Remote sensing estimation of catchment-scale reservoir water impoundment in the upper Yellow River and implications for river discharge alteration

The impoundment and regulation information of reservoirs is essential for water resource management and hydrological analysis. Along the upper reach of Yellow River in northern China, a large number of reservoirs have been built or are being built, while information about the distribution and impoundment of these reservoirs is rather limited. Here we compiled an inventory of reservoirs ( > 0.01 km(2)) in two catchments in the upper reach of the Yellow River and estimated the seasonal water storage variations of reservoirs associated with impoundment during 2000 and 2018, by using a suite of remote sensing observations. The inventory of 149 reservoirs in the study region consists of 102 existing reservoirs (built before 2000) retrieved from Google Earth high-resolution imagery or available publications and 47 new reservoirs (constructed during 2000-2018) which we detected by using time-series analysis of NDWI derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) surface reflectance data. To estimate water storage changes of all reservoirs associated with impoundment, we combined representative inundation extents of reservoirs in each season of the year retrieved from sequential Landsat images on the Google Earth Engine (GEE) platform and shoreline topography information derived from the Shuttle Radar Topography Mission DEM data. The results demonstrate significant seasonal variations (mostly impoundment in autumn/winter and release in spring/summer) as well as a rapid increasing trend of water storage held in reservoirs. Particularly in Catchment A - the headwater of the Yellow River, the total water storage increased by about 16 Gt from 2000 to 2018, about 25% percent of which is due to impoundment of newly constructed reservoirs. The remote sensing-based results were validated by long-term in-situ water level data of 2 reservoirs and construction information of 14 new reservoirs. Uncertainty analysis shows acceptable accuracy in monitoring storage changes on the annual scale, indicated by high confidence in area-volume curves for most reservoirs (102 reservoirs with R-2 of area-volume curve higher than 0.9), while uncertainty is unneglectable on the seasonal scale. In accordance with streamflow variation data from 2000 to 2016 provided by published literature, our estimation indicates that the water resource in the upper Yellow River has been more highly regulated by reservoir impoundment and the seasonal regulation exerts significant impact on catchment runoff process (averagely about 17% of annual discharge).