Fusing MODIS and OLI images: evaluating applications to winterwheat mapping

Abstract

Moderate resolution imaging spectroradiometer (MODIS) series products are widely used to observe Earth’s surface, but the coarse resolution of MODIS images may negatively affect the monitoring accuracy. In this paper, MODIS vegetation index data (NDVI, normalized difference vegetation index; EVI, enhanced vegetation index) were fused with operational land imager (OLI) data (Landsat 8 OLI: pan band) using two image transform methods (WT, wavelet transform; GST, Gram-Schmidt transform) to improve the accuracy of winter wheat mapping. The findings showed that image fusion resulted in increased information entropy for image interpretation; however, the mapping accuracy was improved based on the fused OLI and MODIS images for only the OLI pan image acquired during the winter wheat growing season. Winter wheat was more often misclassified as urban green spaces in the summer than in the spring. For the entire study area, the overestimation of winter wheat based on the fused EVI and OLI data and the WT algorithm reached 5.98%. The time-series curves of the fused images were similar to those of the MODIS images, suggesting that time-series MODIS images might have expanded application in phenology-based wheat mapping.