Identification of winter wheat and its distribution using the CBERS-02B images in an irrigation district along the lower Yellow River, China

Crops and spatial distribution of their planting area are two important factors for agricultural water management. Remote sensing has been proved an effective technique in agricultural monitoring. Crop recognition and sown area monitoring are important topics in agricultural remote sensing using data sources from variety of sensors. This paper made efforts in extracting winter wheat and its sown area in an irrigation district along the lower Yellow River stream using the newly launched CBERS-02B sensor. Based on selection of the winter wheat training samples, spectral features, NDVI, MSAVI, spatial information of soils and texture analysis, a rule sets were developed for extracting winter wheat and its sown area. Google Earth was also employed to identify specific ground truth at a high resolution. It is tentatively concluded that the newly launched CBERS-02B CCD data is a reliable source for remote sensing monitoring for agriculture. The rule-based method proposed in this paper has improved the accuracy of crop monitoring. Integration of the spectral information, texture information, information of soils and land use/cover into the rule sets has strengthened identification capacity of the rule-based method. Compared to the unsupervised classification result, the rule-based crop recognition method achieved a better accuracy. Google Earth is a powerful tool which can be employed in sample selection and accuracy assessment. Its high resolution makes lots of small objects identifiable and identification mixed classes possible.