Mapping South Korea's Paddy Rice During 2001-2015 Based on Landsat Images

Abstract:

Agricultural land use change substantially affects terrestrial water and carbon cycles, and it is also significant for food security. Rice is an important food crop in South Korea, and its planting area has changed over the past years due to climate change and social development. However, the paddy rice maps in South Korea with high spatial, temporal resolutions and quality are still unavailable. In this study, annual paddy rice planting areas in South Korea during 2001-2015 were extracted using the phenology-based and artificial neural networks(ANN)-based method according to the availability of Landsat images. Three types of vegetation indices, Normalized Difference Vegetation Index(NDVI), Normalized Difference Water Index (NDWI) and Land Surface Water Index (LSWI), were calculated for both methods. For the phenology-based method, we calculated difference and ratio of the vegetation indices from the Landsat images before and after the inundating/transplanting phase, and set the extraction thresholds for rice paddy using the training samples. For the ANN-based method, using the vegetation indices of the Landsat image after inundation, slope and soil type data of the training samples as input, an ANN was established for rice paddy identification. The resultant maps of paddy rice were evaluated using validation samples and agricultural statistics. Finally, we analyzed the spatial and temporal patterns of South Korea's rice paddy during 2001-2015, and investigated the possible reasons.