Uncertainty analysis of elevation data based on groundwater modeling system: A case study of Zhaozhou County, China

This study takes Zhaozhou County in the Songnen Plain (a plain area with sparse borehole data) as the study area, and aims to address the issue of elevation data sources' impact on groundwater numerical simulation by comparatively analyzing the application effects of the Shuttle Radar Topography Mission Digital Elevation Model 90M (SRTM DEM 90M) and interpolated elevation derived from 25 actual boreholes and 14 virtual boreholes (synthetic borehole points derived from DEM data and corrected using nearby measured boreholes). The results show that: The elevation of SRTM DEM 90M can reflect regional microtopographic features; borehole interpolation has single-point errors (−5 m to 10 m) and interpolation uncertainty, and the over-smoothed terrain leads to a relative deviation of 40%–50% in unsaturated zone thickness in specific areas, which further results in a maximum reduction of 93% in evaporation (from 93.92 m³/d to 6.34 m³/d). Regarding simulation accuracy, the traditional water level dynamic fitting of both models meets the required standards, but the Nash-Sutcliffe Efficiency (NSE) value of water level fitting for the borehole model is negative, while the statistical indicators of the SRTM model meet the standards. This study confirms that: (1) the "parameter combination effect" plays a dominant role in simulation: Simulation results are essentially the integrated product of multi-parameter interaction; (2) in borehole-sparse areas similar to the Songnen Plain, SRTM DEM 90M can replace borehole interpolation.