Summer Maize Mapping by Compositing Time Series Sentinel-1A Imagery Based on Crop Growth Cycles

The accurate and timely mapping of summer maize is vital for agricultural management and food security. Time-series remotely sensed imagery provides a promising data resource for crop mapping by characterizing growth cycles over time. Therefore, this study explores summer maize mapping with composited time-series Sentinel-1A imagery of a typical crop field in China. First, time-series backscattering coefficients of major land-cover types (i.e. summer maize, peanut, forest, settlements, and water) are extracted from multi-temporal Sentinel-1A imagery. Second, according to the growth cycles of summer maize and peanut, the multi-temporal Sentinel-1A images are composited to enhance the characteristics of the summer maize growth cycle, while also eliminating redundant information and differences in phenology. Third, the decision-tree method is used to perform pixel-level classification; samples with an area of 1 km(2) are collected as validation datasets. The results show that Sentinel-1A VH-polarized images are more sensitive to the summer maize growth cycle than VV-polarized images. The summer maize cropping areas are estimated with an overall accuracy of 96.55% and a kappa coefficient of 0.93. The results suggest that multi-temporal Sentinel-1A imagery is capable of characterizing the growth cycle of summer maize, and provides a promising solution for accurate summer maize mapping, irrespective of weather conditions.

Automated summary

This study successfully mapped summer maize planting areas in China using composited time-series Sentinel-1A imagery. The VH-polarized images were found to be more sensitive to the growth cycle of summer maize, with an overall accuracy of 96.55% and a kappa coefficient of 0.93.