Coupling a sugarcane crop model with the remotely sensed time series of fIPAR to optimise the yield estimation

The objective of this study was to assess the efficiency of the assimilation of the fraction of intercepted photosynthetically active radiation (fIPAR) data derived from Satellite Pour l'Observation de la Terre SPOT images into the MOSICAS sugarcane crop growth model for estimating the yield at field scale on Reunion Island. Over 3 years, time series of SPOT satellite imagery were used to estimate the daily evolution of NDVI for 60 plots located on two climatically contrasted farms. Ground measurements of the fIPAR were performed on 5 reference fields and used to calibrate a relationship with the corresponding NDVI values. Forced and not forced simulations were run and compared with respect to their ability to predict the final observed yield. Forcing MOSICAS with fIPAR values derived from SPOT images improved the accuracy of the model for the yield estimation (RMSE = 12.2 against 14.8 t ha(-1)) closer to the 1:1 line. However, underestimations of the yield by the forced model suggest that some of the model parameters were not optimal. The maximal radiation use efficiency parameter (RUEm) was optimised for each field, and an analysis of variance showed the significant effect of the ratoon number of the field, of its cultivar and of the farm where it is planted. Accordingly, the RUEm was recalibrated for each cultivar for the number of ratoons and farms. New RUEm values ranged from 3.09 to 3.77 gMJ(-1), and new computations were run using the optimised values of RUEm The results indicate that recalibrating the maximal radiation use efficiency according to the number of ratoons improved the yield estimation accuracy by as much as 10.5 t ha-1 RMSE. This study highlights the potential of time series of satellite images to enhance the estimation of the yield by a forced ecophysiological model and to obtain better knowledge about the ecophysiological processes that are involved in crop dynamics with the recalibration method. (C) 2014 Elsevier B.V. All rights reserved.