机场典型霾过程大气消光特性及能见度预测模型

AN ATMOSPHERIC EXTINCTION CHARACTERISTICS AND VISIBILITY PREDICTION MODEL FOR TYPICAL HAZE PROCESSES AT AIRPORTS

大气能见度是机场运行的重要指标之一,探究机场低能见度天气形成机理、准确预测能见度变化趋势对保障航空交通安全高效运行至关重要。选取天津机场为研究对象,开展大气光学参数、污染物浓度及气象条件监测,研究了2020年12月08—23日一次典型霾天气过程中机场大气消光特性,并基于广义加性模型(generalized additive model,GAM)和梯度提升回归树模型(gradient boost regression tree,GBRT)分别构建机场能见度预测模型,对比并确定最优模型。结果表明,在天津机场冬季一次典型霾污染引起的低能见度过程中,大气总消光系数Bext为37.4~891.7Mm^(-1),平均值为346.0 Mm^(-1)。其中,B_(sp)、B_(ap)、B_(ag)、B_(sg)对B_(ext)的贡献占比分别为73.7%、11.7%、5.9%和8.7%,气溶胶污染是造成能见度降低的主要因素,GBRT分析显示1μm以下粒子对总消光的相对贡献最大。同时,大气黑碳(BC)和NO_(2)也会通过消光作用降低能见度。利用气象参数和污染物浓度数据,通过GAM和GBRT模型均可以对机场能见度进行较为准确的预测,其中GBRT模型的拟合效果优于GAM模型,表明GBRT模型可在多发的霾天气中提供准确可靠的机场能见度预测。

Atmospheric visibility is one of the most important indicators of airport operations,and it is vital to investigate the mechanism of low visibility weather formation at airports and to accurately predict visibility trends for the safe and efficient operation of air traffic.By monitoring the atmospheric optical parameters,pollutant concentrations,and meteorological conditions at Tianjin Airport,we studied the extinction characteristics of the airport atmosphere during typical haze weather from 8th to 23rd December 2020,constructed airport visibility prediction models based on the generalized additive model(GAM)and the gradient boost regression tree(GBRT)respectively,and compared the prediction results to determine the optimal model.The results indicated that during a typical winter haze pollution-induced low visibility at Tianjin Airport,the Bext ranged from 37.4Mm^(-1)to 891.7Mm^(-1),with a mean value of 346.0Mm^(-1).The contributions of B_(sp),B_(ap),B_(ag)and B_(sg)to B_(ext)respectively accounted for 73.7%,11.7%,5.9%,and 8.7%,with aerosol pollution being the visibility reduction,and GBRT analysis showed that the relative contribution of particles below 1μm to total extinction was the largest.Meanwhile,BC and NO_(2)can also reduce visibility through extinction.Using meteorological parameters and pollutant concentration data,both the GAM and GBRT models can provide more accurate prediction on airport visibility,and the GBRT model fitting better than the GAM model,indicating that the GBRT model can provide accurate and reliable airport visibility predictions in frequent haze weather.