摘要:近年来,青藏高原多年冻土区生态环境呈现出逐年恶化趋势,从而对多年冻土活动层水热过程造成显著影响。此外,如何构建更加有效、针对寒区的陆面过程模式成为寒区研究的重点、热点之一。作为一种有效的参数估计方法,Bayes参数估计算法具有准确估计陆面过程模式参数的能力。因此,基于2005--2008年观测数据,利用CoupModel模型对青藏高原风火山流域土壤水热运移过程进行模拟;同时,利用Bayes参数估计方法估计部分水热运移参数。结果显示:模型对土壤温度(ST)的模拟效果较好,NSE系数均在0.90以上;也能够较好模拟浅层(0~40cm)土壤水分,NSE值均达到0.80以上,而对40cm以下土壤水分的模拟结果较差。模型也能够较准确模拟活动层土壤的冻结一融化过程。模型对温度水分极值和40cm深度以下水分的模拟存在一些偏差。值得一提的是,基于重要性采样MCMC方案的Bayes参数估计算法能够有效估计水热运移参数,模型模拟结果得到极大的改进。Bayes算法能够广泛解决陆面过程模式参数估计问题。

In recent years, ecological environment is deteriorating in permafrost regions of the Tibetan Plateau, which dramatically affects the water and heat processes between the earth and atmosphere. In addition, it is a key and hot research field in cold regions how to structure more effective land surface process mode. In Fenghuoshan basin of the plateau, the water and heat processes were simulated by using a land surface model CoupModel, on the basis of observed data sets from 2005 to 2008. The simulation reveals that the model is able to simulate soil temperatures well and capture all soil layer temperature fluctuation trends and all the Nash-Sutcliffe Model Efficiency Coefficient values between observed and simulated soil temperatures are over 0. 90. The mean error values display that the model overestimates the temperatures at 20~55 cm depths and underestimates the temperatures at 65~120 cm depths. The minimum root mean square error value is at 55 cm depth, where temperature fluctuation is less than that at the surface layers. Meanwhile, the model has the capacity of simulating soil water content in shallow layers (0 ~40 cm), where Nash-Sutcliffe Model Efficiency Coefficient value is between observed and simulated soil water content, above 0. 80. However, simulated soil water content is not very good below 40 cm depth, perhaps due to sharp slope. The model can also simulate preferably the freeze-thaw processes within the active layer. The simulation of the soil water content bellow 40 cm depth has a little deviation, and so does the simulation of extreme values of temperature and water content. What's more, as an effective parameter estimation method, Bayesian parameter estimation method has the ability to evaluate model parameters. In this research, parameters include those of meteorology, soil freezing, soil thermal transport, frozen soil and hydraulics. Thus, the simulation will be greatly improved. The Bayesian algorithm can be utilized to complete parameter estimation.