Research On The Usage Of Ensemble Kalman Filter In Synchronously Optimizing State And Parameters In Hydrological Model

A data assimilation method is a technique which is developed to calibrate a heterogeneous hydraulic conductivity field. Hydraulic conductivity accuracy directly affects the precision of soil moisture estimation. It is the key variable to soil moisture in meteorological, hydrological, agricultural research. Soil moisture affects the surface energy fluxes, hydrological cycle, radiation balance, migration and so on. Accurate estimation of soil moisture is critical for researching and understanding energy-hydrological exchange between land surface-atmosphere.In this study, we developed an assimilation scheme based on vadose zone hydrological model HYDRUS-1D and aquifer Characterization. This assimilation scheme, which was established with Ensemble Kalman filter algorithm, can synchronously optimize the state variable and soil hydraulic parameters synchronously. In the one-dimensional soil moisture assimilation scheme, the monitoring data gained in the Linze National Agricultural Experiment Station (located in the northwest arid and semi-arid regions of the Heihe River Basin) during the wheat growing season verify the vadose zone hydrological model-based collection Kalman algorithm. Furthermore, the vadose zone hydraulic characteristics parameters obtained in the inversion of the Kalman algorithms and heuristics shuffle complexes evolutionary algorithm (SCE-UA) were compared.2D assimilation using COMSOL modeling tool creates the aquifer characterized in model test and the data compared with the initial data generating sample (true value) are random. The results show the developed data assimilation scheme of soil moisture can synchronously estimate soil moisture dynamic change and optimize soil hydraulic parameters by statistical characteristics of the state variables and parameters error through ensemble forecasting. Data assimilation scheme can correct the simulating trajectory of model (model parameters and state variables) to improve the accuracy of forecasting variables in real-time with integrating simulated state variables and observations by ensemble method, so the accuracy of forecast variables are improved.