Chlorophyll_a algorithms for MODIS and MERIS

full resolution imagery: an analysis of Ligurian

and North Tyrrhenian waters

Chiara Lapucci

Gohin F., Ampolo Rella M., Brandini C., Gozzini B., Maselli F., Massi L., Nuccio C.,

Ortolani A., and Trees C.

MELBA MILONGA

ENVIRONMENTAL CHARACTERIZATION

Chlorophyll_a is a key element to assess the sea ecological status

(WFD, MS descriptor 5 - eutrophication )

http://www.mo-mar.net/

OCEANOGRAPHIC CAMPAIGNS

NURC

Ifremer

University of Florence Ecologia Vegetale

True Color MODIS 17 March 2009

In situ HPLC Chlorophyll_a data:

Campaign Period

Number

of

stations (matchup)

[Chl_a]

Average

(mg m-3)

[Chl_a]

Standard

deviation

(mg m-3)

[Chl_a]

minimum

(mg m-3)

[Chl_a]

maximum

(mg m-3)

BP09 14 – 23

March 2009 33 1.011 0.712 0.396 2.837

MOMAR 19 – 22

July 2010 9 0.126 0.076 0.045 0.311

REP10

19 August -

3

September

2010

21 0.298 0.536 0.050 1.899

Main features and relevant [CHL] statistics of the three oceanographic

campaigns

MODIS AQUA – TERRA NASA MODerate Resolution

Imaging Spectro-radiometer

Spatial resolution: 1km

(FUTURE: VIIRS NPP)

MERIS Envisat ESA Medium Resolution Imaging Spectro-Radiometer

Full resolution: 0.3m

(FUTURE: OLCI Sentinel 3)

ENVISAT

European Space Agency

EOS Terra

Chlorophyll

CDOM

SPM ALGORITHMS

SATELLITES

Case 1 Case 2

OC3M

Bio optical empirical MODIS global algorithm, Case 1 waters. O'Reilly, J.E., and 24 Coauthors, 2000: SeaWiFS Postlaunch Calibration and Validation Analyses, Part 3. NASA Tech. Memo.

2000-206892, Vol. 11, S.B. Hooker e E.R. Firestone, Ed. NASA Goddard Space Flight Center, 49 pp.

MedOC3

Bio optical empirical MODIS algorithm, OC3 regionally adapted on North Western

Mediterranean, Case 1 waters. Santoleri R., Volpe G., Marullo S., Buongiorno Nardelli B., "Open waters optical remote sensing of the Mediterranean Seas“,

Remote sensing of the European Seas, 103-116, Springer Netherlands, 2008.

OC5

Bio optical empirical MODIS (MERIS and SeaWIFS) algorithm, suited for Biscay Bay and the

English Channel, Case 1 and Case 2 waters. Gohin F., Druon J. N., Lampert L., A five channel chlorophyll concentration algorithm applied to SeaWiFS data processed by

SeaDAS in coastal watersInt. J. Remote Sensing, 2002, vol. 23, no. 8, 1639–1661.

SAM-LT

Bio optical semi – analytical MODIS algorithm, locally tuned for Tyrrhenian - Ligurian sea,

Case 2 waters. Maselli F., Massi L., Pieri M., and Santini C., Spectral Angle Minimization for the Retrieval of Optically Active Seawater

Constituents from MODIS Data Photogrammetric Engineering & Remote Sensing, Vol. 75, No. 5, May 2009, pp. 595–605.

ALGORITHMS

OC5 LUT taking into consideration MBR, corrects for

CDOM (Rrs412) and SPM (Rrs555).

SAM_LT inversion algorithm which identifies simulated Rrs

which best corresponds to measured spectral data.

EMPIRICAL ALGORITHMS

are based on direct regression of the ratio of reflectances at two

wavelengths - BLUE/GREEN - to chlorophyll concentration. The remote-

sensing reflectance can be assumed as proportional to bb/a; so if the band

ratio is taken, the influence of scattering is weakened.

OC3M, MedOC3 MBR=(Rrs443>Rrs488)/Rrs547

SEMI ANALYTICAL ALGORITHMS

Radiative transfer theory provides a relationship between upweling

radiance and the inherent optical properties of water (Sathyerdranath et al.,

1997). Rrs is function of the inherent optical properties of sea water.

Rrs=βbb(λ )/[a(λ)+bb(λ)]

17 March 2009

MODIS AQUA Chlorophyll_a mg/m3

MODIS AQUA

OC3M

0,01

0,1

1

10

0,01 0,1 1 10Measured [CHL] (mg m-3)

Esti

mat

ed

[C

HL]

(m

g m

-3)

BP09

MOMAR

Rep10

R2 = 0.7671

RMSE = 1.22 mg/m3

MBE = 0.36 mg/m3

MedOC3

0,01

0,1

1

10

0,01 0,1 1 10Measured [CHL] (mg m-3)

Esti

mat

ed

[C

HL]

(m

g m

-3)

BP09

MOMAR

Rep10

R2 = 0.795

RMSE = 2.91 mg/m3

MBE = 0.91 mg/m3

SAM_LT

0,01

0,10

1,00

10,00

0,01 0,1 1 10Measured [CHL] (mg m-3)

Esti

mat

ed

[C

HL]

(m

g m

-3)

BP09

MOMAR

Rep10

R2 = 0.4333

RMSE = 0.75 mg/m3 MBE = -0.22

63 matchups

OC5

0,01

0,1

1

10

0,01 0,1 1 10Measured [CHL] (mg m-3)

Esti

mat

ed

[C

HL]

(m

g m

-3)

BP09MOMARRep10

R2 = 0.7414

RMSE = 0.74 mg/m3

MBE = 0.08 mg/m3

Lee and Hu (2006) Case 1

0,00

0,10

0,20

0,30

0,40

0,50

0,60

0,70

HPLC OC3M MedOC3 OC5 SAM

[CH

L] (

mg

m-3

)

Lee and Hu (2006) Case 2

0,00

0,50

1,00

1,50

2,00

2,50

3,00

3,50

HPLC OC3M MedOC3 OC5 SAM

[CH

L] (

mg

m-3

)

Lee Z P, Hu C (2006) Global distribution of Case-1 waters: an analysis from SeaWiFS

measurements. Remote Sensing of Environment 101 270 – 276.

Optical relations typical of Case 1

waters

Chl_a - CDOM

Chla_a - SPM

15 Case 1

31 Case 2

MERIS FR and MODIS AQUA Chlorophyll_a mg/m3

17 March 2009

22 matchups

MODIS AQUA vs MERIS OC5

0,01

0,10

1,00

10,00

0,01 0,10 1,00 10,00

MERIS [CHL] (mg m-3

)

MO

DIS

[C

HL

] (m

g m

-3)

R2 = 0.7708

22 matchups

MERIS FR MODIS

OC5

0,01

0,1

1

10

0,01 0,1 1 10Measured [CHL] (mg m-3)

Esti

mat

ed

[C

HL]

(m

g m

-3)

BP09

MOMAR

Rep10

R2 = 0.5826

RMSE = 1.23 mg/m3

MBE = 0.32 mg/m3

OC5

0,01

0,1

1

10

0,01 0,1 1 10Measured [CHL] (mg m-3)

Esti

mat

ed [C

HL]

(mg

m-3

)

BP09MOMARRep10

R2 = 0.7639

RMSE = 0.95 mg/m3

MBE = 0.23 mg/m3

MERIS OC5

0,01

0,1

1

10

0,01 0,1 1 10Measured [CHL] (mg m-3)

Est

ima

ted

[C

HL]

(m

g m

-3)

BP09

MOMAR

Rep10

R2 = 0.7695

RMSE = 0.50 mg/m3

MBE = -0.10 mg/m3

Arno river plume strong

Chl_a oversetimation by

OC5

MODIS OC5

0,01

0,1

1

10

0,01 0,1 1 10Measured [CHL] (mg m-3)

Esti

mat

ed

[C

HL]

(m

g m

-3)

BP09MOMARRep10

R2 = 0.7869

RMSE = 0.38 mg/m3

MBE = -0.09 mg/m3if we get rid of those stations

OC5 MODIS chlorophyll_a

8-day average

http://www.lamma.rete.toscana.it/en/node/3431

• The current analysis confirms the overestimation of “conventional” MODIS algorithms (OC3M and MedOC3)

• SAM_LT is poorly sensitive to low chlorophyll concentrations, while it performs better than the others on critical pixels, such as on river plumes

• OC5 overall shows the better performances, but has a tendency to

the overestimation

• A larger dataset of in situ and satellite observations is needed.

Conclusions

lapucci@lamma.rete.toscana.it

Thank you