AFFILIATION : Mohammed VI Polytechnic University - Morocco
THEME : Different approaches of using remote sensing to estimate water consumption for agriculture
SUMMARY A good management of water resources in agriculture requires being able to monitor and control the crops water budget. This information is useful for example to estimate the water use at regional scale, or to help applying the right amount of irrigation at the right time for each specific plot. In this regard, remote sensing can bring a key information which is the evapotranspiration (ET) of cultivated areas, which cannot be obtained accurately by other means. ET estimates can be obtained using basically two approaches. The first family of approaches uses thermal remote sensing data (TIR), i.e. observation of the temperature of the field surface, coupled with ground meteorological data, to estimate ET in solving the energy budget of the surface. The main variations in these TIR approaches are related to the assumption of homogeneity of the surface, either a mixture of vegetation and bare soil (“single source” approach) either a juxtaposition of bare soil avec vegetation modelled separately (“dual source” approach). There are also various means to interpret the temperature obtained from satellite, either sticking to the sole energy budget to interpret the temperature, or comparing it to all pixels in the area assuming that dry and wet surfaces are encountered simultaneously in the area (“contextual” approach). The main problem with the TIR approach is that the estimates are made at the time of satellite overpass and needs assumptions to be extrapolated to the day, then to several days. Another major pitfall is that TIR data is today poorly available, with daily data at very low resolution (1 km, MODIS), and 90 m resolution data only every 16 days (Landsat). However, there is a good probability that a high resolution high repetitivity TIR mission will be launched in the coming years. The second approach is to use water budget simulation of the crop (ex FAO-56 approach), feeding it with ground meteorological data and with remote sensing estimates of the actual vegetation at the surface using vegetation indices (ex NDVI). This approach has the advantage of providing a continuous simulation of the surface behaviour at hourly to daily time step, but its main limitation relies in the lack of information regarding the soil moisture, i.e. the irrigation input cannot be known for each plot and must be either simulated, either estimated based on thermal or radar data. It is clear to day that improved estimates of ET requires the combination of both above approaches, adding information about soil moisture obtained using radar data (Sentinel1). A typical operational configuration would be a land surface model driven by vegetation estimates using NDVI, assimilating in one hand information about vegetation stress and instantaneous ET brought by thermal data, and in other hand soil moisture estimates brought by SAR data. BIOGRAPHY

Dr. ABdelghani Chehbouni is a doctor in hydrology and remote sensing, he received his B.S. in fluids Mechanics from University Paul Sabatier, Toulouse, France in 1987 and his M.S. in Environment Science from the National Polytechnic Institute of Toulouse in 1989, and a PhD in Hydrology and Remote Sensing from University Paul Sabatier in Toulouse in 1992. He has been working at the University of Arizona for 18 months and the Jet Propulsion Laboratory for 2 years. He is co-investigator of several Remote Sensing programs (VEGETATION, ERS2/ATSR2, EOS). His main research interests include the application of remotely sensed data in land-surface-atmosphere models especially in arid semi-arid regions. He participates in several international experiments (MONSOON'90, Hapex Sahel) and he wasa co-leader of the Semi-Arid-Land-Surface-Atmosphere (SALSA) international program. He joined CESBIO in 2000 as the leader of hydrology group.