论文部分内容阅读
针对不同陆面模式对土壤湿度方程求解方法以及对土壤分层结构的差异,本文选取了三类陆面模式(CLM,CABLE和ECMWF陆面模式)作比较研究。为了避免不同模式参数化方案引起的上边界误差,上边界采用固定蒸发率、入渗率和固定表层土壤湿度三类边界条件。土壤分层采用101层(细网格)和11层(粗网格)两种,并考虑土壤性质沿深度变化。结果表明:当土壤性质均匀时,求解的差别主要在第三类边界条件下CLM求出的水分入渗速度比其它两种快;改用粗网格后由于土壤深层厚度加大无法与细网格得出的土壤湿度廓线相重合。当土壤性质非均匀时,模拟结果间差别加大,只有ECMWF模式模拟的土壤湿度廓线是严格连续的。对于模式和上边界的不同组合,粗、细网格模拟结果间均方根偏差不一致。一般而言,CABLE模拟的偏差除第一类条件较小外,其它都是最大的。第二三类边界条件引起的偏差较大,第一类最小。上述结果提示我们,在比较不同陆面模式以及用观测资料来检验模拟结果时应充分考虑土壤分层及土壤性质非均匀性的可能影响。
According to the method of solving the equation of soil moisture for different land surface models and the difference of the soil layer structure, this paper selects three types of land surface models (CLM, CABLE and ECMWF land surface models) for comparative study. In order to avoid the upper boundary error caused by different mode parameterization schemes, the upper boundary adopts three kinds of boundary conditions: fixed evaporation rate, infiltration rate and fixed surface soil moisture. Soil stratification using 101 (fine grid) and 11 layers (coarse grid) two, and consider the soil properties along the depth changes. The results show that when the soil properties are uniform, the difference between the two solutions is faster than the other two under the condition of the third type of boundary conditions. Because of the thicker soil depth, The resulting soil moisture profiles coincide. When the soil properties are not uniform, the differences between simulation results increase. Only the soil moisture profile simulated by ECMWF model is strictly continuous. For different combinations of modes and upper bounds, root mean square deviations between coarse and fine grid simulations are inconsistent. In general, the CABLE simulation deviations are the largest except for the first type. The second three types of boundary conditions caused greater deviation, the first category is the smallest. The above results suggest that we should give full consideration to the possible influence of soil stratification and soil heterogeneity when comparing different land surface patterns and using the observed data to test the simulation results.