| 霾天能见度参数化方案改进及预报效果评估 |
| 摘要点击 1720 全文点击 800 投稿时间：2018-08-30 修订日期：2018-10-19 |
| 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 霾 能见度 参数化方案 数值预报 预报检验 |
| 英文关键词 haze visibility parameterization scheme numerical forecast performance test |
|
| 中文摘要 |
| 为了选择适合京津冀地区能见度预报的参数化方案，为霾的预报提供更准确的能见度预报产品.根据实际应用需求，改进了基于气溶胶体积浓度建立的Chen等能见度参数化方案（S1），并利用2017年2月快速更新多尺度分析和预报系统-化学子系统（RMAPS-CHEM v1.0）的预报结果，对比评估了该方案与基于PM2.5浓度建立的参数化方案（S2）和Mie散射计算方案（S3）在京津冀地区的预报效果.结果表明：①模式系统对京津冀地区的PM2.5浓度预报总体较好，预报值与观测值非常接近，二者的相关系数在大部分地区可达0.8以上，小时相对湿度的预报值与观测值相关在0.78以上，平均误差低于3.91%.②三套方案计算的能见度都能较好地预报出2017年2月京津冀地区能见度的时间演变趋势，且在大部分时间三者计算的能见度都非常接近.总体上S1计算的能见度最低，S3计算的能见度最高，S2居中.在京津冀大部分地区，S1的均方根误差和归一化平均绝对误差最低，S3的最高，S2居中且在北京地区表现最佳.③能见度大于10 km时三套方案计算结果都偏低，其中S3的平均误差和均方根误差最低，能见度低于10 km时，特别是对出现频率较高的1~5 km低能见度的预报，S1方案的平均误差、均方根误差和归一化平均绝对误差都是最低的，更适合京津冀地区霾天气能见度的预报应用. |
| 英文摘要 |
| In order to select a visibility parameterization scheme that can be applied well to the Beijing-Tianjin-Hebei (BTH) region and provide better forecasting, a modified parameterization of visibility based on the aerosol volume concentration and RH is developed in this study. This upgraded parameterization scheme (S1) and other schemes based on PM2.5 and RH (S2) and Mie theory (S3) are evaluated using forecast data from Rapid refresh Multi-scale Analysis & Prediction System-CHEM (RMAPS-CHEM v1.0). A performance test using data from February 2017 showed that:① The concentration of PM2.5 is forecast well in the BTH region. The correlation coefficients of the observed and forecast daily average PM2.5 in most areas are higher than 0.8, and the forecasted values are close to those observed. The mean errors (ME) are -7.54, -0.46, and -11.0μg·m-3 for the forecast domain, south and north of Hebei province, and 12.04, 2.02, and -13.31 μg·m-3 for the cities of Beijing, Tianjin, and Shijiazhuang, respectively. The correlation coefficients for the forecast and observed hourly relative humidity in the three typical cities are above 0.78, and the mean errors are lower than 3.91%. ② All three parameterization schemes predict the time evaluation of visibility in the BTH region during February 2017 well. In general, the visibility predicted with S1 is the lowest, while that of S3 is the highest; the predictions of S2 are intermediate. In most areas of the BTH region, S1 has the minimum root mean squared error (RMSE) and normalized mean error (NME) between the observed and forecast visibility, while S3 has the maximum RMSE and NME. The error of S2 is between that of S1 and S3, but it shows the best performance in the Beijing area. ③ When the observed visibility is higher than 10 km, the predicted visibilities of the three schemes are all lower than the observed visibility, and S3 has the lowest mean error (ME) and RMSE. S1 has the lowest MB, RMSE, and NME when the visibility is lower than 10 km, especially for visibilities of 1 km to 5 km, which occurred more frequently during heavy haze episodes. The comparison of the results indicated that S1 is best for application to haze forecasting in the BTH region. |
|
|
|
