Presenter
Miss. Chen Li, Tsinghua University
Co-author(s)
Prof. Di Long, Tsinghua University
Mr. Yiming Wang, Tsinghua University
Dr. Tejasvi Hora, University of Waterloo
Ms. Esha Zaveri, The World Bank
Ms. Aude-Sophie Rodella, The World Bank
Sub-theme
4. Supporting Aquatic Ecosystem Health and Functions
Topic
4-7. Groundwater and ecosystem
Body
The Haihe River Basin (HRB) has experienced severe groundwater overexploitation over the past four decades. Accurate monitoring of groundwater storage (GWS) changes is critical for aquatic ecosystem protection and water resource management. Traditional approaches using networks of observation wells to calculate GWS may be subject to spatiotemporal variability and inaccuracy in parameters associated with aquifers such as specific yield and storage coefficients. Gravity Recovery and Climate Experiment (GRACE) satellites launched in 2002 and its follow-on mission launched in 2018 have provided a revolutionary way for groundwater monitoring globally and regionally. The GRACE mascon product provided by the Jet Propulsion Laboratory (JPL) shows that GWS in the HRB has declined at a rate of -1.97 cm/year (i.e., 6.26 km3/yr) from 2003 to 2021 (p<0.05). Because of the South-to-North Water Diversion and climatic variability that induced increases precipitation, declining GWS trends in the HRB have been slowed down in recent years, and even got largely recovered in 2021 due to higher than normal precipitation of ~ 830 mm/yr. However, this recovery is not sufficient to compensate for overexploitation. Groundwater overexploitation in the HRB still needs considerable attention. Due to relatively low native resolution of GRACE (~ 90,000 km2, 3°×3°), spatial downscaling is required when focusing on smaller-scale variability. To figure out GWS changes at finer spatial scales, we implement a statistical downscaling approach using Random Forests (RF) to obtain finer GRACE GWS (0.25°×0.25°) over the HRB. Hydrological fluxes and its accumulation, including precipitation, ET, runoff, and SM, NDVI are used as predictors. Results show that the RF model perform well in terms of a correlation coefficient (CC) of 0.98 and a root mean square error (RMSE) of 2.25 cm for the validation set. From the downscaled results, GWS changes over the Daqing River basin and Ziya River basin where marked groundwater overexploitation has been derived. GWS in these two sub-basins shows a significant decreasing trend (p<0.05), with rates of -1.68 cm/yr (i.e., 0.76 km3/yr) and -1.95 cm/yr (i.e., 0.92 km3/yr), respectively. The findings from this study provide a better understanding of GWS changes in the HRB and its sub-basins, and can offer reference for formulating local groundwater management strategies in the future.