Improved 1-D hydrodynamic simulation of reservoir release based on coarse resolution topography data using ISS-SRC-assisted ensemble Kalman filter

Abstract

Evaluating the risk of flows related to reservoir releases requires a comprehensive grasp of the associated hydrodynamic process. Due to the obstacle posed by bathymetry in mountainous areas, the accuracy of the hydrodynamic simulation is hindered by the absence of reliable terrain data. To evaluate both the overall parameter sensitivities (e.g., sensitivities to cross-section geometry and roughness) within the feasible domain and the varying responses induced by changes in the input forcing (e.g., inflow series), we introduce a generalized global sensitivity indicator, namely, the input space sampled standardized regression coefficients (ISS-SRC). Through mathematical derivation, we found that the parameter whose ISS-SRC index is close to zero will not be significantly updated by the EnKF. Thus, ISS-SRC can be employed to assist the EnKF by predicting the correction performance and reducing the dimensionality of identifiable factors. The results of synthetic experiments confirm the connection between the evolution of the parameter ensemble and the dynamic sensitivity indicators. The ISS-SRC-assisted EnKF is employed to modify the physical and geometrical parameters of riverbed sections in the hydrodynamic model of a real-world reservoir release case and the simulation accuracy is greatly enhanced. This success demonstrates the potential of obtaining reliable flood simulation results with only coarse-resolution topographic information.