Résumé
In West African Soudano-Sahelian areas, agriculture is highly vulnerable to climate and populations suffer recurrent food crises related to droughts. To cope with climate variability majority of farmers still crop local photoperiod sensitive varieties [1] having medium grain yield potential (around 2500 kg.ha-1) and low harvest index, under traditional non-intensive practices (no fertiliser, neither pesticide, own seeds, no tillage in many situations, sandy soils, fertility managed through crop rotation and/or manure) [2]. For those areas yields forecasting tools improvement is an important challenge. Yet, addressing the question of climate impacts on crops requires measurements at regional scale [3] and adaptation of crop models. Those issues were addressed within AMMA project from 2005 to 2009, aiming at adapting the SarraH crop model [4] to farmers varieties, assessing and understanding farmers yields, and improving yield forecasting methodology using SarraH. Studies were carried out in Senegal, Mali, Burkina Faso and Niger, in 8 meso-scale sites (from 2.500 to 10.000 km2 each) with different agroclimatic (from 400 to 1000 mm annual rainfall) and agricultural features.