IWRA Proceedings

AbstractIrrigated agriculture makes a major contribution to food security, producing nearly 40% of food and agricultural commodities on 17% of cultivated lands (FAO 2002). However, serious water shortages occur in arid and semi-arid areas as existing resources reach full exploitation. The design of tools providing with regional estimates of water balance and crop yield is necessary to ensure a sustainable development of these areas. Crop simulation models are used to describe the effect of climate, soil and agricultural practices on crop growth and crop production. Although the performance and accuracy of crop models have continuously made progresses over the past few years, applications for yield forecasting over large areas are limited. In particular, there are a generally large number of model parameters compared to the amount of observation available for their identification over each agricultural unit (field). Furthermore, it is difficult to cope with the lack of adequate and sufficient input data to run the model at a regional scale. This is particularly true for the information about technical practices such sowing, irrigation and fertilisation schedules, which know large space time variations. In this context, Earth Observation Systems designed to provide both high spatial resolution (10 m) and frequent (daily) time of revisit offer strong opportunities. At the present time, two of them appear of particular interest: 1) the FORMOSAT-2 Taiwanese satellite (http://www.nspo.org.tw), which provides high spatial resolution (8m) images in 4 bands ranging from blue to near-infrared spectral domains ; 2) the Advanced Synthetic Aperture Radar (ASAR), onboard ENVISAT mission (http://envisat.esa.int/), which operates in the C-band with 7 different incidence angles between 15 and 45 degrees at a spatial resolution of about 12.5 m in the Alternating Polarisation mode. The orbital cycle of ASAR is 35 days, but the combination of acquisitions with different incidence and path configurations allows a revisit time of a few days. In this study, we combine FORMOSAT-2 and ASAR data in order to monitor wheat crops over a 10 km² irrigated area in the Tensift/Marrakech plain. We use two time series of images acquired during the 2005/2006 agricultural season together with experimental data collected on wheat crops (see Duchemin et al., International Journal of Remote Sensing, in press). We first investigate the potential of these satellite data to characterise both soil management practices (ploughing, sowing, irrigation) and the vegetation dynamics (inversion of biophysical variables such as the leaf area index). In a second step, this information is incorporated into a simplified canopy functioning model to provide spatial estimates of evapotranspiration and wheat production. The outputs of the model are evaluated at the field scale using soil moisture and biomass data collected at ground on 10 wheat fields as well as at the scale of irrigation unit using data provided by the regional agricultural agency.