Résumé
A generic crop model (STICS) was adapted (STICS-Grassland) to model growth, yield, and environmental impacts of grasslands in France. It is a semimechanistic model combining equations of physiological processes and mathematical relationships between processes in a daily time step. The Information et Suivi Objectif des Prairies (ISOP; Grassland Information and Objective Survey) application was developed to estimate and map the real-time status of grass growth and forage production in the 200 forage regions of France, to help decision makers anticipate forage availability in case of severe deficits. Initially, the model could not simulate long, severe droughts typical of Mediterranean regions when plant and tiller densities are significantly reduced. A tiller density module models the dynamics of tiller death during droughts. Since STICS is based on the concept of a mean plant (tiller) covering the whole field, variability was introduced through a law distribution of transpiration deficit by tiller, a threshold value imposing desiccation, and death to tillers reaching this level. Its second part generates incomplete or total recovery of tiller density through new tillering in the next wet season. Species and contrasting groups of cultivars within species can be characterized with a limited number of parameters.