Identifying major crop types in Eastern Canada using a fuzzy decision tree classifier and phenological indicators derived from time series MODIS data.

Liu, J., Huffman, E.C., Shang, J., Qian, B., Dong, T., and Zhang, Y. (2016). "Identifying major crop types in Eastern Canada using a fuzzy decision tree classifier and phenological indicators derived from time series MODIS data.", Canadian Journal of Remote Sensing, 42(3), pp. 259-273. doi : 10.1080/07038992.2016.1171133  Access to full text

Abstract

This paper presents a methodology for identifying major crop types in southwestern Ontario in eastern Canada using a fuzzy decision tree classifier and phenological indicators derived from remote sensing data. Phenological indicators were derived from time series Normalized Difference Vegetation Index (NDVI) calculated from 250-m surface reflectance data of the Moderate Resolution Imaging Spectroradiometer (MODIS). Training and testing samples were derived from crop classification maps at 30-m resolution for 2011, 2012 and 2013. Training samples for 2013 were used for discrimination rule development, and the classifier was then applied to all three years. Results showed that the classifier was able to discriminate major crop types such as winter wheat, corn, soybean, and forage crops with an overall accuracy of 75.3% for 2013 and comparable accuracy for 2011 and 2012. Confusion exists mainly between corn and soybean, and between winter wheat and forage crops. This indicates that phenological indicators derived from optical remote sensing data are intrinsic to a crop and may be more indicative than the commonly used remote sensing features that are susceptible to environmental and management impacts. This methodology provides an opportunity for discriminating general crop types without requiring a year-specific training sample set.

Date modified: