IWRA Proceedings

AbstractIrrigation is the world major consumer of water, using about 65% of the resources. As more and more arid areas are facing water shortage problems, the optimization of irrigation is a major challenge in these areas. Besides the improvement of irrigation techniques at the field level (crop budget calculation, drip irrigation, etc.), the design of tools providing with spatialized estimates of water budget over a whole region is useful for manager. This is one of the objective of the SudMed project, which is carried out by CESBIO (Toulouse, France) and the University Cadi Ayyad (Marrakech, Morocco) in the Haouz plain around Marrakech (10000 km2). Evapotranspiration (ET) of the irrigated crops is one of the major fluxes conditioning the water budget. ET assessment basically requires information about climate, soil moisture and vegetation state, the latter being efficiently obtained from satellite images (vegetation type and development stage). High resolution remote sensing time series are now available quite easily (Landsat, SPOT) and ongoing missions will improve the offer providing images at higher frequency (4 days frequency : FORMOSAT, Venµs). These images allow an accurate monitoring of vegetation development over large areas, which is critical for estimating ET. We developed a tool for spatializing ET and water balance based on the FAO method (Allen et al. 1998 mettre la réf entière) and satellite data. This method requires three types of data, linked to climate, land cover and crop development stages. The climatic data is obtained from ground measurements or from meteorological model outputs, and is used to compute the reference evapotranspiration ET0. The land cover in the area is complex and characterised by a high inter annual variability, especially for annual crops. The tool is based on simplified classes, namely trees, annual crops and bare soil, which are suitable at this stage for water budget. The trees class was mapped using a very high resolution image and is considered stable from one year to another. The annual crops distribution is unknown at the beginning of each season and thus obtained from scenarios based on water availability in dams and rainfall hypothesis. Then, this land cover is adjusted along the season when images are acquired. The crop coefficients (Kc) used in the FAO method are also derived from images using NDVI time series The tool has the ability to compute past budget from the beginning of the season until the date of observation. Moreover, forecasts of the water needs for the rest of the season are computed based on simulation of crop phenology.