Frozen Ground Response To Climate Change And The POD-based Reduced-order Extrapolating Model

Frozen soil is an important part of the cryosphere,not only widely distributed,but also has great influence on the process of water and energy during the phase change process.It is an important surface in the land ecosystem.The movement of the soil frost and thaw frontdepth affects the terrestrial ecological carbon cycle process as well as local and even global climate change,and is the sensitive indicator to climate change due to its sensitivity to climate.Obtaining accurate frost and thaw front depth information plays an important role in surface energy balance,alpine ecology,hydrological runoff,cold zone engineering and greenhouse gas emissions.In the actual engineering calculation,the numerical solution is an effective way to solve the partial differential equation.However,the numerical solution of the solution of the dimension of the equation is positively correlated with the number of regional split nodes.In order to obtain sufficient accuracy,the resolution area needs to be finer divided,the corresponding calculation will increase.This means that there will be a huge amount of computation in the calculation process,while the computer needs a huge memory capacity.It is important to study how to reduce the computational freedom,simplify the calculation,save the calculation time and the required memory space under the premise of ensuring sufficient precision.In order to solve the above scientific problems,this study has developed a parameterization scheme of frozen soil considering the dynamic change of frost and thaw front depth,and coupled it with land surface model CLM4.5,and then used the new model to reveal the response of frozen soil to global climate change.At the same time,the POD method is used to construct the reduced order finite difference model.The POD base is constructed by using the numerical results of the initial simulation for a short period of time,and then the simulation of the future period is extrapolated.This avoided the shortcomings of the existing POD reduction model.The main conclusions of this study are as follows:(1)The maximum soil freezing depth change process in Inner Mongolia was analyzed using 117 site observations.With the change of temperature,the maximum soil freezing depth in Inner Mongolia has obvious seasonal variation,and the permafrost mainly appears in winter and spring.And with the increase of latitude,the depth of permafrost has a lag for the temperature response.The permafrost in Inner Mongolia has a tendency to increase first and then decrease from west to east,and the maximum value appears in the Hulunbuir grassland and Xilingol grassland,and the extreme value is more than 260 cm.At the same time,the spatial distribution of the maximum permafrost and the spatial distribution of air temperature reflect a high correlation.The correlation coefficients of year,winter and spring were-0.819,-0.782,-0.787 respectively.In the past 36 years,the maximum soil freezing depth in Inner Mongolia showed a decreasing trend.Correspondingly,the average temperature in Inner Mongolia is on the rise.These conclusions suggest that soil freezing depth is closely related to climate change.Especially in some ecologically fragile permafrost regions,small climate change may lead to greater impact.This further proves the necessity of studying the spatial and temporal changes of soil freezing and melting.(2)The two-directional Stefan method is coupled with the vertical one-dimensional soil thermal diffusion equation,and a series of sensitivity experiments and experimental verification of the D66 observation site of the Qinghai-Tibet Plateau are carried out.The results show that the new model can simulate the dynamic change of freeze-thaw interface accurately.At the same time,the new model can also improve the simulation of soil temperature.This is because in the coupling process,the frost thaw front position is seen as 0 ? isotherm depth,thus updating the vertical distribution of soil temperature profile.Land surface model CLM4.5 also applies the thermal diffusion equation to simulate the vertical stratified soil temperature.Therefore,the above experimental results provide a theoretical basis for the new frozen soil parameterization scheme in CLM4.5 coupled with frost and thaw front change.(3)Using the two-directional Stefan frost and thaw front simulation method and CLM4.5's own algorithm,the new land surface model CLM4.5_FTF considering the freeze-thaw interface is developed.In order to evaluate the simulation capability of the new model,we simulated the daily frost thaw fronts at the Hulugou site in the seasonally frozen ground area and D66 site in the permafrost region.At the same time,the active layer of the permafrost region in the northern hemisphere was simulated on the interannual scale,and the distribution of different permafrost types in China was simulated.The results are also consistent with the observed data.This proves the rationality and effectiveness of the new development model.On this basis,we use this model to simulate the spatial distribution and trend of global permafrost in 1970-2010.The results show that the thickness of the active layer in the permafrost region is gradually deepened under the background of global warming,and the maximum freezing depth of the seasonal frozen soil is decreasing in the black sea and the surrounding area of the Caspian Sea.Further research results show that the simulation results are reasonable due to the deepening of the freezing depth due to the decrease in winter air temperature in the region.(4)We construct a new reduced-order extrapolation finite difference model by combining the POD reduction method with the classical finite difference scheme of the two-dimensional shallow water equation containing the sediment concentration and the classical FDTD scheme of the Maxwell equation.In order to solve the problem of repetitive calculation,which is usually existed in the existing POD reduction model,this study uses the numerical results of the initial simulation for a short period of time to calculate the POD basis,and then constructs the reduced order model and then extrapolates the simulation.The study proves the error estimates of the newly developed two POD reduced-order difference model numerical results,based on which the number of POD bases is selected and whether the POD base is replaced.For each new model,we prove that the POD reduced-order extrapolation difference model is very effective in solving the problem of shallow water equation containing sediment concentration and the Maxwell's equation,by two numerical examplesrespectively.