Parameter Estimation Of The Hydrological Model Using Satellite-based Soil Moisture Data

The increased availability of remote sensing products provides the opportunity to introduce these data sources into hydrological models.Previous studies have shown the integration of satellite-based data with hydrological modelling improved the model forecasting skill by data assimilation or model calibration.Although the approaches of data assimilation and hydrological model calibration are able to utilize the temporal variability information of the soil moisture data over the whole catchment,they fail in fully exploiting the spatial information hidden in the soil moisture data.In this study,we assessed the feasibility of another approach that uses satellite-based soil moisture data to directly estimate the parameter in the hydrological model.Parameters of hydrological models are generally related to specific basin characteristics.The approach of the parameter estimation might be another potential approach to exploit the added value of spatial information provided by the soil moisture data.The lumped and semi-distributed Hymod models were chosen and case studies were carried out at four study regions,i.e.the Xiang River basin,and the Gan River basin,the Fu River basin,and the Zi River basin.The main contents and findings of this study are summarized as follows:(1)The Soil Moisture Active Passive(SMAP)root-zone soil moisture data were used to yield the samples for the soil moisture storage capacity among each basin,and the spatial distribution of the soil moisture storage capacity is obtained.The caparison between empirical distribution and the spatial distribution of soil moisture storage capacity showed that the observations did not follow the Pareto distribution.Therefore,this study proposed to use the Lognormal distribution to fit the soil moisture storage capacity.Comparing the Hymod models using different storage capacity distributions,the simulation results showed that the Hymod model using Lognormal distribution as the capacity distribution performed better in terms of NSE and KGE values than the original Hymod in the calibration period and validation period.The streamflow simulation in the Zi River Basin has been improved most,with the NSE value increasing by 3.1%.By comparing the results of flow duration curves,it is found that the Hymod model using Lognormal had better simulation at the medium and low flow portions of the hydrograph.At high flows,the Hymod model using Pareto distribution is better.The simulation results indicated that the Lognormal distribution is more suitable than Pareto distribution to describe the spatial distribution characteristics of soil moisture storage capacity among the study basins.(2)The soil moisture storage capacity of each grid in the basin calculated using SMAP soil moisture data can be used as a sample of the spatial distribution of storage capacity.Therefore,the parameter of the storage capacity curve can be estimated by the method of moments,and then the calculated parameter values can be brought into the hydrological model.In the Gan River basin and the Xiang River basin,comparisons are made with the Hymod model using all the parameters obtained through calibration.The results showed that the models using the prior estimation of the parameter were able to produce consistent streamflow predictions in study catchments in terms of NSE/KGE and the prediction uncertainties are smaller.(3)A relationship was established between the total water storage capacity calculated by the Hymod model and the observed total water storage capacity,so that the corresponding parameter values in the ungauged basin can be calculated based on the parameters calibrated in the donor basin.The Gan River Basin and Xiang River Basin are used as donor basins,and their neighboring basins are used as receiver basins.Compared with the parameters directly transplanted from the considered watershed,the calculated parameters of the ungauged basin calculated by this method have a better simulation effect in the ungauged basin,indicating that the use of observational data to establish a connection with the model parameters can effectively estimate the parameter value of the area without data.