Integration of remote sensing and crop modeling for the early assessment of durum wheat harvest at the field scale

In the Mediterranean, durum wheat is one of the major crops, but a high variability of grain yield and protein concentration (GPC) prevents an adequate agronomic planning at the farm or consortium level. Although there are many stud- ies on monitoring of crop production and early prediction of yields, little has been done at the local scale. The aim of this study was to assess simpli ed integration algorithms (SIAs) for inte- grating remote sensing information with a crop model, to forecast the GPC and grain yield at the eld scale. To this end, the CERES-Wheat model was run to simulate the seasonal aver- age of grain yield (AVE) and GPC in Val d’ Orcia (Tuscany Region, Italy) during the 2009–2010 and 2010–2011 growing seasons. The perfor- mances of different vegetation indices from MODIS imagery in harvest forecasting were assessed and compared. The SIA formulation was based on the simulated AVE and GPC, and on their spatialization in relation to the intraan- nual variability between the elds described by vegetation indices. The simulated AVE traced the observed trend. The fraction of absorbed photosynthetically active radiation (fPAR) was the best index in describing grain yield, and the related SIA showed at validation good perfor- mance at the eld scale (r2 = 0.74). Conversely, the SIA was unable to predict GPC due to the low performance of CERES-Wheat in capturing the interannual variability and to the failure of the fPAR in describing the GPC inter elds variability at intermediate canopy re ectance values.