Study On Soil Moisture Sensitivity And Assimilation Based On Noah LSM

Soil moisture is an important factor affecting weather and climate,it is generally believed that soil moisture has a greater influence on climate than the weather,however,as an important thermodynamic factor,its influence on weather is real,specially its specific role during the development of mesoscale weather over eastern China is unclear,this makes it quite difficult for understanding the fundamentality of soil moisture from the meterological view.Numerical simulation is one of the key ways to verify the impacts of soil moisture on weather,while numerous parameters in the complex land surface model can cause great uncertainties in simulation,although auto-calibration algorithms supply important strategies for solving the land parameter-simulation problem,various obstacles during real-world application should be studied.Therefore,by using the mesoscale weather model coupled with Noah land surface scheme(WRF;Noah-WRF)and the "off-line" Noah land surface model(Noah LSM),this paper has mainly investigated the sensitivities between soil moisture and weather development over eastern China,LSM error correction,land assimilation scheme of remote sensing data and their application.Main conclusions are as follows:Based on the coupled Noah-WRF model,the sensitivities of multi-cases to soil moisture are studied,the differences of which and their possible reasons are analyzed.Results show that in the planetary boundary layer(PBL),the sensitivity of different variables to soil moisture is quite different,among which the PBL height(PBLH)is quite sensitive to soil moisture in the daytime and weak at night;in some cases,the vertical velocity(W)is relatively sensitive to soil moisture while the convection is developing,but the convective available potential energy(Cape)and the equivalent potential temperature(?se)generally show weak sensitivities to soil moisture;meanwhile,a positive feedback between soil moisture and the near-surface thermodynamic conditions can be identified;surface heating over dry soil and abundant water evapotranspiration over wet can be found.However,soil moisture has weak influence on precipitation intensity over eastern China,and only affects the patterns of the precipitation systems.Besides data and the physical process that can affect land simulation,parameter is an important error source of which,and model calibration can effectively improve model parameters,thus,based on particle swarm optimization(PSO)and shuffled complex evolution(SCE)algorithms,two schemes for calibrating Noah LSM are constructed,and further compared on their abilities in correcting parameters and simulations.Results show that PSO can converge more quickly to better fitness values than SCE and its sensible heat flux simulation is more consistent with observation than SCE,therefore,the calibrated simulation of PSO is more accurate and fast;also,it can achieve more stable and consistent calibration but with lager parameter ranges.This indicates that PSO is of higher efficiency and stability than SCE.As the accuracy of soil moisture simulation can significantly affect the weather development,based on the Ensemble Square-Root filter(EnSRF)method,an integrated assimilation scheme is constructed,and the impacts of observation and model corrections on soil moisture assimilation are compared.Results show that observation correction can effectively reduce the seasonal differences between observation and simulation,which has indirectly reduced observation errors during assimilation,while model correction has indeed reduced assimilation errors resulted from model parameters;the assimilation schemes with corrected information can outperform that without,however,the integration of either or only one correction with EnSRF cannot provide best corrected information,thus,by using both the scaled observation correction and the correction with optimized model parameters,the traditional soil moisture assimilation scheme can be greatly improved.Based on EnSRF algorithm,the assimilation experiments of the moderate-resolution imaging spectroradiometer(MODIS)land surface temperature(LST)products has been carried out in this paper finally,results show that when the localization parameter r of background covariance that slightly smaller than the model resolution,the best innovation can be achieved,meanwhile more stable spread can be achieved by using the smaller inflation parameter inf,which can obtain the best assimilation results;according to the MODIS LST assimilation,WRF LST simulations can be greatly improved,which are also mostly close to station observations.