improved automated land cover updating approach by integrating with downscaled NDVI time series data

Land cover change monitoring is important for climate and environmental research. An automated approach for updating land cover maps derived from Landsat-like data is urgently needed to process large amounts of data. Change detection is an important part of the updating approach; however, pseudo-changes commonly occur because satellite images acquired in different seasons can capture phenological differences. Change detection based on normalized difference vegetation index (NDVI) time series data could avoid this problem; nevertheless it suffers from the much lower spatial resolution of the NDVI data. To address the resolution issue, this study improves an automated land cover updating approach by integrating downscaled NDVI time series data. First, Moderate Resolution Imaging Spectroradiometer (MODIS) NDVI data at 250-m resolution are downscaled to 30 m using the NDVI linear mixing growth model. Then, the NDVI-based change detection method is used to detect the changed/unchanged areas, and the unchanged areas are removed from the changed areas that were detected using the original land cover updating approach. A case study shows that the NDVI-based change detection module sufficiently removed the pseudo-change caused by seasonal differences and improved the land cover updating result, with an increase in overall accuracy from 76.6% to 89.6%.