol
species
atmosphere
were
tested.
The
results
show
that
in a single
aerosol
species
atmosphere
(with
constant
RA) the
algorithm
accurately
calculates
the [3A(z) profde
and the correct
R A (and
thus
accurate
CA(Z) and
AOD
prof'fles)
even
if the concentration
of
the aerosols
varies
vertically
and
the aerosols
are separated
into
different
layers.
For
cases
with
two
aerosol
species
with
different
R A values
and
different
backscattering
coefficient
profiles,
the results
show
that the algorithm
was
found
to calculate
an R A value
that
was
an
accurate
average
of
the
two
different
true
R A values
whether
the
species
were
in
one
continuous
layer
(but
not
mixed
together)
or
separated
into
two
distinct
layers.
For
real
situations,
different
aerosol
species
are often
mixed
together
and it is expected
that
the final
BER
calculated
will
be dependent
more
on the relative
amounts
of each
aerosol
and
will
not
produce
a direct
average
of the different
individual
BER
values.
However,
the
algorithm
will
produce
an
accurate
columnar
value
of the R A in real
situations.
This
is an
important
result
since
other
ground-based
instruments
that
also
measure
R A related
quantities,
such
as
the aerosol
phase
function,
also
produce
columnar
values
because
they
measure
the entire
atmospheric
column.
Errors
were
present
in the resulting
I_A(Z) prof'des
when
the constant
R A inversion
algorithm
was
applied
to an inhomogeneous
R A atmosphere.
The
initial
13A(Z) profile
values
Draft
version
12-09-98
27
Draft:
for
TeUus
ACE-2
special
issue
near
zm are correct
but
successive
values
of
[3A deviate
from
the correct
value,
this
and
the
calculation
of an average
R A then
effect
the calculation
of the ffA(Z) profile.
The
algorithm
forces
the f'mal
CA(Z) profile
to integrate
to the correct
"cA value.
The
value,
_3AIRA will
be
iterated
continually,
until
the correct
"cAvalue
is reached.
If the R A value
used
is incorrect,
then
the resulting
1_A profile
will have
errors.
These
types
of R A related
errors
have
been
studied
by
other
researchers
in depth
(Klett,
1985;
Sasano
et al.,
1985;
Kovalev,
1993;
Kovalev
and
Moosmuller,
1994).
In
order
to
attempt
to
overcome
errors
associated
with
the
choice
of
a constant
R A, these
researchers
have
constructed
algorithms
using
range
dependent
R A values.
However,
for
these
algorithms
an R A profile,
from
model
or data,
must
be used.
The
choice
was
made
to
use
a constant
R A algorithm
for
this
study
since
neither
data,
nor
models
of
R A, were
available
during
the lidar
campaigns.
Acknowledgments
This
research
is a contribution
to the
Intemational
Global
Atmospheric
Chemistry
(IGAC)
core
project
of the International
Geosphere-Biosphere
Programme
(IGBP)
and
is
part
of the IGAC
Aerosol
Characterization
Experiments
(ACE).
The
authors
extend
great
appreciation
to E. Cuevas,
R.
Juega
Buide,
and
the rest
of the Izafia
observatory
staff
tbr
their support
during
ACE-2.
Thanks
is also
given
to O.
Dubovik
for
providing
the
cimel
phase
functions,
and
to A. Chapin
for
invaluable
work
in preparing
the micro-pulse
lidar
system
for
field
use.
Funding
for
the measurements
and
analyses
presented
in this
work
was
provided
by the National
Aeronautics
and
Space
Administration,
US
National
Science
Foundation,
Office
of Naval
Research,
European
Commission
DG
XII
(Environment
and
Climate),
Max
Planck
Society,
and the National
Oceanic
and Atmospheric
Administration.
Draft
version
12-09-98
28
Draft:for
Tellus
ACE-2
special
issue
References
Carlson,
T.N.,
and
Prospero,
J.M.
1972.
The
large-scale
movement
of
Saharan
outbreaks
over
the Northern
Equatorial
Atlantic.
J. Appl.
Meteorol.
11,283-297.
air
Fernald,
F.G.,
Herman,
B.M.,
and
Reagan,
J.A.
1972.
Determination
of
aerosol
height
distributions
by lidar.
J. Appl.
Meteorol.
11,482-489.
Fernald,
F.G.
1984.
Analysis
of atmospheric
lidar
observations:
some
comments.
Appl.
Optics
23,
652-653.
Formenti,
P., Andreae,
M.O.,
and
Lelieveld,
J.
1998.
Saharan
Dust
Occurrence
at a mid-
tropospheric
site
in Tenerife.
J. Aerosol
Sci.
29,
S1047-S1048.
Formenti,
P.,
Andreae,
M.O.,
and
Lelieveld,
J.
1999.
Measurements
of
aerosol
optical
depth
in the North
Atlantic
free
troposphere:
results
from
ACE-2.
Tellus,
submitted
this
issue.
Heintzenberg,
J.,
and Russell,
P.B.
1999.
An
Overview
of the ACE
2 Clear
Sky
Column
Closure
Experiment
(CLEARCOLUMN).
Tellus,
submitted
this issue.
Hansen,
J.E.,
and
Travis,
L.D.
1974.
Light
scattering
in planetary
atmospheres.
Space
Sci.
Reviews
16,
527-610.
Draft
version
12-09-98
29
Draft:
for
Tellus
ACE-2
special
issue
Holben,
B.N.,
Eck,
T.F.,
Slutsker,
I.,
Tanre,
D.,
Buis,
J.P.,
Setzer,
A.,
Vermote,
E.,
Reagan,
J.A.,
Kaufman,
Y.J.,
Nakajima,
T.,
Lavenu,
F., Jankowiak,
I.,
and
Smimov,
A.
1998.
AERONET-A
federated
instrument
network
and
data
archive
for
aerosol
characterization.
Rein.
Sens.
Environ.
66,
1-16.
Karyampudi,
V.M.,
and
Carlson,
T.N.
1988.
Analysis
and
numerical
simulations
of
the
Saharan
Air Layer
and its effects
on easterly
wave
distrubances.
J. Atmos.
Sci.
45,
3102-
3136.
Klett,
J.D.
1985.
Lidar
inversion
with
variable
backscatter/extinction
ratios.
Appl.
Optics
24,
1638-1643.
Kovalev,
V.A.
1993.
Lidar
measurement
of
the vertical
aerosol
extinction
prof'des
with
range-dependent
backscatter-to-extinction
ratios.
Appl.
Optics
32, 6053-6065.
Kovalev,
V.A.,
and
Moosmuller,
H.
1994.
Distortion
of
particulate
extinction
prof'des
measured
with
lidar
in a two-component
atmosphere.
Appl.
Optics
33,
6499-6507.
Marenco,
F.,
Santacesaria,
V.,
Bais,
A.F.,
Balls,
D.,
di Sarra,
A.,
Papayannis,
A.,
and
Zerefos,
C.
1997.
Optical
properties
of
tropospheric
aerosols
determined
by
lidar
and
spectrophotometric
measurements
(Photochemical
Activity
and
Solar
Ultraviolet
Radiation
campaign).
AppL
Optics
36,
6875-6886.
Mishchenko,
M.I_,
Travis,
L.D.,
Kahn,
R.A.,
and
West,
R.A.
1997.
Modeling
phase
functions
for
dustlike
tropospheric
aerosols
using
a shape
mixture
of
randomly
oriented
polydisperse
spheroids.
J. Geophys.
Res.
102,
16831-16847.
Draft
version
12-09o98
30
Draft:
for
Tellus
ACE-2
special
issue
Prospero,
J.M.,
Schmitt,
R.,
Cuevas,
E.,
Savoie,
D., Graustein,
W.,
Turekian,
K.,
Volz-
Thomas,
A., Oltmans,
S., Levy,
H., and Diaz,
A.
1995.
Temporal
variability
of ozone
and
aerosols
in the free
troposphere
over
the Eastern
North
Atlantic.
Geophys.
Res.
Lens.
22,
2925 -2928.
Raes,
F.,
Van
Dingenen,
R.,
Cuevas,
E.,
Van
Velthoven,
P.F.J.,
and
Prospero,
J.M.
1997.
Observations
of aerosols
in the free
troposphere
and
marine
boundary
layer
of
the
subtropical
Northeast
Atlantic:
Discussion
of processes
determining
their
size
distribution.
J. Geophys.
Res.
102,
21315-21328.
Raes,
F.,
Bates,
T.,
Verver,
G.,
Vogelenzang,
D.,
and
Van
Liedekerke,
M.
1999.
The
Second
Aerosol
Characterization
Experiment
(ACE-2):
introduction,
meteorological
overview
and main
results.
Tellus,
submitted
this
issue.
Sasano,
Y.,
Browell,
E.V.,
and
Ismail,
S.
1985.
Error
caused
by
using
a
constant
extinction/backscattering
ratio
in the lidar
solution.
Appl.
Optics
24, 3929-3932.
Schmid,
B.,
Livingston,
J.M.,
Russell,
P.B.,
Durkee,
P.A.,
Jonsson,
H.H.,
Collins,
D.R.,
Flagan,
R.C.,
Seinfeld,
J.H.,
Gass6,
S.,
Hegg,
D.A.,
OstrSm,
E.,
Noone,
K.J.,
Welton,
E.J.,
Voss,
K.J.,
Gordon,
H.R.,
Formenti,
P.,
and
Andreae,
M.O.
1999.
Clear
sky
closure
studies
of
lower
tropospheric
aerosol
and
water
vapor
during
ACE-2
using
airborne
sunphotometer,
airborne
in-situ,
space-borne,
and
ground-based
measurements.
TeUus,
submitted
this
issue.
Smirnov,
A.,
Holben,
B.N.,
Slutsker,
I.,
Welton,
E.J.,
and
Formenti,
P.
1998.
Opticnl
Properties
of Saharan
Dust
During
ACE
2. J. Geophys.
Res.
103,
28079-28092.
Draft
version
12-09-98
31
Draft:
for
Tellus
ACE-2
special
issue
Spinhirne,
J.D.
1993.
Micro
pulse
lidar.
IEEE
Trans.
Geosci.
Remote
Sens.
31,
48-55.
Spinhirne,
J.D.,
Rail,
J.,
and
Scott,
V.S.
1995.
Compact
eye-safe
lidar
systems.
Rev.
Laser
Eng.
23,
26-32.
Welton,
E.J.
1998.
Measurements
of
Aerosol
Optical
Properties
over
the
Ocean
Using
Sunphotometry
and Lidar.
Ph.D.
Dissertation.
University
of Miami,
Coral
Gables,
1-150.
List of Figure
Captions
Figure
1. Schematic
diagram
of the micro-pulse
hdar
system
(not to scale).
Figure
2a. Aerosol
total
mass
concentrations
(_tg m _) measured
during
the early
mornings
of June
29 and July
15 1997 at IZO.
Figure
2b.
Aerosol
extinction
coefficients
(m 1)
measured
during
the early
mornings
of
June
29 and July
15 1997
at IZO.
Figure
3a.
Overlap
functions,
O(z),
calculated
on June
29 and July
15 1997.
Figure
3b. Afterpulse
functions,
A(z),
calculated
on June
29 and July
15 1997.
Figure
4.
The
measured
MPL
signal,
a calculated
Rayleigh
lidar
signal,
and
the
final
calibrated
MPL
signal
are shown
for 00:00
GMT
on June
30,
1997.
Draft
version
12-09-98
32
Draft:for
Tellus
ACE-2
special
issue
Figure
5. Three-day
attenuated
backscatter
signal
(ABS)
(m sr) 1 time
series
from
June
29
through
July
1, 1997.
Each
major
tickmark
separates
the
days,
and
each
minor
tickmark
is
one hour
(GMT).
Figure
6a. Attenuated
Backscatter
Signals
(m sr) -_ on June
29,
1997.
Each
prof'de
is a 15
minute
average,
starting
15 minutes
prior
to the
time
shown.
The
06:15
GMT
(before
sunrise)
profile
shows
no evidence
of
upslope
aerosols
and
is representative
of
a clean,
Rayleigh-only
lidar
profile.
Upslope
aerosols
are visible
up to approximately
5 km
for
the
07:15
GMT
and
10:15
GMT
prof'des
(during
daytime).
Figure
6b. Attenuated
Backscatter
Signals
(m sr) -_ on June
29,
1997.
Each
prof'fle
is a 15
minute
average,
starting
15 minutes
prior
to the
time shown.
Upslope
aerosols
are visible
up to approximately
5 km
for
the
17:15
GMT
and
19:15
GMT
prof'des
(during
daytime).
The
22:15
GMT
(after
sunset)
prof'fle
shows
no
evidence
of
upslope
aerosols
and
is
representative
of a clean,
Rayleigh-only
lidar
prof'de.
Figure
7a.
Lidar
aerosol
extinction
coefficient
(m l)
profiles
at 08:15,
10:15,
17:15,
and
18:15
GMT
on June
29,
1997.
The
columnar
AOD
for each
profile
is 0.017,
0.018,
0.015,
and 0.014
respectively.
The calculated
columnar
backscatter-extinction
ratios
(sr -1) for
each
prof'fle
are
0.018,
0.038,
0.031,
and
0.017
respectively.
The
Rayleigh
extinction
coefficient
profile
is
shown
for
comparison.
The
average
aerosol
extinction
coefficient
measured
by the IZO
nephelometer
from
07:00
to 19:00
GMT
is also
shown.
Draft
version
12-09-98
33
Draft:for
Tellus
ACE-2
special
issue
Figure
7b.
Lidar
aerosol
optical
depth
(AOD)
profiles
at 08:15,
10:15,
17:15,
and
18:15
GMT
on June
29,
1997.
The
columnar
AOD
for
each
profile
is 0.017,
0.018,
0.015,
and
0.014
respectively.
The
calculated
columnar
backscatter-extinction
ratios
(sr-')
for
each
profile
are 0.018,
0.038,
0.031,
and
0.017
respectively.
The
average
AOD
measured
by
the Teide
shadowband
is also
shown.
Figure
8.
Horizontal
lidar
profiles
of
the
natural
logarithm
of
the
attenuated
backscatter
coefficient
(m
sr) _.
A 15 minute
average
horizontal
profile
at
10:45
GMT
on
June
29
is
shown
along
with
a calculated
Rayleigh
horizontal
profile
for the IZO
altitude.
Figure
9. Three-day
attenuated
backscatter
signal
(ABS)
(m sr) _ time
series
from
July
16
through
July
18, 1997.
Each
major
tickmark
separates
the days,
and each
minor
tickmark
is
one
hour
(GMT).
Figure
10.
Lidar
aerosol
extinction
coefficient
(m _)
profdes
at 08:15,
10:15,
17:15,
and
18:45
GMT
on July
17, 1997.
The
columnar
AOD
for each
profile
is 0.161,
0.205,
0.226,
and 0.217
respectively.
The calculated
columnar
backscatter-extinction
ratios
(sr -_) for
each
prof'de
are
0.026,
0.048,
0.073,
and
0.027
respectively.
The
Rayleigh
extinction
coefficient
profile
is shown
for comparison.
Figure
1 la. Lidar
aerosol
extinction
coefficient
(m L) profiles
at 10:15
and
18:45
GMT
on
July
17,
1997.
The
columnar
AOD
for
each
profile
is 0.205
and 0.217
respectively.
The
calculated
columnar
backscatter-extinction
ratios
(sr -_) for each
profile
are 0.048
and
0.027
respectively.
The
average
aerosol
extinction
coefficients
measured
by the IZO
nephelometer
from
07:00
to 18:00
GMT
(daytime)
and after
18:00
GMT
are also
shown.
Draft
version
12-09-98
34
Draft:
for
Tellus
ACE-2
special
issue
Figure
1 lb.
Lidar
aerosol
optical
depth
(AOD)
profile
at 18:45
GMT
on July
17, 1997.
The
columnar
AOD
is 0.217
and
the
calculated
columnar
backscatter-extinction
ratio
(sr 1)
is
0.027.
The AOD
profile
measured
with
AATS-14
from
18:30
to 18:45
GMT,
and
the AOD
measured
by the IZO
cimel
and
the Teide
shadowband,
are also
shown.
Figure
12.
Horizontal
lidar
prof'des
of the
natural
logarithm
of the
attenuated
back.scatter
coefficient
(m sr) q.
A
15 minute
average
horizontal
profile
at 11:15
GMT
and
a 5 minute
average
horizontal
prof'de
at
18:50
GMT
on
July
17 are
shown
along
with
a calculated
Rayleigh
horizontal
profile
for the IZO altitude.
Figure
13. Schematic
representation
of the lidar
inversion
algorithm.
Draft
version
12-09-98
35
Figure 1
A
A
Laser
Pulses
(Transmitted
& Received)
T-R
Window
MPL Control
Housing
/
MPL
Telescope
MPL
Optics
&
Detector
/VlPL
Transmitter
-
Receiver
(T-R)
MPL
Scalar
Unit
I
Laser
Power
Supply
MPL
Control
Laptop
Computer
Figure
1. Schematic
diagram
of the rnicro-pulse
lidar
system
(not
to scale).
Figure2a
=1.
....¢
O
o
120
110
100
90
80
70
60
50
40
30
20
10
June
29
-.-----
July
15
_nnumlmul_umn__
__ a|
i m
nom
r__
._lilili'll_l_l_.
--
0
r
.ll{ll
t'-_
--.,tit'-
{
i
--ll|II'"
1111_
.Ill|;
0
2
4
6
8
Time
(GMT)
Figure
2a. Aerosol
total
mass
concentrations
measured
during
the early
mornings
of
June
29 and July
15 1997
at IZO.
Figure2b
o
ov,,q
8
0
u_
0
c_
<
1E-04
-
8E-05
6E-05
4E-05
2E-05
0E+00
June
29
-,-,-.-
July
15
_111P
I Ill
II Illll
ill
I ill
I
air I
Hi I IR iiill
I |
II till I111i ti |
a |
a |
mm
I _
I _
I
i
_
IIIIIII1_
tI
r
r
i
0
2
4
6
Time
(GMT)
Figure
2b. Aerosol
extinction
coefficients
measured
during
the early
mornings
of
June
29 and July
15 1997
at IZO.
Figure3a
1E+02
1E+01
1E-w0
0
_
1E-O1
0
1E-02
1E-03
1E-04
I
I
I
I
i
i
0
1
2
3
4
5
6
10
O(z),
June
29
"'"'"'-
O(z),
July
15
I
I
I
7
8
9
Range
(kin)
Figure
3a.
Overlap
functions,
O(z),
calculated
on June
29 and
July
15 1997.
Figure
3b
,<
#.
¢,O
0
-1
-2
31
I AzJun2
,,,,,o'*."
A(z),
July
15
-4
i
i
i
I
i
i
i
i
i
0
1
2
3
4
5
6
7
8
9
10
Range
(kin)
Figure
3b. Afterpulse
functions,
A(z),
calculated
on June
29 and July
15 1997.
Figure4
10
_
6-
4-
2
0
Illll
Signal(Measured)
Signal (Rayleigh Only)
Signal(Calibrated)
I
I
I
0.5
1
1.5
2
Signal
Figure
4. The
measured
MPL
signal,
a calculated
Rayleigh
lidar
signal,
and
the final
calibrated
MPL
signal
are shown
for 00:00
GMT
on June
30,
1997.
q3
-,--t
<
8.0
7.5
7.0
6.5
6.0
5.5
5.0
4.5
4.0
3.5
3.0
2.5
Figure
5
0
96
192
Minutes
(GMT)
June
29
to
July
i,
1997
0.0e+00
i. 5e-06
3.0e-06
4.5e-06
6.0e-06
Attenuated
Backscatter
i/(m
sr)
Figure
5. Three-day
attenuated
backscatter
signal
(ABS)
(m sr) 1 time
series
from
June
29 through
July
1, 1997.
Each
major
tickmark
separates
the days,
and each
minor
tickmark
is one
hour
(GMT).
Figure6a
10
g
w
\
\
\
iHiiiilunlmllll
NUmnmum
0615
GMT
0715
GMT
1015
GMT
I,.,.¢
2
I
I
I
0
5E-07
1E-06
1.5E-06
2E-06
Attenuated
Backscatter
Signal
1/(m sr)
Figure
6a. Attenuated
Backscatter
Signals,
l/(m
sr),
on June
29,
1997.
Each
prof'de
is a 15 minute
average,
starting
15 minutes
prior
to the time
shown.
The
06:15
GMT
(before
sunrise)
proftle
shows
no evidence
of upslope
aerosols
and
is representative
of a clean,
Rayleigh-only
lidar
prof'tle.
Upslope
aerosols
are visible
up to
approximately
5 km
for the 07:15
GMT
and
10:15
GMT
profiles
(during
daytime).
2
10
Figure6b
8_
1
4_
m|m|H|l
2
,
,
,
0
5E-07
1E-06
1.5E-06
1715
GMT
1915
GMT
2215
GMT
2E-06
Attenuated
Backscatter
Signal
1/(m sr)
Figure
6b. Attenuated
Backscatter
Signals,
1/(m
sr), on June
29,
1997.
Each
profile
is a 15 minute
average,
starting
15 minutes
prior
to the time
shown.
Upslope
aerosols
are visible
up to approximately
5 krn for the 17:15
GMT
and
19:15
GMT
prof'des
(during
daytime).
The
22:15
GMT
(after
sunse0
profile
shows
no evidence
of upslope
aerosols
and is representative
of a clean,
Rayleigh-only
lidar
prof'de.
q_
Figure
7a
4
<
81111 I1|11 ||111
jlBIn
D
nnlnlmmm
0815
GMT
1015 GMT
1715
GMT
1815 GMT
Rayleigh
IZO Neph.
(Daytime)
j,,,,,,,.,,,,,,"::"
I
I
I
I
0
5E-06
1E-05
1.5E-05
2E-05
2.5E-05
Extinction
Coefficient
(I/m)
Figure
7a.
Lidar
aerosol
extinction
coefficient
(l/m)
profdes
at 08:15,
10:15,
17:15,
and
18:15
GMT
on June
29,
1997.
The
columnar
AOD
for each
prof'de
is 0.017,
0.018,
0.015,
and
0.014
respectively.
The
calculated
columnar
backscatter-extinction
ratios
(l/sr)
for each
prof'de
are 0.018,
0.038,
0.031,
and 0.017
respectively.
The
Rayleigh
extinction
coefficient
proftle
is shown
for comparison.
The
average
aerosol
extinction
coefficient
measured
by the IZO
nephelometer
from
07:00
to 19:00
GMT
is also
shown.
Figure7b
7
.
0815
GMT
•,,,',,,',
1015
GMT
• • " • •
1715
GMT
•",-'-,•
1815
GMT
•
Teide
AOD
I
I
I
0
0.005
0.01
0.015
0.02
Aerosol
Optical
Depth
(AOD)
Figure
7b. Lidar
aerosol
optical
depth
(AOD)
profdes
at 08:15,
10:15,
17:15,
and
18:15
GMT
on June
29,
1997.
The
columnar
AOD
for each
profile
is 0.017,
0.018,
0.015,
and 0.014
respectively.
The
calculated
columnar
backscatter-exfinction
ratios
(l/sr)
for each
profile
are 0.018,
0.038,
0.031,
and
0.017
respectively.
The
average
AOD
measured
by the Teide
shadowband
is also
shown.
Figure 8
-13
-13.5
"_
-14
.9
-14.5
-15
June
29,
1045
GMT
•,.°,.,,,,
Rayleigh
I
iiiiiiIiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiIiIIiiiIiiiiiiiiiIiiIiii
I
t
_
_
_
Range
(kin)
Figure
8. Horizontal
lidar
profiles
of the natural
logarithm
of the attenuated
backscatter
coefficient,
l(m
st).
A 15 minute
average
horizontal
profile
at
10:45
GMT
on June
29 is shown
along
with
a calculated
Rayleigh
horizontal
profile
for the IZO
altitude.
6
5
n3
5
-,'4
,_
4
0
5
0
5
0
5
0
5
0
5
0
5
0
Figure
9
96
Minutes
(GMT)
July
16
r
0.0e+00
1.5e-06
3.0e-06
Attenuated
Backscatter
192
to
July
18,
1997
4.5e-06
6.0e-06
i/(m
sr)
Figure
9. Three-day
attenuated
backscatter
signal
(ABS)
(m sr) _ time
series
from
July
16 through
July
18, 1997.
Each
major
tickmark
separates
the days,
and each
minor
tickmark
is one hour
(GMT).
Figure 10
"._-I
|lllnnllll
IIIIII
in|l|ii
0815
GMT
1015
GMT
1715
GMT
1845
GMT
Rayleigh
2-
r
r
T
0.0E+O0
5.0E-05
1.0E-04
1.5E-04
2.0E-04
Extinction
Coefficient
(l/m)
Figure
10. Lidar
aerosol
extinction
coefficient
(l/m)
prof'des
at 08:15,
10:15,
17:15,
and
18:45
GMT
on July
17, 1997.
The
columnar
AOD
for each
prof'de
is 0.161,
0.205,
0.226,
and
0.217
respectively.
The
calculated
columnar
backscatter-extinction
ratios
(l/sr)
for each
profile
are 0.026,
0.048,
0.073,
and 0.027
respectively.
The
Rayleigh
extinction
coefficient
profile
is shown
for comparison.
Figure 1la
6
Olllllllll
1015
GMT
1845
GMT
IZO
Neph.
(Daytime)
IZO
Neph.
(After
1800
GMT)
IIII
Illllllllllllllllllllpot
IIIIIIIiii
"'llll_ll
I
J;
iiiiiiiiiiiiiiiiiiiiiiiiiii1'
2
i
i
i
0.0E+00
5.0E-05
1.0E-04
1.5E-04
2.0E-04
Extinction
Coefficient
(l/m)
Figure
1 la.
Lidar
aerosol
extinction
coefficient
(l/m)
proftles
at 10:15
and
18:45
GMT
on July
17, 1997.
The
columnar
AOD
for each
prof'de
is 0.205
and 0.217
respectively.
The
calculated
columnar
backscatter-extinction
ratios
(l/sr)
for each
profile
are 0.048
and 0.027
respectively.
The
average
aerosol
extinction
coefficients
measured
by the IZO
nephelometer
from
07:00
to 18:00
GMT
(daytime)
and after
18:00
GMT
are also
shown.
Figure 1lb
¢)
6
1
4_
I
I
I
I
I
1845
GMT
'"'"'"'
AATS-14
IZO AOD
O
Teide
AOD
0
0.05
0.1
0.15
0.2
0.25
0.3
Aerosol
Optical
Depth
(AOD)
Figure
1 lb.
Lidar
aerosol
optical
depth
(AOD)
profde
at 18:45
GMT
on July
17,
1997.
The columnar
AOD
is 0.217
and the calculated
columnar
backscatter-extinction
ratio
(l/sr)
is 0.027.
The
AOD
prof'de
measured
with
AATS-14
from
18:30
to 18:45
GMT,
and
the AOD
measured
by the IZO
cimel
and the Teide
shadowband,
are also
shown.
Figure 12
-11
-12-
-13
-14
-17
illllllgll
i
in
an
in
0
July
17,
1115
GMT
July
17,
1850
GMT
Rayleigh
Q
|||l||ii|llBln
1
Range
(km)
Figure
12. Horizontal
lidar
profiles
of the natural
logarithm
of the attenuated
backscatter
coefficient,
l(m
st). A 15 minute
average
horizontal
profile
at 11:15
GMT
and a 5 minute
average
horizontal
prof'fle
at 18:50
GlVlT on July
17 are shown
along
with
a calculated
Rayleigh
horizontal
profile
for the IZO
altitude.
Figure
13
START
Xm
Xm°|
Xm-2
.gL
Set RA= 1 [
Set [_A(Xm)= 0 [
I Calculate
_A(X_-I)
]
Calculate I_^(xr2)
[
]Calcula_l_^(xL)].
Determine new RAvalue
I._"/3a(x"_'
RAm,
=
T,t
'cAfrom indep, measurement
Xm
Xm-
1
xr2
XL
Set l_a(x_) = 0 [
Calculate 13A(X=_2)
Calculate _^(xO ]
all
% Diff.
in
R^
> 0.5
END
• Final
13A proffie
• Final RA determined
• Final o A profile
• OA= I_A(P_)-_
• Final
xA profile
Integrate o Aprofile
from
x_
to
_.
t
Figure
13. Schematic
representation
of the lidar
inversion
algorithm.