The final yield predictions are a combination of biomass estimations obtained by (i) the agrometeorological model B-CGMS, (ii) those obtained from satellite data and (iii) a time technological trend function.
Each one of these components is taken into account according to the cultural period for which the root mean square error representing the prediction error (PRMSE) is the smallest one. The B-CGMS system can be subdivided in three operational levels:
- level I : Level of weather data interpolation. At this stage, raw data from weather stations are spatialized on a regular grid.
- level II :
- Crop growth simulation: This is the heart of CGMS. It is based on a deterministic model of crop growth (WOFOST).
- Yield forecasting: based on the output of crop growth simulations. Data resulting from remote sensing and technological trend functions of crop yields are also taken into account.
- level III : Level of aggregation in standard administrative units at various spatial scales.