银川紫外辐射特征及TUV模式适用性研究

Research on the characteristics of ultraviolet radiation in Yinchuan and applicability of TUV model

利用2013-2019年银川紫外观测数据,分析了银川紫外辐射变化特征,基于TUV模式,结合云光学厚度、云顶高度、气溶胶光学厚度、单次散射反照比、波长指数、臭氧柱浓度和NO2柱浓度等遥感资料,研究了TUV模式在银川紫外辐射模拟中的适用性.结果表明,银川多年年平均紫外辐照度为8.7 W·m^(-2),7月最高为17.7 W·m^(-2),1月最低为3.1 W·m^(-2),13:00前后达到日最大值;不同参数输入方案下TUV对银川紫外辐照度随时间变化趋势的把握能力均较好,但平均值和标准差都普遍存在高估现象,2-5月TUV模拟效果最差,1月和6-10月模拟效果较好,11和12月模拟效果最好;模拟层数的增加不能有效提高模拟精度,改善TUV模拟效果;TUV模式模拟误差随着云光学厚度、相对湿度的减小及臭氧柱浓度、云顶高度的增加而增大;TUV对日照时数小于10 h时的紫外辐照度的模拟效果较日照时数大于10 h时好;PM10浓度为0~50μg·m^(-3)时TUV模拟均方根误差最小.

Yinchuan ultraviolet observation data was used to analyze the variation characteristics of ultraviolet radiation, and based on the TUV model, combined with remote sensing data such as cloud optical thickness, cloud-top height, aerosol optical depth, single scattering albedo, wavelength exponent, ozone column concentration and NO2 column concentration, the applicability of TUV model in Yinchuan ultraviolet radiation simulation was studied. The results showed that the annual average ultraviolet irradiance in Yinchuan was 8.7 W·m^(-2). The peak of the monthly maximum of ultraviolet radiation appeared in July with 17.7 W·m^(-2) and the minimum appeared in January, was only 3.1 W·m^(-2), it reached the daily maximum around 13:00. The TUV model showed better grasp in the variation trend of UV irradiance with time, but the average and standard deviation were generally overestimated. The TUV model worked the worst from February to May, better in January and June to October, and the best in November and December. The simulation error of TUV model was affected by cloud optical thickness, relative humidity, ozone column concentration and cloud top height;the smaller the first two were or the higher values of the latter two were, the greater error was. The simulation effect of TUV on ultraviolet irradiance was better when the sunshine duration were less than 10 h than the sunshine duration greater than 10 h. When the PM10 concentration was between 0~50 μg·m^(-3), the TUV simulation had the smallest root mean square error.