Mathematical Integration of Remotely-Sensed Information into a Crop Modelling Process for Mapping Crop Productivity

Abstract

Remote sensing is a useful technique to determine spatial variations in crop growth while crop modelling can reproduce temporal changes in crop growth. In this study, we formulated a hybrid system of remote sensing and crop modelling based on a random-effect model and the empirical Bayesian approach for parameter estimation. Moreover, the relationship between the reflectance and the leaf area index was incorporated into the statistical model. Plant growth and ground-based canopy reflectance data of paddy rice were measured at three study sites in South Korea. Spatiotemporal vegetation indices were processed using remotely-sensed data from the RapidEye satellite and the Communication Ocean and Meteorological Satellite (COMS). Solar insulation data were obtained from the Meteorological Imager (MI) sensor of the COMS. Reanalysis of air temperature data was collected from the Korea Local Analysis and Prediction System (KLAPS). We report on a statistical hybrid approach of crop modelling and remote sensing and a method to project spatiotemporal crop growth information. Our study results show that the crop growth values predicted using the hybrid scheme were in statistically acceptable agreement with the corresponding measurements. Simulated yields were not significantly different from the measured yields at p = 0.883 in calibration and p = 0.839 in validation, according to two-sample t tests. In a geospatial simulation of yield, no significant difference was found between the simulated and observed mean value at p = 0.392 based on a two-sample t test as well. The fabricated approach allows us to monitor crop growth information and estimate crop-modelling processes using remote sensing data from various platforms and optical sensors with different ground resolutions.