Ability for a model to predict crop production variability at the regional scale:: an evaluation for sugar beet

The spatial application of crop models for the prediction of plot yields on a regional scale has specific constraints. Firstly, it supposes accurate input into the model from sources concerning the spatial variability of crop performance. Furthermore, it is necessary for the model to be suitable for simulating interactions between the environment, agricultural practices and the crop which explain this variability. The aim of this article is to evaluate a sugar beet growth model, SUCROS, in terms of its ability to predict the spatial variability of yields on the scale of a beet-producing region. The model was adjusted to the agronomic context of northern France by taking into account the effects of moisture stress on leaf senescence and on the partitioning of assimilates. This modification reduced by more than a half the error of LAI and storage organ dry weight estimates. The modified model was then tested on thirty four plots of two beet production region, and we were able to demonstrate that its performance was closely linked to the accuracy of input variables concerning the climate and soil water properties of each plot.