Wheat growth modelling by a combination of a biophysical model approach and hyperspectral remote sensing data

摘要

The study presented here investigates the potential of improvement for a physically based model approach, when the static input data is enhanced by dynamic remote sensing information. The land surface model PROMET (Processes of Radiation, Mass and Energy Transfer) was generally applied, while the remote sensing input data was derived from hyperspectral data of the CHRIS (Compact High Resolution Imaging Spectrometer) sensor, which is operated by ESA (European Space Agency).rnThe PROMET model, whose vegetation routine basically applies the Farquhar et al. photosynthesis approach, was set up to a field scale model run (10 x 10m) for a test acre tilled with wheat (Triticum aestivum L.) mapping the crop development of the season 2005. During the model run, information on the absorptive capacity of the leaves for two canopy layers (top, sunlit layer and bottom, shaded layer) was updated from remote sensing measurements, where angular CHRIS images were available. Control data were acquired through an intensive field campaign, which monitored the development of the stand throughout the vegetation period of the year 2005, also accompanying the satellite overflights.rnWhile the model without additional dynamic input data was able to reasonably reproduce the average development of the crop and yield, the spatial heterogeneity was severely underestimated. The combination of remote sensing information with the vegetation model led to a significant improvement of both the spatial heterogeneity of the crop development in the model and yield, which again entailed an overall improvement of the model results in comparison to measured reference data.