Optical Remote Sensing of marine and inland waters · 2018-09-17 · 2. Laboratoire d’Ecologie...

Belgian Earth Observation Day

[1/28]

www.mumm.ac.be/BELCOLOUR

Oudenburg, 25th May 2011

Optical Remote Sensing of marine and inland waters Main results of the BELCOLOUR-2 project (2007-2011)

1. Management Unit of the North Sea Mathematical Models (MUMM/RBINS)

2. Laboratoire d’Ecologie des Systèmes Aquatiques (ULB/ESA)

3. Vlaamse Instelling voor Technologisch Onderzoek (VITO)

4. Unité d’Océanographie Chimique, Université de Liège (ULg)

5. Laboratoire d’Océanographie de Villefranche, France (LOV)

6. CSIRO Land and Water, Australia (CSIRO)

MUMM


[2/28]



“What‟s in the water”?

(How much information can we extract?)

Mass concentration of particles?

Size distribution of particles?

Algal or non-algal?

What species of algae?

How much carbon is it fixing/day?

Dissolved CO2 concentration?

Conc/typ of Dissolved organic matter?

How transparent?

+ space-time variation (and cause/effect)

BUT

Viewing through atmosphere

Possibly also viewing sea bottom

Sediment

transport

Ecosystem

dynamics

(phytoplankton)

Carbon cycle

Water quality

(EU monitoring)

Marine science

(inc. benthos)

1. Introduction

2. Application Results

3. Science Results

4. Conclusions

Applications


[3/28]



History

Forel-Ule measurements [Schott, 1944!] in [Wernand, 2010]

SeaWiFS chlorophyll a [NASA]

1. Introduction


3. Science Results

4. Conclusions


[4/28]



The BELCOLOUR-2 project

Objectives:

– To improve the quality of chlorophyll and total suspended matter-based

products

– To develop new optical remote sensing products, including pCO2 and

primary production

Project duration: January 2007 – December 2011

Funded by Belgian Federal Science Policy Office, STEREO

programme (SR/00/03): 1.07M€ (+0.16 M€ RECOLOUR)

5 years funding

Easy admin (yearly reporting, budget transfers)

Constructive support and Steering Committee

Funding for international partners

Partners: MUMM, ULB/ESA, VITO/TAP, ULg-Chim., LOV (Fr), CSIRO

(Aus) + RECOLOUR (2007-8): ULg/GHER + ad hoc intl. (ULCO, etc.)

1. Introduction


3. Science Results

4. Conclusions


[5/28]



The Partners

BELCOLOUR-2 BE:

– MUMM: algorithms, satellite imagery, env . apps

– ULB/ESA: phytoplankton, primary production, IOPs

– V ITO/TAP: airborne, atmospheric correction, inland waters

– ULg-Chim: carbon dynamics, air-sea flux, CO2

BELCOLOUR-2 foreign:

– LOV: scattering, turbid waters, suspended matter

– CSIRO: algorithms, satellite imagery, env. apps

RECOLOUR

– ULg-GHER: statistical analysis, EOF, data filling

1. Introduction


3. Science Results

4. Conclusions


[6/28]



Project structure

7. APPLICATIONS EXTERNAL

PROJECTS/USERS

4. IMAGERY

2. THEORY

5. QUAL. CONTROL 3. MEASUREMENTS

RECOLOUR

1. Introduction


3. Science Results

4. Conclusions


[7/28]



Applications

1. Introduction


3. Science Results

4. Conclusions


[8/28]



App #1: Eutrophication monitoring (optimisation)

Reorganisation of BE monitoring network

2006: Satellite data used to optimise sampling,

but not trusted for actual monitoring

2011: Further reduction of resources, increase in

monitoring needs => interest in satellite data for

actual monitoring (+species detection??)

In situ: 4-10 data/year, Satellite: 30-40 data/year

1. Introduction


3. Science Results

4. Conclusions

Before 2007 (17+2) After 2007 (10+2)

[Ruddick et al, in “Remote Sensing of the European Seas”, 2008]


[9/28]



App #2: Eutrophication assessment

90 percentile CHL in 2006 (OSPAR Eutrophication cttee)

[MERIS data from

ACRI/ MARCOAST,

processed MUMM]

Satellite data is:

Neutral, transparent

Spatially extensive

Cross-boundary

Relevant for:

OSPAR “screening”

Supplementing in situ

[F. Gohin]

… Marine Strategy Framework Directive (2014)

1. Introduction


3. Science Results

4. Conclusions

[In “Second OSPAR Integrated Report on the Eutrophication Status …”, 2008]


[10/28]



App #3: Ecosystem model support

1. TSM for light forcing

2. CHL for validation

[Lancelot et al, 2005]

[Lacroix et al, 2007]

2008 4-season TSM

climatology

2010 RECOLOUR daily

TSM 2003-6

2012 SEVIRI/DINEOF

15-min KdPAR?

1. Introduction


3. Science Results

4. Conclusions


[11/28]



App #4: Algal bloom timing in the North Sea

MAR APR MAY

1. Early blooms in Norwegian coastal current, central North Sea

2. Late blooms in turbid water in east UK coast, German bight, western channel

=>Significant spatial and interannual variabilities in the North Sea

ICES fisheries

working groups

1. Introduction


3. Science Results

4. Conclusions

[Park et al, Int. J. Rem Sens. 2011]


[12/28]



App#5: Inland water monitoring

Review of Water Framework Directive reqts [VITO/BELCOLOUR report]

[Map from http://www.ciwvlaanderen.be/]

1. Introduction


3. Science Results

4. Conclusions


[13/28]



App #6: Underwater light climate (fish genetics)

Wavelength of maximally

transmitted light (WMTL)

from MODIS data of 2007

Light climate influences

fish evolution (vision)?

412 550 700nm

Future perspectives:

Use of optical remote sensing

data of light climate for

applications in habitat

adaptation of visual predators

… evolution studies??

1. Introduction


3. Science Results

4. Conclusions

[Larmuseau et al, Molecular Ecology, 2009]


[14/28]



Science results (selected)

1. Introduction


3. Science Results

4. Conclusions


[15/28]



)700(*)700(*)]480(/57.0

))450(/43.0())467(/1[()467(3

www

wwc

a

a

Reflectance algorithm

)480(*57.0)450(*43.0)467()467(3 tttc aaaa

Absorption algorithm

0.0

0.5

1.0

1.5

2.0

2.5

3.0

440 450 460 470 480 490 500

Wavelength (nm)

467 Extra chl c3

450 480

Detection of Phaeocystis globosa

1. Introduction


3. Science Results

4. Conclusions

[Astoreca et al, J. Plankton Research, 2009]

Wavelength (nm)

Normalised

absorption Phaeocystsis

Diatoms

Abso

rption

(m

-1)


[16/28]



0

0.05

0.1

0.15

0.2

0.25

350 450 550 650 750 850

wavelength (nm)

wa

ter-

lea

vin

g r

efl

ec

tan

ce

(-)

noctiluca 29/06/2005 few noctiluca 23/06/2005 BRC4 (turbid!)

few noctiluca 23/06/2005 BRCT3 (turbid!) phaeocystis 23/04/2007 702

turbid 25/04/2007 HARE intermediate turbid 19/09/2006 330

clear 20/09/2006 MH2E noctiluca experiment full bucket

noctiluca experiment half bucket noctiluca AHS

Detection of Noctiluca scintillans

Main differences

- high red and NIR

- sharp increase 520 -580 nm

1. Introduction


3. Science Results

4. Conclusions

[Van Mol et al, EARSEL eProceedings, 2007]


[17/28]



Spectral variation of scattering from near infrared to visible

The relation between Δbp and a p is function of particle size dist.

Next step: similar analysis for backscatter (with new instrument) …

1. Introduction


3. Science Results

4. Conclusions

[Doxaran et al, Limnology and Oceanology, 2009]

Pa

rtic

ula

te s

ca

tte

rin

g (

m-1

)

Pa

rtic

ula

te a

bso

rptio

n (

m-1

)


[18/28]



• Spectral variation depends on

particle type and size

• Phytoplankton ~7µm give

anomalous positive slope

• In situ measurements confirmed

by lab experiments and Mie theory

simulations

[Astoreca et al, submitted to Continental Shelf Research, 2011]

1. Introduction


3. Science Results

4. Conclusions

Spectal variation of particulate scattering

Beam

att

enuation (

m-1

)

“Normal” spectrum “Algal bloom”

spectrum


[19/28]


Oudenburg, 25th May 2011 Oostende, 19th May 2011

Mass-specific Particulate beam attenuation (cp) and backscatter (bbp)

[Neukermans et al, submitted to Limnology and Oceanography, 2011]

Mass-s

pecific

backscatter

(m2g

-1)

Mass-s

pecific

beam

att

en

ua

tio

n (

m2g

-1)

Apparent density (m2g-1) Particle diameter (m2g-1) Organic fraction (Carbon)

!

Regions:

S. North Sea

(BE/UK/FR/NL)

Channel (FR/EN)

Atlantic (FR/SP)

Ligurian (IT/FR)

Fr. Guyana

1. Introduction


3. Science Results

4. Conclusions


[20/28]



One band TSM retrieval algorithms

Remote-sensing reflectance, Rrs, at any single wavelength, λ

, is almost linearly related to Total Suspended Matter, S

1rs

rs

AS R

R C

0.000

0.010

0.020

0.030

0.040

0.050

0.060

0.00 50.00 100.00 150.00 200.00

Total Suspended Matter (mg/l)

Re

mo

te S

en

sin

g R

efl

cta

nc

e 620nm665nm708nm778nm865nm

LINEAR (optimal)

SATURATION

WEAK SIGNAL

1. Introduction


3. Science Results

4. Conclusions

[Nechad et al, Rem. Sens Env., 2010]


[21/28]



Estimated pCO2@10

C

Apr

2007

[Borges et al, ESA Living

Planet Proceedings, 2010] In situ

pCO2@10

C

Dissolved CO2

0

100

200

300

400

500

600

0 100 200 300 400 500 600

Derived pCO2 (ppm)O

bserv

ed

pC

O2 (

pp

m)

April

July

Sept.Jul

2007

Sep

2007

1. Introduction


3. Science Results

4. Conclusions

pCO2 algorithm has input:

- hydrodynamic model salinity

- satellite Chlorophyll


[22/28]


[Schroeder et al, in prep]

Atmospheric correction inter-comparison

1. Introduction


3. Science Results

4. Conclusions

2 examples

All 47 matchups

Diffe

ren

ce s

ate

llite

- in

situ


[23/28]



Adjacency detection and correction with the NIR similarity

spectrum: application to airborne and MERIS data

0

10

20

30

40

50

60

70

400 450 500 550 600 650 700 750 800 850 900

wavelength (nm)

Ra

dia

nc

e (

mW

/ m

2 S

r n

m)

uncorrected

NIRsim correction

ICOL correction

background (5by5 average)



Adjacency detection Adjacency correction

[Sterckx et al, Int. J. Rem. Sens, in press]

1. Introduction


3. Science Results

4. Conclusions


[24/28]



Pure water absorption coefficient (Pope & Fry,

1997; Kou et al., 1993)

In situ evidence of non-zero SWIR reflectance in the Scheldt

Simulated water reflectance with Hydrolight

APEX water reflectance

ASD water reflectance

[Knaeps et al, submitted to Rem. Sens. Env]


[25/28]


Oudenburg, 25th May 2011 Oostende, 19th May 2011

TSM mainly driven by tidal

resuspension with

max(TSM) occurring about

3h after max(BottomStress)

Good overall correlation

between SEVIRI TSM

and CEFAS SB TSM, but

large deviations occur.

[Neukermans, in prep.]

Mapping of tidal variability of TSM by SEVIRI

TSM from CEFAS buoy (g/m3)

10

TS

M f

rom

SE

VIR

I (g

/m3)

100 1

1. Introduction


3. Science Results

4. Conclusions


[26/28]



Rio de la Plata

(TSM)

Vietnam

(SST) Congo River Plume

(TSM)

Bay Of Bengal

(TSM)

Elbe Mouth,

Germany

(CHL)

Automation of image processing [Vanhellemont, in prep]

BELCOLOUR-ARG

postdoc [Dogliotti et al]

1. Introduction


3. Science Results

4. Conclusions


[27/28]



Web site publications and products

1. Introduction


3. Science Results

4. Conclusions


[28/28]



Conclusions

BELCOLOUR-2 is the research behind current/future applications of

marine optical remote sensing products (CHLa, TSM, etc.)

Project team is progressing in various areas of aquatic optics

research (absorption/scattering, adjacency effects, bottom effects,

exploitation/design of geostationary sensors, plankton species

detection, pCO2 estimation, Primary Production, data filling/QC, etc.)

Publications (includes RECOLOUR spin-off):

– 19 peer-reviewed accepted/published (inc. 5 external collaborations)

– 3 peer-review submitted

– 7 peer-review drafted

– 13 other publications, conference proceedings

– ~10 international conference presentations/year

Training:

– 1 Masters course (RUG/VUB), 6 Masters theses, 2 Phd theses

1. Introduction


3. Science Results

4. Conclusions

BELCOLOUR has established expertise and

strong internal and external collaborations

• Will future be fragmented?

• What happens to the researchers?

Our heritage


[29/28]



Reserve slides


[30/28]



Applications

Research leads to use of optical remote sensing for:

– Marine science support (e.g. NRT for cruise ships,

multitemporal composites, time series extraction via web site)

– Ecosystem functioning, aquaculture and fisheries (e.g. primary

production maps, algae bloom timing, monthly CHL)

– Eutrophication (OSPAR/WFD) via forcing/validation of

ecosystem modelling (AMORE-3) and optimising sampling

– Sediment transport support via initialisation/validation of

models

– HAB group support (Marcoast, ICES, AMORE-3)

– pCO2 and CO2 fluxes

– Inland waters

1. Introduction

2. Science Results


4. Dissemination

5. Plans for 2011


[31/28]


Atmospheric correction inter-comparison

Validation completed incorporating MUMM's reflectance data base until end of 2008

Five algorithms compared (MUMM, NIR-6.1, SWIR-6.1, MOD09, ANN)

SeaDAS 6.1 standard atmospheric correction significantly improved compared to v5.4 (major changes introduced with v6.0)

Results indicate overall best performance of ANN algorithm at all time differences ±30, ±60, ± 120, ±180 min to satellite over pass

Peer reviewed publication in preparation (Optics Express)

[Schroeder, in prep]


[32/28]



Project structure e.g. 1/2: OSPAR,

Eutrophication, HAB

7. APPLICATIONS EXTERNAL

PROJECTS/USERS

4. IMAGERY

2. THEORY


RECOLOUR

OSPAR/EU

Multitemp Chl

P90+mean

Assess

Report

Validation

Operational

Future Phaeocystis

detection

AMORE

modelling

Geostationary

Space Agencies

1. Introduction

2. Science Results


4. Dissemination

5. Plans for 2011


[33/28]



7. APPLICATIONS

4. IMAGERY

2. THEORY


RECOLOUR

7. APPLICATIONS

4. IMAGERY

2. THEORY


Project structure e.g. 2/2 Inland Waters

EXTERNAL

PROJECTS/USERS

Adjacency

Bottom effects

Turbidity

Airborne/UAV

Hi res sat

EU-WFD

NIR Sim.

Spectrum

1. Introduction

2. Science Results


4. Dissemination

5. Plans for 2011


[34/28]



Spectral variations of scattering: Do phytoplankton blooms play a role?

[Astoreca et al.]

All data

-0.1< < -0.9

: 0.1-0.7

y = -4.98x + 0.65

R2 = 0.72

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

0.00 0.10 0.20 0.30 0.40

ap/cp(440)

(412-7

15)

<0 for

a/c>0.14

Ac9 particle scattering


[35/28]



0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

4:19 6:43 9:07 11:31 13:55 16:19

Time

Ju

ng

e e

xp

on

en

t an

d c

p s

pectr

al

slo

pe

junge-3 gamma_bp_nir

Tidal variations over mudbanks

(Belgian coast – July 2007)

Particulate attenuation spectral slope vs. Particle Size

Distribution (Junge exponent – 3)

Tidal variations at Scheldt river mouth

(Belgian coast – September 2009) data in process

[Doxaran et al]

Large

particles


[36/28]



0

0.5

1

1.5

2

600 700 800 900

wavelengths (nm)

AT:

AT(7

80

nm

) AT

0

0.5

1

1.5

2

600 700 800 900

wavelengths (nm)

aw [Kou et al1993]

aw [Pope & Fry 1997]

aw [Buiteveld 1994]

Generic one-band TSM and turbidity algos (hyperspectral

calibration)

T-calibrated algorithm:

npwTT

bTw

*

AT B

aρ= + =

1 -ρ γb/A

C

Laboratory pure water absorption measurements

AS

0

0.5

1

1.5

2

600 700 800 900

wavelengths (nm)

[Nechad et al, 2009]


[37/28]



Comparison of observations from a totally independent data-set (Luctor) at

the mouth of the Scheldt and values computed with the algorithm

(pCO2@10°C = f(SSS; Chla) based on the 2008 Belcolour data-set shows the

ability to reconstitute a full annual cycle of pCO2. The two outsiders are due to

SSS value outside range of algorithm application, and suspicious input value

(Chla value)

[Borges et al]


[38/28]


Oudenburg, 25th May 2011 Brussels, 28-29 December 2009

Atmospheric correction

inter-comparison

29.07.2002

05.08.2003

(MAPE=Mean Absolute Percentage Error)

North Sea

North Sea

[Schroeder et al]


[39/28]



TSM mapping and uncertainty (SEVIRI)

June 29th 2006 at 13:00 UTC

2)6.0(

)6.0(

)(w

w

C

ACTSM

TSM (mg/l) TSM (mg/l) [Neukermans et al, 2009]


[40/28]



App #5: Aquaculture support

CHL, TSM and SST time series extracted from BC archive for

ILVO (mussel production in Belgian waters)

2008 original

2010 updated

1. Introduction

2. Science Results


4. Dissemination

5. Plans for 2011


[41/28]



• Particle size distributions (PSDs) observed in

the Southern North Sea show a power-law „flat‟

distribution in the TCZ and SCZ while a bimodal

distribution with peaks at 7 and 200 m is

observed in the MSNS.

• Mie computations were used to explain the

occurrence of the cp spectral negative slopes.

The observed PSDs were used as inputs.

• Results show that cp follows a power-law shape

with positive cp in the SCZ and TCZ.

• a negative cp is obtained for the MSNS with

absorbing particles

• when modifying the PSD of the MSNS by

removing the peak at 7 m, the result is a

positive cp as in the TCZ.

• when removing the peak at 200 m from the

MSNS PSD, the result is a negative cp

=> the negative cp results from the bloom of

phytoplankton particles of size 7 m.

1.00E+012

1.00E+013

1.00E+014

1.00E+015

1.00E+016

1.00E+017

1.00E+018

1.00E+019

1.00E+020

1.00E+021

1.00E+022

1.00E+023

1.00E+024

0 50 100 150 200 250

Particle size ( m)

Nu

mb

er

of

part

icle

s

SCZ_700

MSNS_W09

TCZ_TH01

MSNS_W09_low bump

MSNS_W09_highbump

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

1.6

400 500 600 700

Wavelength (nm)

cp

no

rmali

zed

at

715 n

m

[Astoreca et al, submitted]


[42/28]



1. Introduction

2. Science Results


4. Dissemination

5. Plans for 2011

Rrs (sr-1) at bands 665 and 443 nm vs TSM (g m-3),

generated from Hydrolight for aNAP*=0.04 m2g-1,

SNAP=0.012 nm-1, bbp*=0.012 m2g-1 and γ=0.5, CHL=1 mg

m-3. The sun zenith angle is 0°

High Rrs (443nm) sensitivity up to 20 g m-3

High Rrs (665nm) sensitivity up to 46 g m-3

Wavelengths

(nm)

TSM ranges

(g m-3)

412-560nm 0-10

620-665nm 10-30

682-705nm 30-100

775nm 100-500

865-900nm 500-1500

Is there a general algorithm for TSM estimation from satellites?

Hydrolight simulations

Non linear regression (NLR)

analysis to fit (Rrs//TSM)SIOP,GEO

(1) Rrs = G TSM / (A – TSM/C)

using (1), parameterise:

Saturation

Sensitivity

Uncertainties due to SIOPs

and sun/viewing geometry

[Nechad, in prep]


[43/28]



ECOLOGIE DES SYSTEMES AQUATIQUES

Development and validation of a primary production

algorithm for coastal waters of the Southern North Sea

[Rousseau et al, in prep]

dzdtePChlaPPdz

z

P

Esunsett

sunriset

BB

ztB

*)(1**0

)*(

maxmax

,

103,2][*037,4*463,0 4max POTPB

zPARK

zdeEE

)*(

0 *

BB Pmax*003,0017,0

995,19)ln(*191,6])ln([*157,0])ln([*454,0))(ln( salChlaMESPARKd

http://www.ulb.ac.be/contacts.html


[44/28]



ECOLOGIE DES SYSTEMES AQUATIQUES

0

0,5

1

1,5

2

2,5

3

3,5

4

0 0,5 1 1,5 2 2,5 3 3,5 4

PP

pré

dit

e (

gC

./m

².j)

PP in situ (gC/m².j)

Average Min Max

51% 7% 220%

PP algo performance [Rousseau et al, in prep]

• Regional algorithm based on specific data

• Hyp: PBmax, αB and Kd(PAR) are constant along the day; Salinity > 31

• Good predictive models for PBmax, αB and Kd(PAR) Good for PP

But need to improve the description of PBmax

• GPP vs NPP : respiration taken into account : coupling with 3D-MIRO model

• Climatology vs In situ measurements of Et (satellite) and z (3D-model)

Relative errors

1. Introduction

2. Science Results


4. Dissemination

5. Plans for 2011

http://www.ulb.ac.be/contacts.html


[45/28]



1. Introduction

2. Science Results


4. Dissemination

5. Plans for 2011

App #8: La Plata: TSM and Turbidity maps

(BELCOLOUR-ARG) A band-difference TSM and turbidity algorithm was developed based on

Nechad et al. (2009/10) one-band algorithm.

Good results were obtained for low Turbidity (<100 NFU)

ρrc

(859)

Future perspectives:

- Improve atmospheric

correction algorithm to

apply one-band TSM

algorithms

- Start collecting in situ

radiometric measurements

[Dogliotti, in prep]


[46/28]



Reorganisation of image archive (in progress)

BELCOLOUR image archive was just graphics images ...

reorganising to give time series, digital data downloads

[J-R. Diouf]


[47/28]



Ed Lu,

Lsky

MUMM

abovewater

system

JRC underwater

system

Water-leaving radiance protocol

intercomparison on AAOT [Zibordi et al, in prep]

Also VITO/MUMM comparison (Sint Anna pontoon)

1. Introduction

2. Science Results


4. Dissemination

5. Plans for 2011


[48/28]



Automation of image processing (1/2)

Flexible software to process satellite imagery

– New region: 4 bounding coordinates

– MODIS/SeaWiFS

– MERIS RR and FR

– Custom products easily implemented (TSM, Kd)

Outputs

– gridded scenes (NetCDF/png)

– daily and multitemporal binning

– time-series

Fast

– processing 8 years of MODIS data for SNS: 20 minutes

– extraction of time-series, mean maps: matter of seconds

[Vanhellemont, in prep]


[49/28]



0

1

2

3

4

5

6

400 450 500 550 600 650 700 750 800 850

Dep

th (

m)

wavelength (nm)

____ 5% signal level- - - - 10% signal level

'average' water quality

'turbid' water

'clear' water

WQ based on CASI image

WQ based on in situ Rw measurements Ecolight simulations: look-up-table

Effect of bottom substrate on water leaving reflectance and water quality retrieval Tote Carolien, Sterckx Sindy, Knaeps Els, Raymaekers Dries

0 5 10 15 20 25

field concentration (µg/L)

0

5

10

15

20

25

be

st fit con

centr

ation

(µ

g/L

)

0 4 8 12 16 20

field concentration (mg/L)

0

4

8

12

16

20

best fit concentr

ation (

mg/L

)

600 800 1000

wavelength (nm)

0

0.004

0.008

0.012

0.016

Rrs

in situ

LUT best fit

0 5 10 15 20 25

field concentration (µg/L)

0

5

10

15

20

25

CA

SI derive

d c

oncentr

ation (

µg/L

)

0 4 8 12 16 20

field concentration (mg/L)

0

4

8

12

16

20

CA

SI

de

rive

d c

on

ce

ntr

atio

n (

mg

/L)

Optical Remote Sensing of marine and inland waters · 2018-09-17 · 2. Laboratoire d’Ecologie...

Documents

Transcript of Optical Remote Sensing of marine and inland waters · 2018-09-17 · 2. Laboratoire d’Ecologie...