CROP MODELING AND REMOTE-SENSING FOR YIELD PREDICTION

Methods for the application of crop growth models, remote sensing and their integrative use for yield forecasting and prediction are present ed. First, the general principles of crop growth models are explained. When crop simulation models are used on regional scales, uncertainty and spatial variation in model parameters can result in broad bands of simulated yield. Remote sensing can be used to reduce some of this un certainty. With optical remote sensing, standard relations between the Weighted Difference Vegetation Index and fraction ground cover and LA I were established for a number of crops. The radar backscatter of agr icultural crops was found to be largely affected by canopy structure, and, for most crops, no consistent relationships with crop growth indi cators were established. Two approaches are described to integrate rem ote sensing data with crop growth models. In the first one, measures o f light interception (ground cover, LAI) estimated from optical remote sensing are used as forcing function in the models. In the second met hod, crop growth models are extended with remote sensing sub-models to simulate time-series of optical and radar remote sensing signals. The se simulated signals are compared to measured signals, and the crop gr owth model is re-calibrated to match simulated with measured remote se nsing data. The developed methods resulted in increased accuracy in th e simulation of crop growth and yield of wheat and sugar beet in a num ber of case-studies.