AMMA (African Monsoon Multidisciplinary Analyses) aims at providing the African decision makers with improved assessments of rainfall changes which are likely to occur during the 21st century due to natural fluctuations and as a result of anticipated global climate change. An essential step in that direction is to improve the ability to forecast the weather and climate in the West African region.

AMMA is funded by EU/FP6 (AMMA-EU), and by national institutions (AMMA-France, AMMA-Africa, AMMA-UnitedKingdom, AMMA-UnitedState). POSTEL is involved in AMMA-EU and AMMA-France as provider of biogeophysical variables derived from satellite data useful to characterize continental biomes in Soil-Vegetation-Atmosphere transfer schemes.

Practically, POSTEL is involved in many actions :

• the customisation and distribution to the AMMA community of Land SAF products provided by the Institute of Meteorology of Portugal.
  • The customisation of radiative variables (downwelling shortwave and longwave radiation fluxes, Land Surface Temperature, albedo) and their upload to the AMMASAT database are carried out in real time.
  • These products are also provided to CNRM which uses them as atmospheric forcing variables as inputs of Land Surface Models which simulate land-atmosphere and hydrological processes over West Africa in the framework of the AMMA Land surface Model Intercomparison Project ALMIP . Recents results (Boone and de Rosnay, 2007) show that using satellite-based products combined with NWP data as atmospheric forcings improves the simulations of the West African monsoon (Figures below). These products are also used as atmospheric forcing for meso-scale simulations studies performed by many AMMA teams (e.g. Guichard et al., 2007).

Evapotranspiration simulated by 2 Land Surface Models (ISBA from CNRM, ORCHIDEE from LMD) averaged over the monsoon season (June-September) 2004, using satellite-based products combined with NWP data as atmospheric forcings (Figures a and b). Figures c and d show the differences between the aforementioned simulations and control runs using NWP forcings only. The increased evapotranspiration in the Northern part of the area is associated with errors in the NWP data which displace the monsoon too far South compared to satellite-based products (from Boone and de Rosnay, 2007).

• the production, in real time, of a daily NDVI, corrected for directional effects.
  • This product has been required by AMMA users. It is generated from Land SAF albedos. It is also available through the AMMASAT database.

• the geometric and atmospheric corrections of high resolution (20m and 2.5m) SPOT images acquired in 2005 and 2006 over regional sites (Benin, Mali, Niger) over the AMMA study area.
  • This action has been initiated by CETP with the goal to supply the user community with standardized data, in terms of format, and geometric and radiometric processing. The POSTEL Service Center managed this action, with strong links with various team (SI/MO, PS/TIS, SI/EI) of Cnes, the CS-SI company, and many people from IRD, and CESBIO in Benin, Mali, and Niger. The use of the TARIFA processing line, developed by the team PS/TIS of Cnes, to perform the geometric corrections, allowed to get results of a very good accuracy (+/- 0.2 pixel over Bénin and Mali, +/- 0.1 pixel over Niger (Figure below)). All details of this work are available in a final report (in French only). The 50 geo-referenced, superimposable, calibrated images (11 over Benin, 15 over Mali, 24 over Niger) corrected for atmospheric effects are accessible through the AMMASAT database.

Color composite of 3 SPOT images (XS2 band) acquired at 3 different dates (June 27, September 12, October 25, 2005) over Niger shows the good quality of geometric corrections.

• AMMA teams use these SPOT time series for various applications related to hydrology, change of land use, health, etc... For instance, CETP has generated a land cover map over Niger (below), a map of water bodies, and dynamic leaf area index maps that are used as input of the SEtHyS_savannah surface model (Saux-Picart, et al., 2007a, Saux-Picart, et al., 2007b) which simulates the heat and water exchanges in sahelian landscapes. At the same time, LTHE and CESBIO have generated a land cover map over Benin, and Mali, respectively. They are currently under validation with ground truth data. All classification works performed over the three AMMA super-sites will be presented in a publication under writing (Zin et al., 2008). Furthermore, the land cover map over Niger has been included in a large study focusing on the land cover evolution over Niger supersite during the last 20 years (Ottlé et al., 2007).

Land cover map over Niger from SPOT images at 20m resolution (from CETP)

• the re-processing of AVHRR archive over the AMMA study area.
  • This actions aims at getting homogeneous time series of reflectance and NDVI over 20 years derived from AVHRR data to analyse the spatio-temporal variability of land surface conditions over West Africa. POSTEL, with the CESBIO and the team SI/MO of Cnes as Expertise Centres, is identified as beta-user of calibrated and geolocated AVHRR data, generated by NASA in the frame of LTDR (Long Terme Data Record) project (examples below). First, the work consists in checking the calibration over desertic sites, and the consistency between successive sensors. A preliminary work on this last topic has been achieved : the report is available. Then, the directional effects are corrected using the processing line of the CYCLOPES project. Finally, the CESBIO will analyse the normalized time series fo reflectances and NDVI, especially by comparison with ground data, before using them as input of a surface model to simulate the heat exchanges at the soil/vegetation/atmosphere interface.

Time profiles of AVHRR reflectances (visible : red ; near-infrared : yellow) and NDVI (bleue) from 1982 to 2000 over 3 sites in west Africa : Agoufou in Mali (top), Banizoumbou in Niger (centre), and Djougou in Benin (bottom) (from NASA).

• the customisation of existing products (LAI, FAPAR, FCover, NDVI, albedo) derived from AVHRR, MODIS, and POLDER sensors.
  • It consists in re-sampling of products on pre-defined grids, then extracting areas of interest. The objective is to standardize products in terms of projection, spatial resolution, and coverage area, so that they are easier to use by modellers. All these products are uploaded to the AMMASAT database.

Boone, A. and P. de Rosnay, Towards the improved understanding of land-surface processes and coupling with the atmosphere over West Africa. iLEAPS Newletter, 3, 33-34, 2007.

Guichard, F., C. peugeot, O. Bock, J.L. redelsperger, N. Asencio, A. Boone, F. Chopin, X. cui, M.A. Gaertner, M. Garvert, S. Jones, J.P. Lafore, B. Lamptey, K. ramage, E. Orlandi, J. Sanders, Mesoscale simulation of MCSs : intercomparison of precipitation and surface evapotranspiration fields for the 28-29 August 2005 case study, 2nd AMMA Conference, Karlsrhue, 26-30 November 2007.

Ottlé, C., C. André, M. Zribi, S. Saux-Picart, R. Lacaze, N. Boulain, B. Cappelaere, J.L. Rajot, L. Descroix, Land cover evolution over the Niger Supersite between 1986 nd 2005 from SPOT-HRV data, 2nd International AMMA Conference, Karlsruhe, 26-30 November 2007.

Saux-Picart, S., Ottlé, C., Perrier, A., Decharme, B., Coudert, B., Zribi, M.,Cappelaere, B., and N. Boulain, Spatialisation of the SEtHyS_Savannah SVAT model using remote sensing data over the AMMA-Niger supersite, 2nd International AMMA Conference, Karlsrhue, 26-30 November 2007a

Saux-Picart, S., Ottlé, C., Perrier, A., Decharme, B., Coudert, B., Zribi, M., Boulain, N., Cappelaere, B., and D. Ramier, SEtHyS_savannah : a multiple source land surfce model applied to sahelian landscapes, submitted to Agricultural and Forest Meteorology, 2007a.

Zin, I., M. Zribi, C. André, N. Dessay, S. Guibert, P. Hiernaux, L. Kergoat, R. Lacaze, S. Mascle-Lehégarat, C. Ottlé, S. Saux-Picart, J. Seghieri, Land cover/Land use assessment on the three AMMA super-sites from SPOT/HRV data, to be submitted to AMMA special issue of Hydrological Science Journal, 2008.