estimation of inherent optical ... - genius.ucsd.edugenius.ucsd.edu/public/reviews/b91_85250f.pdf1.2...

9
1Estimation of inherent optical properties using in-situ hyperspectral radiometer and MODIS data along the East Coast of New Caledonia Hiroshi Murakami a* , Cécile Dupouy b , Rüdiger Röttgers c , Robert Frouin d a Japan Aerospace eXploration Agency, 2-1-1 Sengen, Tsukuba, Japan; b M.I.O. AMU/IRD/CNRS/USTV, UMR 235, IRD Center of Nouméa, New Caledonia; c Institute for Coastal Research, Helmholtz-Zentrum Geesthacht, D-21502 Geesthacht, Germany; d Scripps Institution of Oceanography, UC San Diego, La Jolla, California, USA ABSTRACT Hyperspectral remote sensing reflectance (Rrs) was measured by a TriOS radiometer system along the East Coast of New Caledonia during the R/V Alis 03-13 October 2011 CALIOPE cruise. The TriOS system consists of radiance and irradiance sensors measuring in the spectral range 320-950 nm, at a spectral resolution of about 10 nm (sampled by every 3.3 nm), and within a 7-degree field-of-view for the radiance sensor. The method developed by Froidefond and Ouillon (2005) was used to determine Rrs, i.e., the radiance sensor was mounted on a small raft to measure upwelling radiance just below the surface, and Rrs was calculated by normalizing water-leaving radiance with downward solar irradiance measured on the ship deck. Inherent Optical Properties (IOPs), i.e., absorption coefficients of phytoplankton and detritus+dissolved substances (a ph and a dg , respectively), and particulate backscattering coefficient (b bp ) were estimated from the hyperspectral Rrs data by applying linear matrix inversion (Hoge and Lyon, 1996). The IOP inversion algorithm was adapted to MODIS data and applied to Level 1b imagery at 500 m resolution to demonstrate the feasibility of regular IOP monitoring from space in the study area. Local characteristics of the IOP spectra were used for the candidate spectra in the algorithm. The estimated MODIS Rrs and IOPs were evaluated using TriOS Rrs and in-situ IOP measurements obtained concomitantly during the cruise. Keywords: Inherent optical properties, hyperspectral, TriOS, MODIS, New Caledonia, CALIOPE 1. INTRODUCTION 1.1 The CALIOPE cruise The East coast of New Caledonia is characterized offshore by a coral reef lagoon system (bathymetry 20~30m) connected to the open ocean (100~1000m) by passes in the reef barrier. The water in the lagoon changes from oligotrophic to mesotrophic, with higher chlorophyll-a concentration in winter (July) or summer (February) after rainfall events [1] . The Institute for Research and Development (IRD) conducted the CALedonian IOP’s of the East coast of New Caledonia (CALIOPE) [2,3] cruise during 03-13 October 2011 to acquire bio-optical measurements for satellite ocean-color validation. IOP and Rrs data were collected at various stations along the East Coast, inside and outside the lagoon (Fig. 1). The measurements complemented observations made previously during VALHYBIO/VALHYSAT [1,4,5] cruises off the South and Southwest Coast of New Caledonia. ___________________________ *[email protected]; phone 81 50 3362 6586; fax 81 29 868 2961 Figure 1: location of CALIOPE and ValHySat stations (blue and orange dots). Grey tones indicate bathymetry. Remote Sensing of the Marine Environment II, edited by Robert J. Frouin, Naoto Ebuchi, Delu Pan, Toshiro Saino, Proc. of SPIE Vol. 8525, 85250F © 2012 SPIE · CCC code: 0277-786/12/$18 · doi: 10.1117/12.979245 Proc. of SPIE Vol. 8525 85250F-1 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/13/2013 Terms of Use: http://spiedl.org/terms

Upload: others

Post on 12-Jul-2020

12 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Estimation of inherent optical ... - genius.ucsd.edugenius.ucsd.edu/Public/Reviews/B91_85250F.pdf1.2 IOP estimation from Rrs IOP estimation from Rrs, e.g., Linear Matrix Inversion

1 on

150

200

250

300

350

Longitude1665 16]5

Estimation of inherent optical properties using in-situ hyperspectral radiometer and MODIS data along the East Coast of New Caledonia

Hiroshi Murakamia*, Cécile Dupouyb, Rüdiger Röttgersc, Robert Frouind a Japan Aerospace eXploration Agency, 2-1-1 Sengen, Tsukuba, Japan;

b M.I.O. AMU/IRD/CNRS/USTV, UMR 235, IRD Center of Nouméa, New Caledonia; c Institute for Coastal Research, Helmholtz-Zentrum Geesthacht, D-21502 Geesthacht, Germany;

d Scripps Institution of Oceanography, UC San Diego, La Jolla, California, USA

ABSTRACT

Hyperspectral remote sensing reflectance (Rrs) was measured by a TriOS radiometer system along the East Coast of New Caledonia during the R/V Alis 03-13 October 2011 CALIOPE cruise. The TriOS system consists of radiance and irradiance sensors measuring in the spectral range 320-950 nm, at a spectral resolution of about 10 nm (sampled by every 3.3 nm), and within a 7-degree field-of-view for the radiance sensor. The method developed by Froidefond and Ouillon (2005) was used to determine Rrs, i.e., the radiance sensor was mounted on a small raft to measure upwelling radiance just below the surface, and Rrs was calculated by normalizing water-leaving radiance with downward solar irradiance measured on the ship deck. Inherent Optical Properties (IOPs), i.e., absorption coefficients of phytoplankton and detritus+dissolved substances (aph and adg, respectively), and particulate backscattering coefficient (bbp) were estimated from the hyperspectral Rrs data by applying linear matrix inversion (Hoge and Lyon, 1996). The IOP inversion algorithm was adapted to MODIS data and applied to Level 1b imagery at 500 m resolution to demonstrate the feasibility of regular IOP monitoring from space in the study area. Local characteristics of the IOP spectra were used for the candidate spectra in the algorithm. The estimated MODIS Rrs and IOPs were evaluated using TriOS Rrs and in-situ IOP measurements obtained concomitantly during the cruise.

Keywords: Inherent optical properties, hyperspectral, TriOS, MODIS, New Caledonia, CALIOPE

1. INTRODUCTION 1.1 The CALIOPE cruise

The East coast of New Caledonia is characterized offshore by a coral reef lagoon system (bathymetry 20~30m) connected to the open ocean (100~1000m) by passes in the reef barrier. The water in the lagoon changes from oligotrophic to mesotrophic, with higher chlorophyll-a concentration in winter (July) or summer (February) after rainfall events[1] . The Institute for Research and Development (IRD) conducted the CALedonian IOP’s of the East coast of New Caledonia (CALIOPE)[2,3] cruise during 03-13 October 2011 to acquire bio-optical measurements for satellite ocean-color validation. IOP and Rrs data were collected at various stations along the East Coast, inside and outside the lagoon (Fig. 1). The measurements complemented observations made previously during VALHYBIO/VALHYSAT[1,4,5] cruises off the South and Southwest Coast of New Caledonia.

___________________________ *[email protected]; phone 81 50 3362 6586; fax 81 29 868 2961

Figure 1: location of CALIOPE and ValHySat stations (blue and orange dots). Grey tones indicate bathymetry.

Remote Sensing of the Marine Environment II, edited by Robert J. Frouin, Naoto Ebuchi, Delu Pan, Toshiro Saino, Proc. of SPIE Vol. 8525, 85250F

© 2012 SPIE · CCC code: 0277-786/12/$18 · doi: 10.1117/12.979245

Proc. of SPIE Vol. 8525 85250F-1

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/13/2013 Terms of Use: http://spiedl.org/terms

Page 2: Estimation of inherent optical ... - genius.ucsd.edugenius.ucsd.edu/Public/Reviews/B91_85250F.pdf1.2 IOP estimation from Rrs IOP estimation from Rrs, e.g., Linear Matrix Inversion

1.2 IOP estimation from Rrs

IOP estimation from Rrs, e.g., Linear Matrix Inversion[6,7,8], needs candidate spectra of absorption by detritus and colored dissolved organic matter (adg), backscattering by particles (bbp), and absorption by phytoplankton (aph). Retrieval accuracy depends on the validity of the candidate spectra, and specific selection may be required for the coastal areas. The TriOS sensor provides spectrally continuous Rrs, making it possible to use any wavelength for the inversion process.

Satellite observation is a useful way to characterize IOP spatial distribution and temporal change. MODIS (Aqua) has 500m resolution bands, which have the advantage of viewing coastal areas at a better resolution than the traditional 1-km resolution. However, simplification of the method (number of candidate spectra) is required to adopt the IOP estimation scheme because number of observation bands is limited.

In the standard algorithm, atmospheric correction of satellite ocean-color imagery uses aerosol reflectance to estimate marine reflectance. For consistency between the atmospheric correction and IOP estimation, we have proposed to estimate marine reflectance (or Rrs) simultaneously with the atmospheric correction [9].

1.3 Objective of the study

The objective of this study is to demonstrate the possibility of local optimization of satellite IOP estimation with benchmark in-situ measurements for the coastal observations. The optimal candidate IOP spectra are investigated from in-situ IOPs and analysis of TriOS Rrs for the East Coast of New Caledonia. The spectra are tested practically using the MODIS 500m data, which can have additional errors due to atmospheric correction, sunglint, and radiometric calibration.

2. DATA 2.1 TriOS Rrs

The TriOS system measured Lu(0−,λ) (FOV=7°) and Ed(0+,λ) from wavelength λ of 320 nm to 950 nm with Δλ~10 nm (sampled by every 3.3 nm). Above-surface remote sensing reflectance (Rrs) was then calculated according to the following equation [10]:

Rrs(λ) = c × Lu(0−,λ) / Ed(0+,λ), (1)

where c = F × t/n2 = 1.796 × 0.546 = 0.98. The factor F (equal to 1.796) takes into account the effect of FOV change between water/air.

2.2 PSICAM IOPs

Sampled water was analyzed by a Point-Source Integrating-Cavity Absorption Meter (PSICAM) developed at the GKSS Institute for Coastal Research [11,12]. It has a cavity illuminated by a central light source and fiber optic connected to a spectrometer. Spectral total absorption excluding pure water (apg = at − aw) and ag were obtained from measurements of seawater and 0.22-μm filtered seawater samples.

2.3 Chlorophyll-a concentration

Chlorophyll-a concentration was measured by the filter-pad technique using water samples filtered onto 25-mm GF/F Whatman filters. Sum of chlorophyll-a and divinyl chlorophyll-a concentrations (Chl-a) is used in this study for comparison with the satellite-estimated chlorophyll-a concentration.

2.4 MODIS 500m data

MODIS 500m resolution L1B imagery (version 005) in spectral bands centered on 646nm, 857nm, 466nm, 554nm, 1242nm, 1628nm, and 2114nm was used in this study. Radiance measured in bands 1-4 were corrected slightly (less than 3%) by a cross-calibration scheme[13] based on NASA MODIS Aqua Level 3 Rrs products produced by using 1km resolution bands. We processed the MODIS data from 4 to 15 Oct. 2011 in the region of the CALIOPE cruise.

Proc. of SPIE Vol. 8525 85250F-2

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/13/2013 Terms of Use: http://spiedl.org/terms

Page 3: Estimation of inherent optical ... - genius.ucsd.edugenius.ucsd.edu/Public/Reviews/B91_85250F.pdf1.2 IOP estimation from Rrs IOP estimation from Rrs, e.g., Linear Matrix Inversion

E

0.08

0.07

0.06

0.05

0.04

0.03

0.02

0.01

0350 400 450 500 550 600 650 700

Tri05 rrs

Initial settings: ro=0.5; =a.5; 1:1.03.6;

repeatwhite ctecrew g the ctii Terence

h4tt4mc4rrecticm

(1) dpj, b,,,, (3) apIv adu

hh..= ( wtw; ).-'

r, = 0.0949 U+ 0.0794 u'

N* = ar ny r,+:s + N.h+1(1- rp4:s)

a,1,=eXP(S(w-WO) /= rrh +,11/ (3.-rrn)

3. METHODS 3.1 IOP retrieval by linear matrix inversion

IOPs, i.e., aph(λ0), adg(λ0), bbp(λ0) (,rph, rpico, S, and Y), were estimated by the linear matrix inversion[6,7,8] method using as input rrs calculated from the TriOS hyperspectral Rrs and the following relation between IOP and rrs

[14]:

rrs = 0.0949 u+ 0.0794 u2 (2)

where u= bb/(bb+a), bb = bbw + bbp, and a = aw + aph + adg.

We solved the inversion matrix by separating processes, i.e., steps (1)-(5), as indicated in Fig. 2, because the many variables cause unstable solution, and separating processes allowed us to use different wavelengths of rrs. We tested two ways/approaches: (1)-(3) with fixed rpico, S, and Y, and (1)-(5) with variable rpico, S, and Y.

Figure 2: Operation flow of IOP estimation.

3.2 Model spectra

The model spectra used in this study are shown eq. (3)-(5) and Fig. 3. We used aw and bbw (=bw/2) from [15,16], and aph-Pico and aph-Micro from [17] extrapolated to 340nm-400nm by using [18].

aph(λ) = aph(λ0) × {aph-Pico(λ) × rpico

+ aph-Micro(λ) × (1 − rpico)} (3)

adg(λ) = adg(λ0) × exp{ S (λ−λ0) } (4)

bbp(λ) = bbp(λ0) × (λ/λ0)Y (5)

3.3 Selection of wavelength ranges

Each analysis (1)-(5) in Fig. 2 may exhibit different sensitivity to the wavelength range. We found the optimal wavelength ranges (Fig. 4) by testing error sensitivity of IOP in Eqs. (2)-(5) for various start-end wavelengths. We set 360-420nm for (e) to fit the PSICAM wavelengths. (a)-(e) correspond to processes (1)-(5) in Fig. 2.

3.4 Correction of bottom reflectance

Bottom reflectance influenced the TriOS Rrs in high transmittance and shallow areas (bathymetry <60m). We applied bottom reflectance correction according to [19] using the bottom reflectance spectrum reported in [20].

Figure 3: Candidate IOP spectra used in this study.

Proc. of SPIE Vol. 8525 85250F-3

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/13/2013 Terms of Use: http://spiedl.org/terms

Page 4: Estimation of inherent optical ... - genius.ucsd.edugenius.ucsd.edu/Public/Reviews/B91_85250F.pdf1.2 IOP estimation from Rrs IOP estimation from Rrs, e.g., Linear Matrix Inversion

410 420 EnEwaveieng,nlmi 510 520 o ,n aso awEnd via NM

wo o

e lg1n]w

510

Enwaveenth m

0.024

0.02

0.016

=2012

0.008

0.004

Station = 14

apt! aaW:<W-.iwmiCIp424-474

e19 5,360420nm

estimated Rrs_deep

TriOS R-d r3_deepfrom IOP

400 500 560Warelenqth (nmI

0.9

0.1

E 0.03

0.01

0.003

0.001

0.0003

Station = 14UTO= October 06, 01:36Lon Lot =166.480 - 21.670depth = 26.0m (26.0m, m =0.33)

P/IPaMI=2.00 ap0447=0.0505S=-0.022 adp442=0.0147

ELp442=0.1)043TcMa-0.360

a

a nmrov

bbW

000 650 700 900 400 460 500 500wM

600WepN Elm)

650 700

Figure 4: Optimal wavelength ranges for the IOP estimation. X-axis and Y-axis show start and end wavelength, respectively.

4. RESULTS 4.1 IOP estimation by TriOS Rrs

Figure 5 displays an example of the IOP estimation (at station 14 of CALIOPE). Rrs reproduced by the estimated IOPs (blue line in the left panel) agrees well with the Rrs from TriOS measurements (black line in the left panel).

Figure 5: An example of the IOP estimation from TriOS Rrs. Black line on the left panel show the TriOS observed Rrs, red broken line the Rrs corrected for bottom reflectance (Rrs

deep), and light blue line the Rrs calculated by the estimated IOPs. Right panel show the estimated IOPs, green, violet, and dark yellow lines are estimated aph, adg, and bbp respectively.

4.2 Comparison with in-situ data

Figures 6 and 7 display the in-situ and TriOS derived Chl-a and IOPs. The TriOS estimates revealed relatively high absorption and Chl-a along the main land and in the southern part (< −21.5°S) of the lagoon.

Agreement of shallow bathymetry (e.g., 10m) samples was improved by the bottom reflectance correction. Some samples of ag with 20-40m depths were overcorrected (Fig. 7 f), however, there is no overcorrection in the apg (Fig. 7 e).

4.3 Estimation of rpico, S

Figure 8 displays the spectral slope of ag (S) and aph/Chl-a ratio. The in-situ at−ag includes detritus absorption (ad).TriOS aph/Chl-a ratio was calculated by the estimated aph_pico mixing ratio and aph/Chl-a according to [17]. Both diagrams show large scatter and have no significant correlations.

Proc. of SPIE Vol. 8525 85250F-4

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/13/2013 Terms of Use: http://spiedl.org/terms

Page 5: Estimation of inherent optical ... - genius.ucsd.edugenius.ucsd.edu/Public/Reviews/B91_85250F.pdf1.2 IOP estimation from Rrs IOP estimation from Rrs, e.g., Linear Matrix Inversion

-20

-200

b

-21 fi

1648 165 16'.6 166

Longitude166.4 1668 1672 1648 165.2 1656 16E 1644 1668 1672 1640 1652 1656 166 1664 1668 1672

Langibtla Lmgiiutle

0.03 0.1 0.3 3 0.003 0.01 0.03 01 _, 0.3 0.003 0.01 003 0.1 (.13

In situ TChI -a [mg m-3] In situ at @442nm (m( In situ a @442nm [m

5 10 20 50 100 200

Bathymetry [m] Bathymeóy [m] Bathymetry [m]5 10 20 50 100 200 5 10 20 50 100 200

-n

-224

104 .0 1071 16. 6 500 1004 1000 1

Lo.nwx

0.03 0.1 0.3

Tri05 0C4 Chl-a [mg m

104 0 10.2 105 1 4 : 4 1 7 0 0 10 . 1-v 0 1 0 5 . 1050 100 iF.l 1W0 167.1

0 Lm9160e LmJNh

0303 301 003 01 03Trios @442nm [m 7

0003 001 003 0.1 03Tnos a63 @442nm [rñ tl

Figure 6: Chl-a and IOP distribution at CALIOPE sites. Upper and Lower panels show the in-situ and TriOS derived IOPs.

Figure 7: Scatter diagrams between in-situ and TriOS derived IOPs. Upper and lower panels indicate estimates without bottom reflectance correction and with correction. Statistics, x-axis and y-axis averages (xav and yav), root mean square difference (rmsd), and correlation coefficient (r), are calculated in log10 scale. Dot colors indicate site bathymetry shown by the scale bar.

Proc. of SPIE Vol. 8525 85250F-5

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/13/2013 Terms of Use: http://spiedl.org/terms

Page 6: Estimation of inherent optical ... - genius.ucsd.edugenius.ucsd.edu/Public/Reviews/B91_85250F.pdf1.2 IOP estimation from Rrs IOP estimation from Rrs, e.g., Linear Matrix Inversion

-0.008

0 -013

ES(.0rn

-0.018at°

Ó-0.023

-0 -028-0.028 -0.023 -0.018 -0.013 -0.008

In situ a S [nm 1] in 360 -420nm

y=35xau=-0.0nyav=-0.015rmsd=0.008

.

r=0.124

5 10 20 50 100 200Bathymetry [m]

Q0.24

Úo..o 018

0.12aE

m

m°rnO

0.06

00 0.06 0.12 018 0.24

In situ (ai a8)/Chl -a (m2 mg 1]

N=38xav=0.070yav=0.034rmsd=0.063r=-0.060

5 10 20 50 100 200Bathymetry [m]

satellite= rt (fl-rup.)

Set

Sett)" = O - bbpol:mo

Startfrom here

=NIR, red. SWIRU011-) r(?.)

= -a.= log(rp.)/ r(?.-Nmjj 14:)ge.-/ ?.,Nm)

= \ \ 10"= -

[ÌOP]-) ;Imo' and bbpo'

Ste:rp it laro, - 0.0001

apo. bhp,» rw

bottom reflectancecorrection

Figure 8: Scatter diagrams between in-situ and TriOS retrieved S and aph/Chl-a or (at−ag)/Chl-a

ratio. Dot colors indicate site bathymetry.

5. APPLICATION TO MODIS 500M L1B IMAGERY 5.1 Simultaneous retrieval of IOP and atmospheric parameters

MODIS has seven 500m bands but only two in blue-green wavelength range. We used fixed slope parameters, rph=0.57, rpico=0.51, S=−0.014, and Y=−1 to apply the IOP scheme to the MODIS data. Those rph, rpico, and S values are averages estimated from PSICAM measurements. A schematic description of the inversion scheme is given in Fig. 9.

The bottom correction was applied after the bbp and apg iteration because the correction can make the IOP estimation unstable if we apply within the IOP iteration.

Figure 10 shows an example of the MODIS estimation. We notice higher apg and bbp in the Southern part of the lagoon along the East Coast of New Caledonia, and in regions of outflow through passes in the reef barrier. The outflow structure changed day by day, which could be captured by the estimate during the CALIOPE period (not shown here).

Figure 9: Flowchart of IOP estimation using MODIS 500m L1B data

Proc. of SPIE Vol. 8525 85250F-6

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/13/2013 Terms of Use: http://spiedl.org/terms

Page 7: Estimation of inherent optical ... - genius.ucsd.edugenius.ucsd.edu/Public/Reviews/B91_85250F.pdf1.2 IOP estimation from Rrs IOP estimation from Rrs, e.g., Linear Matrix Inversion

1640 165 1053 105.4 1050 1053 160 1603 106.4 1500 1009 107 1675 107.4

Longitude1648 165 165] 1614 1658 1650 166 1663 1664 1656 1666 167 167E 1670

LunyNtle

0.024

0.018-NCG

0.012

cc

O 0.0062

0

N=39 (a) R,.S 466nxav=0.00608yav=0.00717rmsd=0.00182r=0.551

N=11xav=0.00634yav=0.00854rmsd=0.00080r=0.740

0 0.006 0.012 0.018

In situ Rrs 466nm [sr -1]

0.016

7-0.012

ETrC

Le) 0.008

cc

0.004z

N=39 (b) R,.S 5 54nix(xav=0.00262yav=0.00271rmsd=0.00055r=0.924

oo

N=11xav=0.00185yav=0.00216rmsd=0.00073r=0.029

0.0240

0 0.004 0.008 0.012 0.016

In situ Rrs 554nm [sr -1]

1

r0.3

C0.1

Tr@.)

0.03

0.01

0.0030.003 0.01 0.03 0.1 0.3

In situ au9

@442nm [m 1 ]

N=35xav=- 1.489yav=-1.541rmsd=0.131r=0.883

(c) apg 442nrri

1

3N=38 (d) OC2M Chl-xav=-0.610yav 0.622rmsd=0.184r=0.834

0.3

0.1

0.03

M o

Aso

o:1km L2N=11xav= 0.755yav=-0.856rmsd=0.208r=0.726

0.03 0.1 0.3

3N=38xav=-0.610yav=-0.853

1 rmsd=0.288r=0.870

0.3

(e) am-> Ch1-by NOMA

««so

3N=38xav=-0.610yav=-0.635

1 rmsd=0.107r=0.870

0.3

0.1 + N 0.1

.0.03 - - 0.03

0.03 0.1 0.3 3

(f) apg-> Ch1-by CALIOP

r

0.03 0.1 0.3 3

Figure 10: apg and bbp at 442nm on 8 Oct. 2011 estimated from MODIS 500m data (with the bottom correction).

5.2 Validation of MODIS 500m estimates

Figure 11: Scatter diagram between in-situ and MODIS estimates. In-situ Rrs were obtained by TriOS. Colors show bathymetry. Gray circle (of Y-axis) show MODIS standard level-2 Rrs and Chl-a[21]. MODIS OC2M algorithm was applied to our Rrs estimates from MODIS 500m L1B data with the bottom reflectance correction (d).

Proc. of SPIE Vol. 8525 85250F-7

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/13/2013 Terms of Use: http://spiedl.org/terms

Page 8: Estimation of inherent optical ... - genius.ucsd.edugenius.ucsd.edu/Public/Reviews/B91_85250F.pdf1.2 IOP estimation from Rrs IOP estimation from Rrs, e.g., Linear Matrix Inversion

Figure 11 shows the validation results for the MODIS estimates. Chl-a and apg could be derived with accuracy similar to that of the MODIS 1km standard products[21] and of our TriOS estimates. The number of the MODIS 1km samples is less than the number of the 500m samples, because land and small islands in the lagoon sometimes masked the near coastal area. We tried to estimate Chl-a from apg by using correlation between apg and Chl-a. When we used the relation in NOMAD IOP dataset[22], Chl-a was underestimated (Fig. 11 e). The relation obtained from the CALIOPE data improved the estimates from the apg (Fig. 11 f).

6. SUMMARY AND CONCLUSIONS 6.1 TriOS Rrs inversion

The IOP estimated from TriOS Rrs showed good agreement with in-situ IOPs for apg (r=0.88) and ag (r=0.62) on log10 scale. However, we could not achieve sufficient accuracy for the shape parameters, S, rpico, (and Y).

The bottom correction improved the results generally; however, sometimes it produced error on adg. This error affects less the retrieval of apg (i.e., total of aph and adg). We need to investigate more about the candidate IOP spectra and rb, which was not derived from our target area, the East Coast of New Caledonia.

6.2 Application to MODIS 500m data

Our inversion scheme could be applied to MODIS 500m resolution data with simplification of the variables and simultaneous simple atmospheric correction. The estimates showed good agreement with in-situ measurements for Rrs (r=0.55 at 466nm and 0.92 at 554nm), apg (r=0.88 in log10 scale), and Chl-a (r=0.87 in log10 scale).

Chl-a could be estimated by apg by using the correlation between apg and Chl-a in CALIOPE measurements. This suggests a possibility to estimate Chl-a avoiding the effect of bbp, which affect the blue-green type algorithms, if we have information about the local relation between apg and Chl-a.

6.3 Optimization of the model spectra

Satellite retrievals are improved by using the optimized model spectra obtained from in-situ IOP (and spectral Rrs) measurements in the target area. The potential and limitation of the local optimization need to be further evaluated for the satellite missions targeting the global coastal area such as GCOM-C/SGLI and Sentinel-3/OLCI.

6.4 IOP along the New Caledonian coast

TriOS (and MODIS) could estimate apg and ag distribution in this study. However, there was no significant freshwater event in the CALIOPE period. We need more observations & investigation in various conditions such as rainfall events. This will be the objective of the second CALIOPE cruise, planned for July-August 2013.

ACKNOWLEDGEMENTS

In-situ data was obtained in the frame of INSU-EC2CO CALIOPE project. The apg-Chla relation was calculated using NOMAD Version 2.0 ALPHA, which was compiled by the NASA Ocean Biology Processing Group (OBPG), Goddard Space Flight Center (GSFC). MODIS 500m resolution L1B (ver. 005) data is obtained from NASA GSFC the Level 1 and Atmosphere Archive and Distribution System (LAADS). MODIS L2 Rrs and chlorophyll-a data were provided by NASA OBPG. R. Frouin is supported by NASA under grants NNX11AR07G and NNX11AQ22G.

REFERENCES

[1] Dupouy, C., Neveux, J., Ouillon, S., Frouin, R., Murakami, H., Hochard, S., and Dirberg, G., "Inherent optical properties and satellite retrieval of chlorophyll concentration in the lagoon and open ocean waters of New Caledonia, " Marine Pollution Bulletin 61, 503–518, doi:10.1016/j.marpolbul.2010.06.039 (2010).

[2] Dupouy, C. , "Bordure partagée inférieure, SISMER, Ifremer," http://www.ifremer.fr/sismer/FR/catal/campagne/campagne.htql?crno=11100090 (2011).

Proc. of SPIE Vol. 8525 85250F-8

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/13/2013 Terms of Use: http://spiedl.org/terms

Page 9: Estimation of inherent optical ... - genius.ucsd.edugenius.ucsd.edu/Public/Reviews/B91_85250F.pdf1.2 IOP estimation from Rrs IOP estimation from Rrs, e.g., Linear Matrix Inversion

[3] Dupouy, C., "Une mission fructueuse, à renouveler en saison humide / CALIOPE 3 au 13 oct 2011," http://nouvelle-caledonie.ird.fr/layout/set/print/toute-l-actualite/carnets-de-bord/caliope-3-au-13-oct-2011/une-mission-fructueuse-a-renouveler-en-saison-humide (2011).

[4] Dupouy, C., Savranski, T., Lefevre, J., Despinoy, M., Mangeas, M., Fuchs, R., Ouillon, S., and Petit, M., “Monitoring chlorophyll of the South West Tropical Pacific,” Communication at the 34th International Symposium on Remote Sensing of Environment, Sydney, 10-14 April 2011 (2011).

[5] Dupouy, C., Wattelez, G., Fuchs, R., Lefèvre, J., Mangeas, M., Murakami, H., and Frouin, R., “The colour of the Coral Sea,” In Proceedings of the 12th International Coral Reef Symposium, 18E– The future of the Coral Sea reefs and sea mounts, Cairns, Australia, 9-13 July 2012, ICRS2012_18E-2 (2012).

[6] Hoge, F.E., and Lyon, P.E., "Satellite retrieval of inherent optical properties by linear matrix inversion of oceanic radiance models: an analysis of model and radiance measurement errors," Journal of Geophysical Research, 101, 16631-16648 (1996).

[7] Hoge, F.E., and Lyon, P.E., "Spectral parameters of inherent optical property models: Methods for satellite retrieval by matrix inversion of an oceanic radiance model," Applied Optics, 38, 1657-1662 (1999).

[8] Lyon, P., and Hoge, F., "The Linear Matrix Inversion Algorithm," Chap. 7 of IOCCG Report Number 5, Remote Sensing of Inherent Optical Properties: Fundamentals, Tests of Algorithms, and Applications, Ed. by Z. Lee (2006).

[9] Murakami, H and Dupouy, C., "Coastal ocean atmospheric correction for AVNIR-2 high resolution images," SPIE Asia-Pacific Remote Sensing, 7858, 785802 (2010).

[10] Froidefond J. M. and Ouillon, S., "Introducing a mini-catamaran to perform reflectance measurements above and below the water surface," Opt Express, 13, 926 (2005).

[11] Röttgers, R., Schönfeld, W., Kipp, P. -R., and Doerffer, R., "Practical test of a point-source integrating cavity absorption meter: the performance of different collector assemblies," Appl. Opt., 44, 5549-5560 (2005).

[12] Röttgers, R., and Doerffer, R., "Measurements of optical absorption by chromophoric dissolved organic matter using a point-source integrating-cavity absorption meter," Limnol. Oceanogr.: Methods 5, 126–135 (2007).

[13] Murakami, H., Yoshida, M., Tanaka, K., Fukushima, H., Toratani, M., Tanaka, A., and Senga, Y., "Vicarious Calibration of ADEOS-2 GLI Visible to Shortwave Infrared Bands Using GLobal Datasets, "IEEE Trans. Geosci. and Remote. Sens. Vol. 43, No. 7, 1571-1584 (2005).

[14] Gordon, H.R., Brown, O.B., Evans, R.H., Brown, J.W., Smith, R.C., Baker, K.S., and Clark, D.K., "A semi-analytic radiance model of ocean color," Journal of Geophysical Research, 93 (D9), 10909–10924 (1988).

[15] Pope, R.M., and Fry, E.S., "Absorption spectrum (380-700 nm) of pure water. II. Integrating cavity measurements," Applied Optics,36, 8710-8723 (1997).

[16] Kou, L., Labrie, D., and Chylek, P., "Refractive indices of water and ice in the 0.65-2.5 μm spectral range," Applied Optics, 32, 3531-3540 (1993).

[17] Ciotti, A.M., Lewis, M.R., and Cullen, J.J., "Assessment of the relationships between dominant cell size in natural phytoplankton communities and spectral shape of the absorption coefficient," Limnology and Oceanography, 4, 404-417 (2002).

[18] Suzuki, K., Kishino, M., Sasaoka, K., Saitoh, S. and Saino, T., "Chlorophyll-Specific Absorption Coefficients and Pigments of Phytoplankton off Sanriku, Northwestern North Pacific," J. Oceanogr., 54, 517-526 (1998).

[19] Lee, Z-P., Carder, K.L., Mobley, C.D., Steward, R.G., and Patch, J. S., "Hyperspectral remote sensing for shallow waters: 2. Deriving bottom depths and water properties by optimization," Applied Optics, 38, 3831-3843 (1999).

[20] Maritorena, S., Morel, A., and Gentili, B., “Diffise reflectance of oceanic shallow waters: Influence of water depth and bottom albedo,” Limnol. Oceanogr., 39(7), 1689-1703 (1994).

[21] Franz, B. A., "Methods for Assessing the Quality and Consistency of Ocean Color Products," http://oceancolor.gsfc.nasa.gov/REPROCESSING/R2009/validation/ (2005).

[22] Werdell, P.J., and Bailey, S.W., "An improved bio-optical data set for ocean color algorithm development and satellite data product validation," Remote Sensing of Environment, 98, 122-140 (2005).

Proc. of SPIE Vol. 8525 85250F-9

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/13/2013 Terms of Use: http://spiedl.org/terms