Effects Of Soil Water And Heat Processes On Regional Weather In Central And Eastern Inner Mongolia Grassland

The Central and eastern Inner Mongolia grassland is an important ecological barrier area in northern China and locates in state key ecological function area.In recent years,under the background of climate change,regional droughts have occurred frequently,and animal husbandry,social and economic development and grassland ecological environment have been severely affected.Soil water and heat processes are important processes that affect the growth of grassland vegetation,changes in the ecological environment,local water-energy cycles,and weather and climate changes.As it is relatively difficult to accurately obtain soil water and heat data in large areas,this restricts the monitoring of grassland droughts,land-atmosphere interactions research and other research progresses.Using land surface hydrological model and atmospheric model WRF/WRF-Hydro to simulate soil water and heat processes is of essential to understand the land surface hydrological cycle,land-atmosphere interactions and grassland drought disaster monitoring and early warning.This dissertation takes the Central and eastern Inner Mongolia grassland as the study area,using the observation data of the national weather station in the study area and multi-source assimilated data to evaluate multiple sets of precipitation,soil moisture,soil temperature and snow depth gridded datasets,investigating the influence of different soil texture data on soil water and heat processes simulated by land surface hydrological model CLM3.5,evaluating soil water and heat simulated by WRF-Hydro,improving land-atmosphere coupling simulation and reforecast capacity of snow of WRF/WRF-Hydro model by initialization of soil temperature and soil moisture were studied.The main research conclusions of this study are presented as follows:(1)Based on the monthly precipitation data observed by the National Meteorological Station in the Central and eastern Inner Mongolia grassland from 1982 to2018,and the statistical indexes such as correlation coefficient,absolute average deviation and root mean square error,etc.,the applicability of 5 gridded precipitation datasets is evaluated.The results show that the spatial distribution of precipitation from CRA40 and FLDAS are the most consistent with that of observed precipitation,and GLDAS is relatively poor.The time correlation coefficients of 5 gridded precipitation datasets are all higher than 0.96,the statistical index of CRA40 is the best,FLDAS is the second,Era5 land is the third.The simulation ability of precipitation datasets in winter is worse than that in other seasons,and the quality in arid area is worse than that in semi-humid area and semi-arid area,and FLDAS is more suitable in study area.(2)Using land surface hydrological model CLM3.5 and four sets of soil texture data from different sources,4 integration experiments were conducted over the study area,a42-year high spatial and temporal resolution land surface variable reference dataset was developed.The influence and mechanism of soil texture on the simulation of soil water-heat of land surface hydrological model are analyzed and investigated.Soil water and heat processes are affected by soil texture obviously.Soil texture mainly affects soil infiltration,surface runoff and underground runoff by changing soil saturated water content and soil saturated hydraulic conductivity.That ultimately leads to changes in soil moisture and soil temperature.The evaluated results show that the IMRA42-BNU dataset based on CLM3.5 land surface hydrological model and BNU soil texture data can better simulate the soil hydrological variables in the study area,but has less effect on the soil thermodynamic variables.This is mainly due to the fact that the BNU soil texture data contain a relatively high proportion of sand,so that the soil moisture and soil temperature simulated by CLM3.5 are optimal.Compared with IMRA42-FAO,the correlation coefficient of 0?10cm soil moisture simulated by IMRA42-BNU increased by 9.68%,the root mean square error decreased by 28.77%,and the RMSE of 5cm soil temperature decreased by 49.62%.The IMAR42-BNU reference dataset was compared with 7 grid soil moisture datasets and 4 grid soil temperature datasets at home and abroad.The results showed that IMAR42-BNU simulated 0?10 cm soil moisture and 5 cm soil temperature were superior.For soil moisture,IMRA42-BNU and SMAP have the best spatial distribution and statistical index.Compared with SMAP soil moisture,the correlation coefficient of IMRA42-BNU increased by 6.58%,the unbiased root mean square error decreased by 35.67%.For soil temperature,the IMRA42-BNU,SMAP and ERA5 Land perform better.And their correlation coefficients are all higher than 0.99,the deviation is small.The performance of GLDAS-NOAH is the worst.(3)The NCAR new generation land surface hydrological model WRF-Hydro which was driven by the ERA5 Land land surface reanalysis datasets have been used to make long time simulation in the Central and eastern Inner Mongolia grassland.A land surface hydrological simulation system was built and generated a land surface hydrological dataset with high temporal and spatial resolution of year 2016?2020.The simulation capability of WRF-Hydro model for soil moisture and soil temperature is analyzed.The accuracy of WRF-Hydro simulation of soil moisture and soil temperature was evaluated using observed daily soil moisture data measured from May to September in 2020 and daily soil temperature data in 2020,and compared with the IMRA42-BNU dataset.The results show that WRF-Hydro can better simulate the spatial distribution of soil moisture and soil temperature with higher correlation coefficient and smaller bias.Compared with IMRA42-BNU,WRF-Hydro has good simulation ability of soil water and heat,especially for 5cm soil temperature in summer,and the correlation coefficients of soil temperature and soil moisture from WRF-Hydro and IMRA42-BNU are very high with small difference.The mean deviation of 0?10cm soil moisture and 5cm soil temperature simulated by WRF-Hydro decreased by 24.40% and 7% respectively.The WRF-Hydro model with improved hydrological module is superior to CLM3.5 in simulating soil water and heat in the study area.(4)The coupling intensity of soil moisture-precipitation is calculated by using IMRA42-BNU soil moisture and CMFD precipitation data,and the Central and eastern Inner Mongolia grassland in summer and autumn is identified as a “Hot region” of land-atmosphere interactions,in winter and spring,the land-atmosphere coupling is weak.It is clear that there is a positive feedback between soil moisture and precipitation in the study area,but it varies obviously in seasons.Improving the initial field of soil temperature and soil moisture can improve the ability of WRF/WRF-hydro model to forecast the precipitation,air temperature and specific humidity in the Central and eastern Inner Mongolia grassland,the air temperature deviation was also reduced,and the spatial distribution of 2 m specific humidity was consistent with the multi-source fusion data.The prediction ability of soil moisture and soil temperature were also improved.The improvement of soil moisture and soil temperature improved precipitation forecast by 65.38% and 57.69% respectively.The feedback mechanism of precipitation is clarified: when the soil moisture increases,the cloudiness and precipitation increase,which is a positive feedback;when the soil moisture decreases,the cloudiness and precipitation decrease,which is a negative feedback.(5)WRF/WRF-Hydro initialized soil moisture and soil temperature simulated by WRF-Hydro was used to simulate a snowstorm in winter 2020.And the effects of soil temperature and soil moisture on snowfall and snow depth were analyzed.The results show that improving soil moisture can improve the forecasting ability of WRF/WRF-hydro model for snowfall and snow depth with 53.85% and 65.38% of the stations were improved respectively,but the improvement is not significant.The improvement of initial soil moisture field is better than the improvement of initial soil temperature field.The qualified rate of snow depth predicted by the four numerical experiments was more than 57%,and the qualified rate of WRF-SM was higher than 65%in the improved soil moisture group.The results show that there is a positive feedback relationship between soil moisture,soil temperature and snowfall,but it is obviously weaker than that in summer.That is,the soil moisture(soil temperature)increases(decreases)and snowfall slightly increases(decreases).Comparing the predicted snow depth of WRF-FNL with 9 snow depth datasets at home and abroad,WRF-FNL and IMRA42-BNU have the best accuracy and the highest qualified rate.