Recent advances in computer with geographic information system(GIS) technologies have allowed modelers to develop physics-based models for modeling soil erosion processes in time and space.However, it has been widely recognized that the effect of uncertainties on model predictions may be more significant when modelers apply such models for their own modeling purposes.Sources of uncertainty involved in modeling include data, model structural, and parameter uncertainty.To deal with the uncertain parameters of a catchment-scale soil erosion model(CSEM) and assess simulation uncertainties in soil erosion, particle filtering modeling(PF) is introduced in the CSEM.The proposed method, CSEM-PF, estimates parameters of non-linear and non-Gaussian systems, such as a physics-based soil erosion model by assimilating observation data such as discharge and sediment discharge sequences at outlets.PF provides timevarying feasible parameter sets as well as uncertainty bounds of outputs while traditional automatic calibration techniques result in a time-invariant global optimal parameter set.CSEM-PF was applied to a small mountainous catchment of the Yongdamdam in Korea for soil erosion modeling and uncertainty assessment for three historical typhoon events.Finally, the most optimal parameter sets and uncertainty bounds of simulation of both discharge and sediment discharge at each time step of the study events are provided. Recent advances in computer with geographic information system (GIS) technologies have allowed modelers to develop physics-based models for modeling soil erosion processes in time and space. Yet, it has been widely recognized that the effect of uncertainties on model predictions may be more significant when modelers apply such models for their own modeling purposes. Sources of uncertainty involved in modeling include data, model structural, and parameter uncertainty. To deal with the uncertainty parameters of a catchment-scale soil erosion model (CSEM) and assess simulation uncertainties in soil erosion, particle filtering modeling (PF) is introduced in the CSEM. proposed method, CSEM-PF, estimates parameters of non-linear and non-Gaussian systems, such as a physics-based soil erosion model by assimilating observation data such as discharge and sediment discharge sequences at outlets. PF provide timevarying feasible parameter sets as well as uncertainty bounds of outputs while tradition al automatic calibration techniques result in a time-invariant global optimal parameter set. CSEM-PF was applied to a small mountainous catchment of the Yongdamdam in Korea for soil erosion modeling and uncertainty assessment for three historical typhoon events. Finally, the most optimal parameter sets and uncertainty bounds of simulation of both discharge and sediment discharge at each time step of the study events are provided.