Estimating River Discharge Across Scales with a Novel Regional Gauging Method Driven by Sentinel Satellite Data

Hong Lin,Xiao Cheng,Junguo Liu,Qian Shi,Teng Li,Lei Zheng,Xuejiao Hou,Jinyang Du
DOI: https://doi.org/10.1016/j.rse.2024.114266
IF: 13.5
2024-01-01
Remote Sensing of Environment
Abstract:Accurately monitoring river discharge is paramount for effective water resource management, and the utilization of remote sensing has emerged as a promising approach. However, estimating river discharge from a single pixel is limited in its ability across various river sizes: suffering from the applicability of single-pixel size to different river scales. Here, we develop an innovative regional gauging method to help overcome these barriers and improve the accuracy of estimates. Our method focuses on periodically flooded areas that are acutely sensitive to alterations in river discharge, denoted discharge-sensitive areas (SAs). For a river reach, we achieved automated extraction of these SAs using Sentinel-1 data and obtained water fraction within SAs from Sentinel-2 images. Water fraction is used to establish a quadratic relationship with in-situ daily mean discharge observations for enabling satellite-based river discharge estimates. The method was rigorously evaluated on fourteen rivers worldwide, encompassing mean widths from approximately 50 m to 3000 m. The relative root mean square error observed in our estimates ranged from 7% to 50%, with the Nash-Sutcliffe efficiency ranging from 0.69 to 0.81 for smaller rivers (< 200 m wide), 0.86–0.98 for medium rivers (200–500 m wide) and 0.81–0.99 for large rivers (> 1 km wide). For the estimated discharge time series, the mean time sampling is 9 days, and the maximum time gap ranges from 2 to 10 months. For some stations, no high river flow could be retrieved. Nevertheless, for most rivers, the proportion of time gaps >30 days is <10%. Consequently, we argue that SA-based regional gauging is a promising method to enhance discharge monitoring across different river widths and discharge scales and provide reliable data for hydrological research and policy-making.
What problem does this paper attempt to address?