| 基于源反演和气溶胶同化方法天津空气质量模式预报能力改进 |
| 摘要点击 1843 全文点击 601 投稿时间：2021-09-29 修订日期：2021-10-09 |
| 查看HTML全文
查看全文 查看/发表评论 下载PDF阅读器 |
| 中文关键词 空气质量数值模式 源反演 气溶胶三维变分同化 天津 天气背景 |
| 英文关键词 air quality numerical model source inversion aerosol three-dimensional variational assimilation Tianjin weather background |
| 作者 | 单位 | E-mail | | 蔡子颖 | 天津市环境气象中心, 天津 300074
中国气象局-南开大学大气环境与健康研究联合实验室, 天津 300074 | 120078030@163.com | | 唐邈 | 天津市生态环境监测中心, 天津 300074 | | | 肖致美 | 天津市生态环境监测中心, 天津 300074 | | | 杨旭 | 天津市环境气象中心, 天津 300074
中国气象局-南开大学大气环境与健康研究联合实验室, 天津 300074 | | | 朱玉强 | 天津市环境气象中心, 天津 300074
中国气象局-南开大学大气环境与健康研究联合实验室, 天津 300074 | | | 韩素芹 | 天津市环境气象中心, 天津 300074
中国气象局-南开大学大气环境与健康研究联合实验室, 天津 300074 | | | 徐虹 | 天津市生态环境监测中心, 天津 300074 | | | 邱晓滨 | 天津市气象科学研究所, 天津 300074 | |
|
| 中文摘要 |
| 为提升天津空气质量数值模式精细化预报能力，基于高分辨率排放源清单，技术应用源反演技术和气溶胶三维变分同化方法开展2020年天津空气质量数值预测分析，评估不同技术对空气质量模式预报能力改进，并结合气象因素评估模式系统性误差，以期提升天津空气质量精细化预报能力，服务分区精细化大气污染防治.结果表明，基于高分辨率排放源清单、源反演技术和气溶胶三维变分同化方法，可有效改进天津空气质量模式预报能力，调整后天津PM2.5、PM10、SO2、NO2和O3浓度预报平均偏差均在2 μg ·m-3以内，其中高分辨率排放源清单应用后PM2.5平均偏差为1.80 μg ·m-3，源反演技术和气溶胶三维变分同化技术应用后平均偏差分别为-1.45 μg ·m-3和-3.98 μg ·m-3，均显著小于原模式的18.75 μg ·m-3；PM2.5浓度预报和实况的相关系数，基于高分辨率排放源清单、源反演技术和气溶胶三维变分同化分别提高至0.77、0.80和0.92，相对误差分别下降至33.71%、30.62%和21.91%，空间差异有效预报天数提高至145、175和360 d，优于原空气质量模式PM2.5浓度预报相关系数的0.73、相对误差的35.66%和空间差异有效预报天数的100 d.其中气溶胶三维变分同化技术将天津中-重度过程预报TS评分由0.46提升至0.72，重污染过程预报TS评分由0.60提升至0.80.方法改进后天津空气质量数值模式预报仍存在一定系统性误差，呈现低污染时预报偏高，高浓度时预报偏低；低相对湿度时预报偏高，高相对湿度时预报偏低；低风速时预报偏低，高风速时预报偏高的特征，尤其锋前低压和低压槽天气时PM2.5浓度预报值比实况显著偏低，可根据上述特征进行系统性调整，进一步提升空气质量数值模式预报准确性，精细服务天津大气污染防治. |
| 英文摘要 |
| In order to improve the fine prediction ability of the Tianjin air quality numerical model, based on a high-resolution emission source list, source inversion technology and an aerosol three-dimensional variational assimilation method were applied to carry out the numerical prediction and analysis of Tianjin air quality in 2020. We evaluated the improvement of the air quality model prediction ability of different technologies, combined with meteorological factor assessment model systematic error, in order to improve the Tianjin air quality fine prediction ability. The results showed that the prediction ability of the Tianjin air quality model could be effectively improved based on a high-resolution emission source inventory, a source inversion technique, and an aerosol three-dimensional variational assimilation method. After adjustment, the average deviations in the mass concentration prediction of PM2.5, PM10, SO2, NO2, and O3 were all less than 2 μg·m-3. Based on the high-resolution emission source inventory, source inversion technique, and aerosol three-dimensional variational assimilation, the correlation coefficients of PM2.5 mass concentration prediction and observation were increased to 0.77, 0.80, and 0.92, respectively; the relative errors were reduced to 33.71%, 30.62%, and 21.91%, respectively; and the effective forecast days of spatial difference were increased to 145, 175, and 360 days, respectively. This was better than the correlation coefficient of the initial air quality model regarding PM2.5 mass concentration forecast; the relative error was 35.66%, and the effective forecast days of spatial difference was 100 days. The aerosol three-dimensional variational assimilation technique improved the TS score of the Tianjin moderate-severe process forecast from 0.46 to 0.72 and the heavy pollution process forecast TS score from 0.60 to 0.80. There were still some systematic errors in the prediction of the quality numerical model in Tianjin, such as high prediction at low pollution, low at high concentration, high at low relative humidity, and low at high relative humidity. The prediction of PM2.5 mass concentration was lower at low wind speed and higher at high wind speed. In particular, the predicted value of PM2.5 mass concentration in pre-frontal low pressure and low-pressure trough weather was significantly lower than that in reality, which could be systematically adjusted according to the above characteristics to improve the prediction accuracy of the air quality numerical model. |
|
|
|
