Remote sensing extraction and change analysis of surface water body over the Middle and Lower Reaches of the Yangtze River based on a multi-element coupling framework
Under global climate change and the long-term high-intensity exploitation of resources,surface water body in the middle and lower reaches of the Yangtze River region(MLRYRR)has become more complex,extreme,and hazardous in recent decades.However,technical inadequacy in remote sensing water extraction still exists because of the complexity of regional land cover,fragmented structures,and inconsistent feature characteristics.This study aims to propose a framework that combines multisource data fusion to extract a water body under complex geographical conditions at the basin scale.Accordingly,it shows the spatiotemporal pattern of surface water body over MLRYRR.First,the spectral features of several land objects with a water-like spectrum were explored based on multispectral remote sensing images.An enhanced remote sensing water index for large-scale spatial and temporal water extraction was introduced.This index can effectively differentiate water bodies with aquatic plants and vegetation in subtropical regions.Second,the proposed index was incorporated into an automatic water extraction model through the decision-level fusion of multisource geographic information data(i.e.,topography,hydrology,and impervious surfaces).Third,considering the seasonal variation of surface water bodies,a frequency-based classification scheme was introduced to estimate the yearly distribution of stable and seasonal water at 30 m spatial resolution in MLRYRR from 1984 to 2020.On the basis of the proposed framework and the Google Earth Engine platform,the annual spatial distribution data of stable and seasonal water bodies at a spatial resolution of 30 m in MLRYRR from 1984 to 2020 were obtained.The produced data were validated using 9000 validation samples in different scenes(e.g.,urban,agricultural,and lacustrine scenes)and achieved a recall of 98.4％.Results showed that the spatiotemporal distribution of surface water and its trends demonstrate regional heterogeneity,with the water area in Jiangsu and Zhejiang Provinces expanding at 35.1 km2·a-1 and 6.5 km2·a-1,respectively,and the water area in Anhui,Jiangxi,and Hunan Provinces decreasing at 46.53,35.6,and 26 km2·a-1,respectively.Moreover,the results of the annual water body area in MLRYRR can spatially reflect the drought and flood situations in different watersheds.The change trends of the water area in Hubei Province and Shanghai were insignificant.The mode and intensity of human disturbance and geo-climatic factors were the driving factors of the pattern differentiation in water evolution.The proposed surface water extraction framework and data results contribute to improving our understanding of the spatiotemporal distribution,evolution processes,and environmental effects of surface water.The results can provide spatial data and monitoring techniques to support surface water resource spatial investigation and the optimization of resource allocation,coordinated development,disaster risk assessment,and early warning.Future studies will focus on the dynamic observation method on surface water bodies through the collaborative processing of optical and synthetic aperture radar images to break through limitations imposed by continuously cloudy and rainy conditions in subtropical regions.
subtropical remote sensingwater resourceswater body extractiontime-series analysisGoogle Earth Enginethe Middle and Lower Reaches of the Yangtze River Region
万方数据
维普
