Nos tutelles

Centre National de la Recherche Scientifique


Accueil > Productions > Téléchargements


par Dessailly David - publié le , mis à jour le

Contact : Séverine Alvain


PHYSAT Look-Up Table :
Mean SeaWiFS and MODIS values to process nLw* (Alvain et Al. 2005 & 2008)

PHYSAT demonstration tool

PHYSAT products download

To access PHYSATdownload directory you’ll need a login and password, please send and e-mail to to receive this authentication information.

Then you have access to
PHYSAT archive (from SeaWifS Level 3 products / NASA/GSFC/DAAC) and relative maps.
If used the physat data or Ra based labellisation approach in a scientific article, please cite PHYSAT with its Inter Deposit Digital Number (License APP) : IDDN.FR.001.330003.000.S.P.2012.000.30300.).

Each file cover the complete SeaWiFS archive (1997-2010), these are NetCDF file with complete data description.

PHYSAT v2013
Map of dominant groups :
File : PHYSAT_GlobalMapOfMonthlyDominantsGroups_1997-2010.tar.gz

Map of the most frequently dominant phytoplankton groups detected over each months. See Alvain et al. 2005 (DSR) and 2008 (GBC) and Ben Mustapha et al. 2013 (RSE) for more details.
Information : NetCDF files with 2160x4320 tables in integer / 1= Nanoeucaryotes, 2= Prochloroccocus, 3= Synechococcus / 4= diatoms / 5 = Phaeocystis-like / 6= coccolithophorids bloom (warning : underestimation due to SeaWiFS thresholds, please see Alvain et al. 2008).

Map of phytoplankton groups frequencies :
File : PHYSAT_GlobalMapOfMonthlyGroupsFrequency_1997-2010.tar.gz

Map of detection frequencies over each month, 0 = group never detected, 1 = all valid pixels were associated with the group.
Information : NetCDF files with 2160x4320 tables in float 32

Monthly Chlorophyll-a concentration from OC-SD algorithm :
File : PHYSAT_GlobalMapOfMonthly_OC-SD_CHloro_a_1997-2010.tar.gz
(with an improvement in function of the PHYSAT groups), please see Alvain et al. 2006 DSR for more information.
Monthly Chlorophyll-a concentration maps / NetCDF files with 2160x4320 tables in float 32.

PHYSAT v2008
You can also download the "old fashion" version of PHYSAT processing (without the Self Organising Map classification of spectral anomalies introduce by Mustapha et al. 2013 (RSE))
File : PHYSAT_GlobalMapOfMonthlyDominantsGroups_1997-2010_v2008.tar.gz
File : PHYSAT_GlobalMapOfMonthlyGroupsFrequency_1997-2010_v2008.tar.gz


In the past, remote sensing has been used to retrieve the signature of photosynthesis in the ocean. In 2005, a new method has been published : PHYSAT. This methodology makes it possible to use ocean color signals to determine the dominant phytoplankton groups in the surface waters.

Phytoplankton and oceanic biological pump

Anthropogenic carbon in the atmosphere is the main cause of present climate change. However, differents phytoplantkon groups [1] have different impact on the carbon cycles as well as for some other biogeochemical elements in the ocean. Understanding how these pump will evolve is a key to working out climate change scenarios for the coming decades.

Phytoplankton and biogeochemical cycles

Phytoplankton make photosynthesis, and for this, they have pigments with which they capture the energy of photons. First among these pigments is chlorophyll a, which absorbs light, especially in the blue part of the visible spectrum. This modifies the flux of photons that penetrate the surface ocean layer, in such way that the light that leaves the ocean is less blue than penetrating light. For more than 20 years now, space agencies have exploited this property and have launched satellites with ocean color sensors to estimate the surface chlorophyll a concentration and its spatio temporal variability.
Note that in the visible part of the spectrum, the atmosphere is responsible for about 95% of the blue light detected by a satellite sensor. However various methods have been established in order to deconvoluted the part of the atmosphere from the ocean one. This was demonstrated after the launch of the CZCS sensor by NASA in 1978. The blue/green ration of the ocean reflectances part was then used to work out the first global maps of marine biological activity over a seasonal cycle. Sea color data are now globally available in near real time after the launch of several sensor OCTS (NASDA), POLDER (CNES) in 1996, SeaWiFS, MODIS-T and A (NASA), in 1997, 1999 and 2002 respectively, and MERIS (ESA) in, 2001.

The first use of the chlorophyll a concentration was to estimate marine photosynthesis, or primary production, which is the first step in the biogeochemical reactions that constitute the ocean carbon cycle.

Chlorophyll concentration in the ocean is the main cause of sea color variability, however, exceptions have been recognized in certain situations. The need to know which ecosystem is active at a given place and a given time, and the possibility that some part of the variability of sea color might be explained by something other than chlorophyll concentration have generated research that has recently made it possible to identify the signatures of some phytoplankton groups in the satellite ocean color data. It is this method, named PHYSAT, that is presented hereafter.

Identification of phytoplankton groups based on satellite detected ocean color

The various phytoplantkon groups have different proportions of accessory pigments (I.e. pigments other than chlorophyll) resulting in slightly different absorption spectra. Identifying groups in ocean color data has long been considered, if not theoretically impossible, then at least very difficult. However, the need to access this knowledge in order to better understand the functioning of the biological pump and its evolution in the context of climate change is such important that we decided to tackle this problem empirically.

The first step : the long preliminary task of collecting in situ data

The first step was to collect the pertinent data for this purpose : in situ measurements with biological information rich enough to indicate the dominant groups, together with ocean color measurements made by satellite at the same time and the same place. All of this should be representative of a wide range of climatic conditions.
Such data already exist but, access to them was restricted or they have been collected mostly during oceanographic cruises and so they over represent small parts of the ocean during limited periods. By contrasts, sampling along a commercial sea route seemed an appropriate solution for building a dataset representative of the world ocean (from IRD studies). A German shipping company agreed to host an oceanographer on board one of its ships, the “Contship London”, on each of its voyages from Le Havre to Noumea, in order to collect and process the needed water samples. This idea is at the origin of the Gep&Co project coordinated by the French national program PROOF and Yves Dandonneau. This long sea route crosses a large variety of water types :

- The North Atlantic spring bloom
- East coast of North America
- The Caribbean Sea and Gulf of Panama
- Equatorial Pacific with abundant nutrient but no iron
- South pacific gyre where plankton is rare
- Temperate waters near New Zealand

Routine measurements of about 20 pigments during each voyages enabled certain phytoplankton groups, with known pigment signatures, to be identified. GeP&CO consisted of 12 campaigns from November 1999 to August 2002.

Finally, only one tenth of the 1400 GeP&CO observations were found to be colocated with SeaWiFS reflectance measurements, mainly because of cloud cover. This underlines the need to collect field data with global and seasonal coverage, like the GeP&CO dataset.

Ocean color data ; the decisive choice : filtering out the effect of chlorophyll concentration

Linking marine reflectances to groups of phytoplantkon identified according to the relative abundance of certain diagnostic pigments was the objective of my phd work. I started this in october 2002 at the LSCE laboratory in collaboration with LOCEAN. My phd advisors were Cyril Moulin and F.M. Breon. With a very strong collaboration with Yves Dandonneau.

My first task was to filter the signature of the chlorophyll a concentration out of the satellite detected marine reflectance spectra. As state above, these spectra respond strongly to chlorophyll a concentration, and the first order variations were like to dominate any second order signal produced by specific phytoplankton populations. For this purpose, I have established an average statistical model of marine reflectances as a function of chlorophyll a concentration, using a large number of SeaWiFS chlorophyll and reflectance data sorted by small range of chlorophyll concentration from 0.04 to 4 mg.m-3. Normalizing the measured reflectances versus this model makes it possible to work out sea color anomalies, not affected by chlorophyll a concentration.

The first results....

Comparison of these reflectance anomalies with the GeP&CO pigments data at the 140 co-located observations revealed a remarkable organization :
The lowest negative spectral anomalies correspond to abundant 19’hexanoyloxyfucoxanthin, a pigment found in nanoeucaryotes. Some waters showed relatively abundant divinyl-chlorophyll a, which exists int the Prochlorococcus. Other observations with slightly positive anomalies have a high zeaxanthine to divinyl-chlorophyll a ratio. Thus this category differs by the abundance of Synechococcus like cyanobacteria. Finally the highest positive spectra anomalies were found for observations with especially high fucoxanthin content, pigment characteristic of diatoms.

Since the above observation requires only information that can be provided by satellite (marine reflectance spectra and chlorophyll a concentration), it can be applied to any satellite ocean color measurement. I thus fixed a chain of data analysis, named PHYSAT, that aims to identify which kind of phytoplankton dominates in a given pixel, and to draw up global maps of the distribution of the identified groups. For the first time, a global view of phytoplantkon groups in the ocean
PHYSAT maps results are in good agreement with previous knowledge of the distribution of phytoplankton groups . Such maps provide unequaled validation datasets for biogeochemical models of the global ocean. A seasonal distribution and succession of dominant phytoplantkon groups observation in the global ocean from more than 10 years of SeaWiFS data has been publishedin 2008 (Alvain et al., 2008).

Monthly climatology (1998-2007) of the dominant phytoplankton groups

This new operational product will also offer tool to observe the influence of climate on marine ecosystems. Thus, exceptionally abundant chlorophyll in the equatorial Pacific during the climate anomaly “La Nina” in Auguste 1998 was caused by a diatoms bloom according to PHYSAT, a result that is all the more striking since this is a region where diatoms are generally scarce.

All of these are pioneering results that open new doors for oceanographic research : from now on, it’s possible to study the variations of the phytoplankton groups at global or regional scale, and their forcing by mesoscale ocean circulation, using only satellite ocean color data. However, the optical mechanisms that produce their specific spectral signatures are still not fully understood as the differences in the pigment composition of detected groups do not explain them.

Main Publications about PHYSAT :

Alvain S., Moulin C., Dandonneau Y., and H. Loisel, Seasonal distribution and succession of fominant phytoplankton groups in the global ocean : A satellite view., Global Biogeochem. Cycles, 22, 2008.

Alvain S., Moulin C., Dandonneau Y. , Loisel H. and Breon FM., A species-dependent bio-optical model of case I waters for global ocean color processing. Deep Sea Res. I, 53, 917-925, (2006).

Bopp L., Aumont O. , Cadule P., Alvain S. and Gehlen M., Response of diatoms distribution to global warming and potential implications : A global model study, Geophysical Research Letters, 32 (19), 1-4, (2005).

Alvain. S., Moulin C., Dandonneau Y. and Bréon F.M., Remote sensing of phytoplankton groups in case 1 waters from global SeaWiFS imagery, Deep Sea Res. I, 52, 1989-2004, (2005).

[1For more information about the ’groups’ definition used here, please see Alvain et al. 2005 and Le Quere et al. 2005.