Efficiency based wheat yield prediction in a semi-arid climate using surface energy budgeting with satellite observations

Abstract

Accurate pre-harvest assessment of a staple food crop is an integral part of policy formulation in relation to food security issues. Here, two different approaches were attempted to estimate wheat yield using time series multi-year satellite (MODIS Aqua) optical-thermal data from a single earth observation (EO) mission. Surface energy budgeting was used to estimate evapotranspiration in terms of latent heat fluxes from net available energy and evaporative fraction to predict wheat yield over four agro-climate zones in semi-arid climate of Gujarat, India. Satellite based estimates of latent heat fluxes were found to show substantially less error with respect to the area-averaged heat flux measurements from LAS (large aperture scintillometer) as compared to measurements from BREB (Bowen Ratio Energy Balance) alone. The deviations in satellite based zonal CWU were found to have a strong correlation ( r = 0.71) with the deviations from zonal wheat yield. Among both the approaches, the radiation use efficiency ( RUE) based approach produced better accuracy in the predicted yield with lower root mean square error (RMSE) of 390 kg ha −1 (14.8% of reported mean) and higher correlation coefficient ( r = 0.92) than the water use efficiency ( WUE) based approach (RMSE 573 kg ha −1, 21.8% of reported mean; r = 0.80). Uncertainties in the satellite based core inputs resulted into a net 10-12% error in predicted yield in case of RUE approach. Our demonstrative case studies recommend that the coupled use of satellite observations from multiple EO missions and radiative transfer simulation would be effective to make efficiency based approaches operationally viable for regional wheat yield forecasting in near real time.