张懿,陈军辉,唐斌雁,范武波,魏永恒,向卫国,金晨阳,钱骏,刘政.基于两种再分析资料的一次四川盆地大气污染过程气象要素数值模拟研究[J].环境科学学报,2020,40(9):3093-3102
基于两种再分析资料的一次四川盆地大气污染过程气象要素数值模拟研究
- Numerical simulation of meteorological elements of a pollution episode in Sichuan Basin based on two reanalysis datasets
- 基金项目：国家重点研发计划（No.2018YFC0214002，2016YFC0206005，2017YFC0212106）；四川省重大科技专项（No.2018SZDZX0023）
- 张懿
- 四川省生态环境科学研究院, 成都 610041
- 陈军辉
- 四川省生态环境科学研究院, 成都 610041
- 唐斌雁
- 四川省生态环境科学研究院, 成都 610041
- 范武波
- 四川省生态环境科学研究院, 成都 610041
- 魏永恒
- 成都信息工程大学大气科学学院, 高原大气与环境四川省重点实验室, 成都 610225
- 向卫国
- 成都信息工程大学大气科学学院, 高原大气与环境四川省重点实验室, 成都 610225
- 金晨阳
- 四川省生态环境科学研究院, 成都 610041
- 钱骏
- 四川省生态环境科学研究院, 成都 610041
- 刘政
- 四川省生态环境科学研究院, 成都 610041
- 摘要：利用ECMWF-ERA5和NCEP-FNL再分析资料作为中尺度气象模式WRF（The Weather Research and Forecasting）初始场，对四川盆地2018年1月一次大气污染过程气象要素进行了模拟，对比分析了气温、风速、风向、相对湿度、边界层高度、温廓线的模拟效果，并结合大气超级站观测数据对模拟结果进行评估.结果表明：两种资料均能较好地模拟出气象要素的变化情况，但由于两套资料时空分辨率、采用的模式、同化方案、数据来源和质量控制方案存在一定区别，导致各要素模拟效果并不一致.与NCEP-FNL相比，ECMWF-ERA5模拟的平均相对湿度（59.23%）与观测值差异更小，且均方根误差、偏差较小，分别为9.83%和-0.83%，但NCEP-FNL模拟的平均气温（8.99℃）更接近观测值，且偏差值较小，为-0.04℃.两组模拟结果均显示盆地内部为模拟区域的低风速区，相对湿度模拟值在60%以上，气温高于西部山地地区.NCEP-FNL模拟的盆地内部气温、相对湿度、风速小于ECMWF-ERA5模拟值，但边界层高度模拟值较大.ECMWF-ERA5模拟的逆温强度相比较小，且温度露点差较小.此次污染过程PM2.5和PM10日均浓度最大值分别为190.1 μg·m-3和261.0 μg·m-3，相对湿度增大引发的颗粒物吸湿增长是导致PM2.5和PM10质量浓度突增的主要原因.
- Abstract：The Weather Research and Forecasting (WRF) model was employed to simulate the meteorological elements of a pollution episode in Sichuan Basin in January 2018. The ECMWF-ERA5 and NCEP-FNL reanalysis datasets were collected to set up the initial condition in the WRF model. The comparison was accomplished upon air temperature, wind speed and direction, relative humidity, boundary layer height, and temperature profile produced by two simulations. The data observed by atmospheric super site was utilized to evaluate the simulation performance. The results indicated that there was substantial agreement between the results of the simulations and observations in terms of meteorological conditions. However, there were differences between the two simulation performances which caused by the different temporal and spatial resolutions, the different assimilation models and schemes, the different observed data and quality control methods. The average relative humidity simulated by the ECMWF-ERA5 was 59.23% with a better observational consistency compared with the NCEP-FNL experiment. Moreover, the ECMWF-ERA5 had the more promising root mean square error (RMSE) and bias, which were 9.83% and -0.83%, respectively. However, the average temperature produced by the NCEP-FNL was 8.99℃ with a lower bias of -0.04℃, which was closer to the observations. The low wind speed region corresponded to the interior of the basin, where the simulated temperature was higher than that in the western mountainous region, and the relative humidity was approximately above 60% according to both of the simulations. In the interior of the basin, the temperature, relative humidity and wind speed produced by the NCEP-FNL were lower compared with the ECMWF-ERA5, while with a higher boundary layer. The ECMWF-ERA5 simulated lower inversion intensity and temperature-dew point deficit., The maximum daily mass concentrations of PM2.5 and PM10 were up to 190.1 and 261.0 μg·m-3 during this pollution episode. Apparently, a conclusion that the increase of the relative humidity was the main factor for the explosive growth of PM2.5, PM10 was reached.
