Yeonsu Kim

Parameter Estimation and Uncertainty Assessment of a Soil Erosion Model Using Data Assimilation Method

This study aims to develop the CSEM-PF by combining Particle Filter (PF) and Catchment-scale Soil Erosion Model (CSEM) for time-variant parameter estimation and uncertainty assessment. The CSEM-PF can provide information regarding uncertainty of simulation output as well as model parameters of nonlinear soil erosion models. The CSEM-PF was applied to the upper Guemho river basin with two rainfall events. Then we estimated the feasible model parameters of CSEM at each time step and generated uncertainty bounds of both hydrograph and sedigraph. The simulation results show that rainfall-runoff parameters are well estimated as well as good agreement with observed hydrograph with relatively small uncertainty bounds. On the other hand, erosion-sediment yield simulation results led to larger uncertainty bounds than hydrograph results. In particular, parameter estimation with a few sediment observations provided less uncertain (more reliable) sediment discharge than the case with inter- or extrapolated discharge-sediment relation equation.