Improved modelled crop water-use and crop productivity using thermal and optical remote sensing data

Accurately monitoring crop water use and crop photosynthesis is essential for predicting crop yield and optimising management strategies, in order to sustainably enhance crop productivity and maintain future food security. The synthesis of  thermal and multi-spectral remote sensing dataset is a promising approach for improving estimates of crop water and carbon fluxes at scales from the individual plant to regional extents. This study uses remotely-sensed data collected from ground-based, drone and satellite sensors to model evapotranspiration (ET) and crop productivity in wheat (Triticum aestivum) across a growing season at a flux tower site in northern England. A ground-based FLIR thermal camera (T530) was utilized to measure the diurnal variations in temperature every 30 minutes between 10AM till 2PM. A Parrot Analfi thermal drone and DJI Matrice 200 Series V2 paired with the Micasense Rededge MX multispectral sensor were flown at an altitude of 20 m, 7 times across the growing season, along with satellite-based Landsat-8 TIRS (100 m) and Sentinel-2 (60 m) multi-spectral data. Crop water and carbon fluxes were modelled using the Biosphere-atmosphere Exchange Process Simulator (BEPS), where Vc_max was dynamically constrained in BEPS through its relationship with leaf chlorophyll content from drone and Sentinel-2 data using a radiative-transfer based approach (PROSAIL). BEPS-modelled ET estimates compared results from a physical-based Surface Energy Balance System (SEBS), 3T, modified 3T, and simplified model which bypasses the requirement of input net radiation by incorporating a dry reference surface. Modelled photosynthesis and ET results were compared with in-situ eddy-covariance flux tower observations. For three out of 7 days of simulated results, temporal variations in modelled ET compared with flux tower observations at half hourly scale demonstrated index of agreement values of 0.74, 0.80, and 0.68, and Pearson’s Correlation values of 0.72, 0.65, and 0.64, respectively, for the 3T, modified 3T, and simplified ET model. Our results demonstrate the potential of synergizing drone, ground-based, and satellite platforms for providing accurate prediction of crop water use and productivity for a sustainable crop yield.