Study on estimating the planting area of winter wheat based on mixed field decomposition of remote sensing

With the significantly improved data availability in remote sensing technology, mid-resolution images have become the primary data source for crop sown area measurement in large scale. However, it is still difficult to solve the problems of spectrum heterogeneity in one field and spectra similarity between fields. This paper developed mixed field decomposition method and tested the method in an urban agriculture region with complex plant structure through several steps: first, distinguishing the mixed parcels by calculating the coefficient of variation of multi-temporal TM image within the parcels; then, operating multivariate regression model and mixed field decomposition model based on support vector machine (SVM) to estimate the sown area of winter wheat in the mixed parcels with different sample size. Results show that the mixed field decomposition of SVM has a higher accuracy than the multivariate regression model both in amount and position.