Prolonged water body types dataset of urban agglomeration in central China from 1990 to 2021

Abstract

Data and knowledge of spatial-temporal dynamics of multiple types of water body are significant for water resources management but remain very limited. Using the Landsat satellite data and weakly supervised deep learning techniques for long term mapping, we report annual maps of multiple types of inland water bodies on urban agglomeration scale in the middle reaches of the Yangtze River (MRYR) during 1990-2021 at 30m spatial resolution. Accuracy assessment from 14000 validation points in seven years indicates an overall accuracy of 94.50%. Quantitative and qualitative comparison with other water related mapping products further demonstrates the superiority on long time span, refined classification system, and the high precision. The Water-MRYR could support the government and the public in the challenging water resources management and wetland conservation.