Phenology Index-Based Method for Mapping Winter Wheat and Summer Maize Rotation Cropping Pattern With Sentinel-2 Imagery

Abstract

As a common agricultural intensification, the winter wheat and summer maize rotation cropping pattern (wheat–maize) plays a crucial role in achieving sustainable food security in China. Reliable regional wheat–maize maps are of great importance to ensure the sustainability of agro-ecosystems. However, conventional previous studies typically depended on vegetation index time-series for detecting wheat–maize, which was challenging for rapid wheat–maize mapping. This study proposed a simpler phenology index-based method for mapping wheat–maize from multitemporal Sentinel-2 data. To better explore the mapping performance, two indices [i.e., normalized difference vegetation index (NDVI) and two-band enhanced vegetation index (EVI2)] and two mathematical combinations (i.e., multiplication and addition) were introduced to generate four uncorrelated indices. The wheat–maize maps obtained using phenology indices were evaluated using samples and high-precision maps derived from random forest. The results showed that the resulting maps achieved high overall accuracy of above 94% and F1-score of over 0.95, as well as agreed well with random forest derived maps (overall accuracy ≥ 91%, F1-score ≥ 0.88). In addition, this study found that EVI2 was better suited for designing phenology difference-based index than NDVI; concerning combination approaches, multiplication performed better than addition in enhancing spectral differences. Our results demonstrated the advantages of index-based method in mapping wheat–maize and its potential to be applied over larger regions. We hope that this study will advance our understanding of phenology-based methods in agriculture mapping.

Keywords

• Crop phenology
• Google Earth engine
• phenology index-based method
• Sentinel-2
• wheat–maize mapping