在WRF模拟中，默认的土地利用数据与实际土地利用情况差异较大，因此会影响模式的模拟效果。为此，许多学者提出了更新城市土地利用数据的方案。最简单的方法是仅就城市建成区面积进行修正。但因城市地表具有非均匀性，进而又提出了将建成区进一步精细化分类。然而，在研究土地利用资料对WRF模式影响的文献中，绝大多数研究仅是就某种资料更新前后的模拟效果进行比较，并未将城市面积改变、城市非均匀性这两个因子进行区分。本文综合考虑了面积修正与精细化分类这两个因子，根据面积修正方案和两种精细化方案生成了3种土地利用的优化数据，并结合默认土地数据共设置了4个算例对上海市2018年8月和2019年8月两次高温天气过程进行了模拟，通过对结果进行比较分析发现：1）对WRF土地利用数据进行优化后，改善了温度、相对湿度和风速的模拟效果。2）城市建成区面积是影响温度最关键的因子，面积修正使温度的平均均方根误差（RMSE）降低了0.86℃，在此基础上的精细化分类使平均RMSE最多降低了0.04℃。3）城市的精细化分类是影响风速和相对湿度的主要因子，面积修正使风速的平均RMSE仅降低0.04 m/s，而精细化分类可使其RMSE再进一步降低最多0.19 m/s；面积修正使相对湿度的平均RMSE仅降低0.23%，而精细化分类可使其RMSE再进一步降低最多2.25%。4）总体说来，精细化分类方案在一定程度上考虑了城市的非均匀性，因此对于温度、相对湿度和风速模拟结果的改善程度更大，且分类越细致，效果越好。
The default land use data used in the Weather Research and Forecasting Model (WRF) differ significantly from the actual land use situation, which affects the simulation results. For this reason, many researchers have proposed schemes for updating land use data prior to running the model. The simplest method involved correcting the size of the urban area. Due to the heterogeneity of the urban surface, it has been suggested that the urban landscape be subdivided into refined classification areas. However, in the literature on the impact of land use data on the WRF model, most studies have only compared the simulation results before and after data updating, and have not distinguished the two factors of changes in the urban areas and urban heterogeneity. In this paper, the authors consider the use of urban area correction and refined classification synthetically. In the area correction scheme and the two refined classification schemes, three kinds of optimized land use data are generated. Combined with default land use data, the authors established four cases to simulate two high temperature weather processes that occurred in August 2018 and August 2019 in Shanghai. The results of these two simulations are: 1) The simulation results for temperature, relative humidity, and wind speed were improved after the land use data in the WRF model had been updated. 2) The size of an urban area is the most critical factor affecting the temperature. The area correction reduces reduced the average root mean square error (RMSE) by 0.86℃, but the refined classification reduced the average RMSE by just 0.04℃ at most. 3) The refined classification method primarily affected the wind speed and relative humidity. Although area correction reduced the average RMSE of wind speed by just 0.04 m/s, the refined classification method further reduced the RMSE by up to 0.19 m/s. The mean RMSE of relative humidity was reduced by just 0.23% by area correction, while the maximum RMSE was reduced by 2.25% by refined classification. 4) Generally speaking, to some extent, the heterogeneity of a city is considered in refined classification schemes, so the simulation results for temperature, relative humidity, and wind speed are improved to a greater extent, and the more detailed is the classification, the better is the effect.
胡婧婷,陈亮.2020.上海市土地利用资料优化方案对WRF模式模拟高温过程的影响[J].气候与环境研究,25(4):443-456. HU Jingting, CHEN Liang.2020. Influence of Land Use Data Optimization Schemes on WRF Model Simulations of High Temperature Processes in Shanghai[J]. Climatic and Environmental Research (in Chinese],25(4):443-456.复制