Multi-land Surface Model, Multi-drive Field Integrated Land Surface Process Simulation Study And Results Of The Xinjiang Region

As the lower boundary of the atmosphere surfaces, land surface exchanges the material,energy and momentum with atmosphere and plays an important role in atmosphere and climate change.Xinjiang province, locating in the hinterland of the Eurasian continent, as one of the largest provinces in China, has a typical temperate continental climate, and it also has a complicated topography. Since the 1980s,in Xinjiang province,the temperature and precipitation both have increased, which is different from the changes in northern China where the temperature increased, but the precipitation decreased.So the research on the land surface processes in Xinjiang is very meaningful.In order to obtain the land surface process and show its characteristics of the land surface process, the work is organized in the following six parts.In section 1,data base on the surface parameters was built basing on the data collection stations in Xinjiang and data on global soil quality and color and vegetation coverage in China, which will be used later as a reference for evaluation of models.In section 2,a land-surface driven field showing a high resolution on time was set up using a five-degree polynomial interpolation basing on data from 99 sites in Xinjiang from 1960 to 2005.Eight inputs are required for this driven field model. Data on pressure, temperature,relative humidity, wind (U),wind (V),precipitation were taken from the Xinjiang Meteorological Bureau four-time-a-day conventional observation data, a downward long-wave radiation data was calculated according to Swinbank method, and sun radiation data was extracted from the TT.Qian global data. Missing observation data are added by averaging the several years' data of same date, different year from the same site.A model predicting 8 times a day with eight 5-degree polynomial interpolation input data was built.And further data analysis on the driven field model showed that in Xinjiang province,temperature, precipitation, humidity,downward long-wave radiation increased in recent years, pressure and short-wave radiation changed slightly, and wind velocity decreased.In section 3,the surface parameters data base and high-resolution time driven filed were applied to three land surface models, BATS,LSM, and CoLM to simulate the heat and precipitation exchanging process off-line between atmosphere and land surface.The results were analysed and the model were verified-using the observation data on soil temperature.Three different modes showed different computational results on the absorbed solar radiation,sensible heat flux,latent heat flux and Bowen ratio.The results presented in this study will provide a reference on the research on land surface models in Xingjiang.These results are also of great importance to deepen the understanding of land surface models in Xingjiang.In section 4, four different atmospheric forcing schemes are first used to drive the land-surface processes simulated by CLM2.0.Then the spatial and annual variabitlities as well as the interannual variabilities for different seasons are thoroughly analyzed with four simulated parameters, such as soil temperature at 5-cm depth, absorbed solar radiation, sensible heat flux, and soil moisture.Spatial distribution of modeled soil temperatures at 5-cm depth with Obs-Q and obs-P forcing schemes are more close to observed features than the other two schemes, reasonably illustrating the logical impacts due to the special topography in Xinjiang Province.There is a similar spatial pattern of modeled sensible heat flux for all four schemes,with only slightly differences in magnitude.Significant spatial variations of absorbed solar radiation are found in this study, with higher values in the south and east of Xinjiang Province.In addition,results simulated by Obs-Q and obs-P forcing schemes show better correspondence between orographic structure and absorbed solar radiation than the other two schemes.In section 5,model result evaluations are performed by comparing 7 model outputs with the soil temperature and humidity data measured in four different stations.Generally, all the model results can reproduce the temporal trend of soil temperature for 4 different sites.However, the magnitude of modeled soil temperature deviates about 5℃from observations, which suggests further improvements for the model configurations are needed,for example,adding observations to the atmospheric forcing fields, etc.Neither the multiple models nor the models with multiple forcing schemes conducted in this study can reasonably reproduce the general trend of soil moisture variations, though the magnitude between model results and observations are much close.However, the results indicate that land-surface model performance for modeling soil moisture can be improved by adding observations to the atmospheric forcing fields.In section 6,the ensembles of different simulation results are analyzed.To compare different ensemble methods, observation data from Aletai station, Wulanwusu station, Tulufan station and Shache station are used.Three kinds of ensemble methods,the mean method (MEAN),multiple linear regression (MLR),and back propagation artificial neural networks (BPANN), are applied in this work. The mean method is the one that employs simple mathematical average to the model results.While the other two methods all consider the combinations of land-surface processes and observations.To conclude, the performance of model simulations can be improved by using proper ensemble simulation methods.This study shows that MLR method does a better job than BPANN and MEAN methods both in the magnitudes and temporal trends.After gaining the confidence on MLR method, this work further analyzed the climatological responses of atmospheric parameters by using combinations of MLR method and measured soil temperature.Simulation results indicate a weak variation of absorbed solar radiation during 1960-2000.Absorbed solar radiation is found be positively correlated with short-wave radiation and atmosphere temperature, but negatively correlated with atmospheric pressure and precipitation.There is a slightly increasing trend for sensible heat flux, which is well correlated with the short-wave radiation, soil temperature and meridional wind positively. Moreover, both atmospheric pressure and latitudinal wind are found to be negatively correlated with sensible heat flux. Soil moisture increases significantly during 1960-2000, especially after 1987.Soil moisture also demonstrates a strong positive correlation with precipitation and atmospheric relative humidity, but a strong negative correlation with atmospheric pressure and latitudinal wind, as well as a weak positive correlation with atmospheric temperature, which is totally different from the climate patterns observed in most parts of China.