Mapping paddy rice using multi-temporal MODIS images

In this study, temporal MODIS data are used to classify double rice crops in Taitung, Taiwan. Empirical Mode Decomposition (EMD) is applied for data noise filtering. Three similarly assessments, including minimum squared Euclidean distance (MSED), cosine similarity, and correlation coefficient methods, are utilized in the classification of double crops. The accuracy assessment of the double crops classification was evaluated. The results indicate that EMD is able to filter out the noises of the time series data and successfully preserve the temporal and spectral patterns for rice paddies classification.