High Resolution Downscaling Simulation In Complex Terrain Using Spectral Nudging And Three-dimensional Variational Method To Assimilate Surface Observations

The high-resolution meteorological data produced by regional climate model can supplement meteorological information in data-sparse areas,whose data could support the study on the field of local climate change.However,how to obtain high-quality and high-resolution(such as kilometer scale)data sets,especially for complex terrain areas,has been the target of dynamic downscaling simulation.For this,this paper selected the Qinling-Daba Mountains with complex terrain as object,and carried out multi-group dynamic downscaling simulation experiments with 5km horizontal resolution using WRF model during two time periods in the winter of 2014 and in the summer of 2015.A combination of spectral nudging and three-dimensional variational methods is proposed for surface observation assimilation,which is expected to provide an effective approach to obtain high-resolution downscaling data sets in complex terrain.Firstly,four sets of eight simulation experiments were conducted with different combination of parameterized schemes,in order to obtain the optimal combination of parameterized scheme for the Qinling-Daba Mountains.Then,two groups of 4-month simulation experiments are carried out to evaluate the effect of spectral nudging method on 5kmhorizontal-resolution downscaling simulation.Finally,through two groups of surface observation assimilation experiments,the effectiveness of surface observation assimilation using the combination of spectral nudging and three-dimensional variational was verified.The main conclusions were drawn as follows:(1)Among eight combinations of parameterized schemes,the optimal simulation scheme for the Qinling-Daba Mountains area is consisted of Dudhia shortwave radiation scheme,YSU boundary layer scheme,K-F cumulus convection parameterization scheme and WSM6 cloud microphysical parameter scheme.The simulations of 2m temperature and precipitation are sensitive to the different parameterization scheme,but simulated 10 m wind fields are less sensitive to parameterization schemes.The selection of shortwave radiation and cloud microphysical schemes have obvious impact on 2m temperature and precipitation.The use of Dudhia shortwave radiation scheme and WSM6 cloud microphysical scheme can get the best simulation for 2 m temperature,while Dudhia and Lin schemes can get best simulation for precipitation.(2)Spectral nudging method performs better than the continuous modelling approach for 5km-horizontal-resolution downscaling simulation in the Qinling-Daba Mountains area.Spectral nudging can introduce large-scale data constraints to correct the bias between simulated and analysis fields.As a result,the simulated geopotential height field,horizontal wind field and temperature field at 500 hPa and 200 hPa are consistent with the NCEP FNL analysis.The usage of spectral nudging can significantly improve the simulation accuracy of 2m temperature in the summer of 2015 and the winter of 2014,especially in the middle and eastern areas of the Qinling-Daba Mountainss.Furthermore,the experiment with spectral nudging method also significantly improved the simulation effect of precipitation in summer of 2015,but the simulated precipitation in winter of 2014 have an obvious deviation from observations,which is worse than the experimental results without spectral nudging.In addition,although spectral nudging method can also have a positive role on the simulation 10 m wind speed,the simulation of 10 m wind speed still has a large deviation from the observation with respect to the simulation effect of 2m temperature.(3)Two assimilation experiments with or without spectral nudging work better than spectral nudging in simulating 2m temperature,but perform worse than spectral nudging in simulating precipitation in the winter of 2014 and the summer of 2015.Further,the combination of spectral nudging and three-dimensional variational methods gets a better result in simulating 2m temperature,10 m wind fields and precipitation than three-dimensional variational method only.However,the combined method gets a worse simulation for precipitation in the winter of 2014,which maybe caused by the poor precipitation simulation from both experiment with spectral nudging and experiment with three-dimensional variational method.