Crop classification using multi-configuration SAR data in the North China Plain

Abstract

Crop classification is a key issue for agricultural monitoring using remote-sensing techniques. Synthetic aperture radar (SAR) data are attractive for crop classification because of their all-weather, all-day imaging capability. The objective of this study is to investigate the capability of SAR data for crop classification in the North China Plain. Multi-temporal Envisat advanced synthetic aperture radar (ASAR) and TerraSAR data were acquired. A support vector machine (SVM) classifier was selected for the classification using different combinations of these SAR data and texture features. The results indicated that multi-configuration SAR data achieved satisfactory classification accuracy (best overall accuracy of 91.83%) in the North China Plain. ASAR performed slightly better than TerraSAR data acquired in the same time span for crop classification, while the combination of two frequencies of SAR data (C- and X-band) was better than the multi-temporal C-band data. Two temporal ASAR data acquired in late jointing and flowering periods achieved sufficient classification accuracy, and adding data to the early jointing period had little effect on improving classification accuracy. In addition, texture features of SAR data were also useful for improving classification accuracy. SAR data have considerable potential for agricultural monitoring and can become a suitable complementary data source to optical data.

Acknowledgements

The work in this article was supported by the Knowledge Innovation Project of the Chinese Academy of Sciences (no. KSCX1-YW-09), the National Natural Science Foundation of China (no. 41071277), the National Key Technology R&D Programme (no. 2008BADA8B02) and the National High Technology Research and Development Programme of China (863 Programme) (no. 2009AA12Z1462). The authors also thank ESA for providing the ASAR data through ESA–NRSCC Cooperational Dragon 2 Programme (ID: 5279) and DLR for providing the TerraSAR data through the TerraSAR-X Science Plan Programme (ID: LAN0563) in this study. We are grateful for the comments and suggestions of two anonymous referees, and for the help of the editor, Prof. Michael Collins, all of which have led to improvements in the presentation of this article.

Notes