Estimating the spatial and temporal distribution of sowing dates for regional water management

Accurate prediction of water demand by agriculture is a key point of water management. Current procedures of water management often use simple models and parameterisation to predict demand, that may fail in particular climatic situations. In a particular area of south-western France, it is suspected that such failure was due to poor estimation of sowing dates. To solve this problem, a model was developed to estimate the spatial and temporal distribution of summer crop sowing dates. This paper presents the development and the test of the model. The model (SIMSEM) is based on a farm typology, a soil water balance model and supplemental expert knowledge. Although many inaccuracies remain, the distributions predicted compared fairly well to observed distributions obtained from a field survey. The model improved estimated sowing dates compared to the method currently used in water management calculations. The use of SIMSEM to provide information about sowing dates seems helpful for better predicting water demand, as shown by tests performed for the critical 1998 year where the current procedure failed.