1987—2022年新安江水库(千岛湖)水面面积时空变化及其与水位、蓄水量的响应关系

水面面积、水位、蓄水量是水库水资源管理的重要基础数据,遥感是湖库水体提取、水位和蓄水量估算的重要技术手段。由于不同水体提取方法的适用性差异、测高卫星数据的有限时间覆盖度和开源数据集的时空分辨率不足等原因,湖库水面面积、水位、蓄水量的长时序、高频率时空变化监测仍存在一定挑战。本研究以新安江水库为研究区,结合多源遥感、气象、水文和土地利用等数据,基于Google Earth Engine云平台,运用水体指数法,分析1987—2022年新安江水库水面面积时空变化特征,构建水位—水面面积、水位—蓄水量和水面面积—蓄水量响应关系,探究水面面积时空变化成因。结果表明:(1)Landsat 5、Landsat 8和哨兵2号数据的最佳水体提取指数分别为AWEI_sh和GNDWI,F1-score分别为91.93%、91.03%和93.14%。相比于开放数据集GSWED(32.61%)、JRC GSW(76.17%)和ReaLSAT(69.76%),基于最优水体指数的水体提取结果具有最高的F1-score(91.26%);(2)时间上,1987—2022年新安江水库水面面积呈显著上升...

Long-term spatiotemporal variation in water area of the Xin''anjiang Reservoir (Lake Qiandao) from 1987 to 2022 and its relationships with water level and water storage

Water area, water level and water storage of reservoirs are fundamental data for water resource management. The method of satellite-based remote sensing has been demonstrated as an effective method for the estimation of water area, water level and water storage. However, the different applicability of the remote water extraction method and the limited time coverage of altimetry data bring difficulties to the study of the spatiotemporal variation characteristics of reservoir water area, water level, and water storage and their response relationships. Although many open-source data of reservoir water area and reservoir water storage at regional- and global-scale were published, these data were still hard to satisfy the long-term and high-frequency remote estimating reservoir dynamics due to its limitation in spatiotemporal resolution. Xin''anjiang Reservoir (Lake Qiandao), the largest man-made reservoir in eastern China, is a popular tourist destination and an important drinking water source in the Yangtze River Delta region. Here, we used the water index to extract the water area of the Xin''anjiang Reservoir from 1987 to 2022 with Landsat series data (1987-2022) and Sentinel-2 data (2019-2022), established response relationships of water level-water area, water level-water storage, and water area-water storage based on in-situ long-term water level and water storage data, and explored the causes of spatiotemporal changes in water area with land use data and hydrometeorological data. The results showed that: (1) The optimal water indexes for the Landsat series data and Sentinel-2 data to extract water extent in the Xin''anjiang Reservoir were AWEI_sh and GNDWI, respectively. Specifically, the F1-score are 91.93%, 91.03% and 93.14% for Landsat 5, Landsat 8 and Sentinel-2, respectively. Our method showed the highest F1-score value (91.26%) than the open-source dataset, GSWED (32.61%), JRC GSW (76.17%) and ReaLSAT (69.76%). (2) Temporally, the water area of the Xin''anjiang Reservoir had exhibited an upward trend from 1987 to 2022 (R²=0.20; P=0.01), with an overall growth rate of 0.96 km²/a. Spatially, permanent water bodies (with a water inundation frequency greater than 75%) accounted for 73.44% of the water area, primarily distributed in open areas. Seasonal water bodies (with a water inundation frequency greater than 25% and less than or equal to 75%) accounted for 10.17% of the water area, primarily distributed in the lake branch area. (3) The third-degree polynomial function showed the best performance in fitting the response relationships of water level-water area, water level-water storage, and water area-water storage in Xin''anjiang Reservoir. (4) The main factors causing dynamic changes in the water area of the Xin''anjiang Reservoir were precipitation in the upper reaches of the Lake Qiandao basin and land use changes caused by human activities.