Estimating regional winter wheat yield with WOFOST through the assimilation of green area index retrieved from MODIS observations
Here, we describe and test a method for optimising winter wheat green area index (GAI) simulated with the WOFOST crop model using MODIS estimates of GAI in the Walloon region of Belgium. Detailed crop type maps during the period of 2000–2009 were used to derive time series of crop-specific GAI by selecting only the 250-m MODIS pixels that have at least 75% purity of the target crop. Two important model parameters were optimised by minimising the difference between the simulated and observed GAI for each individual pixel and year. The resulting year-specific joint parameter distributions were then used to run an ensemble of crop simulations in which the ensemble was initialised by sampling from the joint distribution of the corresponding year. The semi-variograms of the retrieved parameters revealed that the spatial patterns were consistent with agricultural practices and that seasonal characteristics of weather patterns in Wallonia can explain – at least partially – the temporal variability observed in the retrieved parameter distributions. Finally, the results of the average ensemble crop simulation were aggregated to the provincial and regional levels. A validation using yields reported by EUROSTAT over the period 2000–2009 revealed that assimilating MODIS with GAI provides an improved relationship between simulation results and reported yields at the regional level. âº Detailed crop maps are the key to good MODIS 250-m green area index estimates. âº Optimizing WOFOST parameters using MODIS GAI leads to consistent parameter distributions. âº Ensemble simulations with updated parameters perform better than default parameter values. âº Use of the 2000–2009 MODIS archive allows robust testing against EUROSTAT regional reported yields.