基于通用陆面模式（Common Land Model, CoLM），首次评估了两套最新的全球土壤数据集GSDE（Global Soil Dataset for Earth System Model）和SG（SoilGrids）对全球陆面过程模拟的影响。比较分析了两套数据中砂粒、粘粒、砾石、有机碳的含量和容重这五个土壤属性在全球分布上的差异以及这种差异造成的对模式估计的土壤特性参数、水力热力变量的影响。结果表明，土壤特性参数在全球的空间分布主要受土壤粒径分布（砂粒、粉粒和粘粒）影响，同时也受砾石、有机碳和容重的影响。土壤资料对全球模拟结果影响主要体现在区域差异，对水文学变量的影响（Re最大达到±100%）大于对土壤热力学变量的影响（Re<±10%），对地表辐射变量的影响较小（Re<±5%）。其中，土壤体积含水量在加拿大中部及西北部、俄罗斯东南部及中西部和澳大利亚中部地区模拟结果相差较大，总径流在低纬地区模拟结果出现较大的差异，热力学变量在非洲北部、加拿大西北部以及俄罗斯中北部的差异稍大。将模拟的土壤体积含水量与站点观测相比，两套数据的表现接近，与站点观测相比都存在一定的偏差，但SG更接近观测，其中在Molly Caren站点（39°57'N，83°27'W）上SG相比GSDE整体提高约0.01～0.02。本研究表明，模式模拟结果受不同土壤数据集的影响显著，可优先考虑诸如SG较准确的土壤数据。土壤属性对陆面模拟的影响需进一步研究。
This study aims to evaluate the effect of two, new, global soil datasets on global land surface simulation, based for the first time on the Common Land Model (CoLM). The effects of the two soil datasets, namely GSDE (Global Soil Dataset for Earth System Model) and Soil Grids (SG), on the model simulation results were studied. The differences between these two data sets were compared and analyzed for five soil properties, namely sand, clay, gravel, organic carbon, and bulk density, and the impact, caused by those differences, on the estimated soil characteristic parameters as well as the hydraulic and thermal variables in the model were discussed. The results show that the global spatial distribution of soil characteristic parameters is mainly influenced by soil particle size distribution (sand, silt, and clay), and also by gravel, organic matter, and bulk density. The effect of the soil datasets on the global simulation varies across different regions. Their effect on the hydrological variables (the maximum value of Re is ±100%) is greater than that on the soil thermodynamic variables (Re<±10%) and on the surface radiation variables (Re<±5%). The soil volumetric water content in central and northwest Canada, southeastern Russia, and midwest and central Australia is quite different, and the total runoff in low latitudes area shows great variance. Thermal variables show some differences in northern Africa, northwestern Canada, and north-central Russia. Comparing the simulated soil moisture with site observations, the performance of the two datasets is similar and there is a certain deviation from the site observations. More specifically, the values based on the SG data are closer to the observation values. The results show that there is an increase of about 0.01 to 0.02 using the SG data compared with the GSDE data at the Molly Caren site. This study shows that the model simulation results are significantly affected by different datasets and that soil data with higher accuracy, such as the SG data, are preferable for model use. Further studies on the effect of soil properties on land surface modeling are required.
李文耀,魏楠,黄丽娜,上官微.2020.土壤数据集对全球陆面过程模拟的影响[J].气候与环境研究,25(5):555-574. LI Wenyao, WEI Nan, HUANG Lina, SHANGGUAN Wei.2020. Impact of Soil Datasets on the Global Simulation of Land Surface Processes[J]. Climatic and Environmental Research (in Chinese],25(5):555-574.复制