Preliminary results and uncertainties of
scattering measurements for
SORTIE
Michael Twardowski1, Scott Freeman, Jim Sullivan, Ron Zaneveld, Chuck Trees, and
the SORTIE Team
1WET Labs, Inc., Narragansett, RI
SORTIE IOP ObjectivesSORTIE IOP Objectives
1.1. Use IOPs in radiative transfer models to Use IOPs in radiative transfer models to help constrain uncertainties for help constrain uncertainties for radiometric measurementsradiometric measurements
2.2. Map horizontal-vertical spatial Map horizontal-vertical spatial variability in optical properties around variability in optical properties around stationstation
3.3. Evaluate and refine IOP measurement Evaluate and refine IOP measurement protocolsprotocols
4.4. Evaluate and refine IOP uncertaintiesEvaluate and refine IOP uncertainties
MASCOT packageMASCOT package
MASCOT: VSF (10:10:170 deg; 650 nm)ECOVSF: VSF (100, 125, 150 deg; 650
nm)ECOBB3: VSF (117 deg; 470, 532, 650
nm)AUV-B: total scattering (650 nm)
AC9: a and c (9)
Dolphin packageDolphin package
ACS, AC9, ECO BB3, ECO BB2C, SBE49 CTD, DH4 ACS, AC9, ECO BB3, ECO BB2C, SBE49 CTD, DH4 data handler, and Notus gearfinder pingerdata handler, and Notus gearfinder pinger
cpg(532) (m- 1)
freq
uenc
y
Collected while towing around MOBY 103,950 points includes data collected f rom 6-45 m
10 m time series next to MOBY 4,870 points
0.03 0.05 0.10 0.20
Longitude (°W)
MOBY
4.9 h towing track around MOBY site
Lati
tude
(°N
) Spatial Variability:Spatial Variability:
ccpgpg(532) at the MOBY site(532) at the MOBY site
Spatial Variability:Spatial Variability:ccpgpg(532) trace during tow (532) trace during tow
Estimated error
Spatial Variability:Spatial Variability:Hyperspectral Hyperspectral aa and and cc during during
towtow MOBY MOBY sitesite
Spatial Variability:Spatial Variability:Vertical profiles of cpg532Vertical profiles of cpg532
MOBY site Mamala Bay
VSF calibration protocolVSF calibration protocolRelationship between raw VSF counts () and
: angleDO: dark offsetSF: scaling factor (relative gain)L: pathlength (0.2 m): fraction of bp not reaching detectorbp: particulate scatteringat: total absorption
2 unknowns for each 2 unknowns for each : SF and : SF and
() = [() – DO()] * SF() * exp [ L * (bp * +
at) ]pathlength attenuation
term
VSF calibration protocolVSF calibration protocol
[ – DO()] = bp * [P() / SF()]* exp [ - L * (bp * + at ]
Since there is no “standard” for vicarious calibration, introduce particle standard with known phase function, P(). Solve for bp:
Inverting the above to solve for [() – DO()] , we obtain:
() = [() – DO()] * SF() * exp [ L * (bp * + at) ]
bp = [() – DO()] * [SF() / P() ] * exp [ L * (bp * + at) ]
Now there are 3 unknowns for each Now there are 3 unknowns for each : P, SF and : P, SF and
Important Point #1Important Point #1
For VSF measurements, For VSF measurements, dark offsets should dark offsets should
always be measured in-always be measured in-situsitu
MASCOT VSF calibrationsMASCOT VSF calibrations
12/15/06
Arizona Road DustArizona Road Dust
• If we assume apg650~0, we can solve for (P/SF) and with a nonlinear fit to the empirical data for each channel
*but in practice there are relatively large error bars with this method
[ – DO()] = bp * [P()/SF()]* exp [ - L * (bp * + at ]
MASCOT VSF calibrationsMASCOT VSF calibrations
Arizona Road DustArizona Road Dust
But we know something else: should be approximately constant for each detector
( )
All channels normalized to area
Detector field-of-view unimportant
= 0.56
[ – DO()] = bp * [P()/SF()]* exp [ - L * (bp * + at ]
MASCOT VSF calibrationsMASCOT VSF calibrations
Now we can solve for (P/SF) for each angleNow we can solve for (P/SF) for each angle
0
2000
4000
6000
8000
10000
12000
14000
16000
0 0.1 0.2 0.3 0.4 0.5
c650
co
un
ts-D
O
counts_10
counts_20
counts_30
counts_40
counts_50
counts_60
counts_70
counts_80
counts_90
counts_100
counts_110
counts_120
counts_130
counts_140
counts_150
counts_160
counts_170
Microspherical Microspherical beadsbeads
(P/SF)(P/SF)
[ – DO()] = bp * [P()/SF()]* exp [ - L * (bp * + at ]
MASCOT VSF calibrationsMASCOT VSF calibrations
Phase function for 1.992±0.025 um beads
MASCOT VSF calibrationsMASCOT VSF calibrations
Weighting functions for MASCOT angles
MASCOT VSF calibrationsMASCOT VSF calibrations
Phase function values for MASCOT angles
P()
MASCOT VSF calibrationsMASCOT VSF calibrations
theoretical Pempirical P/SF = SF()
Now all calibration parameters are solved
MASCOT VSF calibrationsMASCOT VSF calibrations
Calibrated VSFs from the AZRD exp’t
Concurrent ECO-VSF measurements
10 m binned VSFs from 10 m binned VSFs from HawaiiHawaii
MASCOT
ECO-VSF
profiles from Hawaiiprofiles from Hawaii
MASCOT
ECO-VSF
profiles from Hawaiiprofiles from Hawaii
MASCOT with ECO-VSF overlay
sw(150°,650 nm)
Pure water scatteringPure water scattering
Twardowski et al. 2007
Important Point #2Important Point #2
In clear water, accurate In clear water, accurate pure seawater VSF values pure seawater VSF values
are are criticalcritical for deriving for deriving particulate VSF valuesparticulate VSF values
Agreement with theoretical Agreement with theoretical modelingmodeling
Fournier-Forand phase functions
IntercalibrationIntercalibration
ECO-BB3 comparisonsECO-BB3 comparisons
3 different devices
More VSF comparisonsMore VSF comparisonsNY Bight: May 2007
MVSM (Ukrainian device)
MASCOT
ECO-VSF
MVSM (Ukrainian device at NRL)
ECOVSF
MASCOT
More VSF comparisonsMore VSF comparisonsNY Bight: May 2007
Backscattering analysis: Backscattering analysis: from Hawaii and NY Bightfrom Hawaii and NY Bight
apparent underestimation of bb by ECOVSF by few percent
Backscattering analysis: Backscattering analysis: from Hawaii and NY Bightfrom Hawaii and NY Bight
ECOVSF 3rd order polynomial (polyfit) method for obtaining bb
MA
SC
OT p
oly
fit
MASCOT fully integrated bb
ECOVSF polyfit
~8% underestimation of bb by polyfit method
~4% difference now
Important Point #3Important Point #3
Currently recommended
“polyfit” method appears to
underestimate bbp by a few percent
This is because the polyfit extrapolation This is because the polyfit extrapolation to 90 degrees from the 100-125-150 to 90 degrees from the 100-125-150
degree measurements is not quite steep degree measurements is not quite steep enough.enough.
Backscattering analysisBackscattering analysis
So why do we use it?
Ocean OpticsOcean Optics20002000
MonacoMonaco
Analysis of shape of VSF in Analysis of shape of VSF in backwardbackward direction direction
Sullivan et al. 2005: 532 nm ECO-VSF
150° vs 125°for 9 different
coastal US sites
Analysis of shape of VSF in Analysis of shape of VSF in backwardbackward direction direction
Adding 657 nm ECO-VSF NY Bight data
Analysis of shape of VSF in Analysis of shape of VSF in backwardbackward direction direction
Adding 650 nm MASCOT NY Bight data
Analysis of shape of VSF in Analysis of shape of VSF in backwardbackward direction direction
100° vs 125°for 9 different
coastal US sites
Sullivan et al. 2005: 532 nm ECO-VSF
Analysis of shape of VSF in Analysis of shape of VSF in backwardbackward direction direction
Adding 657 nm ECO-VSF NY Bight data
Analysis of shape of VSF in Analysis of shape of VSF in backwardbackward direction direction
Adding 650 nm MASCOT NY Bight data
MASCOT % variation inMASCOT % variation inbbbpbp normalized data normalized data (2 (2(()/b)/bbpbp))
[
Lowest prediction errors
in estimating backscattering
coefficient
Consistent with Oishi 1990; Boss and Pegau 2001
Important Point #4Important Point #4
a) a) pp(110-120) is best range to (110-120) is best range to pick a single angle pick a single angle measurement for estimating bmeasurement for estimating bbpbp
b) using 1 or a few angles to b) using 1 or a few angles to estimate bestimate bbpbp has merit has merit
c) No obvious spectral variability c) No obvious spectral variability in VSF shape was observedin VSF shape was observed
VSF UncertaintiesVSF UncertaintiesTable 3. Parameters from scattering measurements in the South Pacific central gyre. All values expressed in 10-4.
parameter
(nm)
462 532 650
t(117°) (m-1 sr-1)
raw uncertaintya0.17 0.044 0.016
t(117°) (m-1 sr-1)
estimated uncertaintyb0.2 0.05 0.02
swB(117°)c (m-1 sr-1) 2.72 1.52 0.66
bbswBc (m-1) 18.7 10.5 4.6
t(117°), mean ± (m-1 sr-1)
central gyre, 0-500 m3.2±0.3 1.77±0.16 0.79±0.15
bbp (m-1)
estimated uncertaintyd1.4 0.51 0.22
bbp, mean ± (m-1)
central gyre, 0-500 m2.7±1.5 1.42±0.87 0.71±0.81
bbp, mean ± (m-1)
central gyre, 300-500 m2.0±1.2 0.68±0.39 0.04±0.37
bbp, lowest measured (m-1) 0.92 0.37 ~0ai.e., random electronic errorbcomputed over 1-m depth bins; see textcpure water scattering computed from Buiteveld et al. (1994) at 20°C; [1 + 0.3(35)/37] adjustment for dissolved salts applied after Morel (1974)
Twardowski, M.S., H. Claustre, S.A. Freeman, D. Stramski, and Y. Hout. 2007. Optical backscattering properties of the “clearest” natural waters. Biogeosciences, 4, 1041–1058.
• Detailed Methodology• Detailed uncertainties
analysis • Dark offsets measured
in-situ for the first time• New values for pure
seawater scattering recommended
Summary Summary
1. For VSF measurements, dark offsets should be 1. For VSF measurements, dark offsets should be measured IN SITUmeasured IN SITU
2. In clear ocean water, accurate pure seawater VSF 2. In clear ocean water, accurate pure seawater VSF values are values are criticalcritical for deriving particulate VSF for deriving particulate VSF valuesvalues
3. More work may be needed to refine method of 3. More work may be needed to refine method of estimating bestimating bbpbp from 3-angle measurements from 3-angle measurements
4. Shape of VSF in the backward direction is 4. Shape of VSF in the backward direction is remarkably consistentremarkably consistent
i. i. pp(110°-120°) is the best range for picking a (110°-120°) is the best range for picking a singlesingle
angle measurement for estimating bangle measurement for estimating bbpbp
ii. Using 1 or a few angles to estimate bii. Using 1 or a few angles to estimate bbpbp has has meritmeritiii. No obvious spectral variability in VSF shape iii. No obvious spectral variability in VSF shape was was
observedobserved
