A Phenology-Based Cropping Pattern (PBCP) Mapping Method Based on Remotely Sensed Time-Series Vegetation Index Data

Abstract

Cropping patterns are closely related to food production, cropland intensification, water resource management, greenhouse gas emission and regional climate alteration. Timely and accurate mapping of cropping patterns is urgently needed in many disciplines. However, the existing cropland-related datasets are informative at the global level, but lack regional-scale details about cropland utilizations. Thus, there is a need for better information on the area and distribution of cropping patterns at regional scales. In this study, we developed a phenology-based cropping pattern (PBCP) mapping method based on remote sensing vegetation index time series. The new method first extracted vegetation phenological metrics (start of season (SOS), end of season (EOS), growing season length (GSL) and growth amplitude (GA)) from the vegetation index time series. Then, it identified crop seasons by using the minimum crop GSL, the minimum crop GA and the maximum crop GSL, which were derived from the training samples. Finally, cropping patterns were classified based on a set of decision rules. The case study in Henan province of China showed that, the results indicated that: (1) compared with cropping index derived from the supervised classification of Landsat-5 TM images, the PBCP method provided cropping index with satisfactory accuracy of 85.3%. (2) Validation sample analysis indicated that the cropping pattern mapping accuracy was 84% for the PBCP method. Different to current cropping pattern mapping methods, the PBCP method considered crop planting information in three years in deciding the cropping pattern to map the dominant cropping patterns. It can provide new insights in agriculture related land use analysis.