基于随机森林算法对贵阳龙洞堡机场能见度的可预报性研究

Research on Visibility Predictability of Longdongbao Airport in Guiyang Based on Random Forest Algorithm

摘要: 利用2017年贵阳龙洞堡国际机场常规观测逐小时数据以及同期贵阳市新华路站点(1446A)环境污染物逐小时数据，研究在气温、湿度、风速等常规气象要素与环境污染物共同作用下基于随机森林算法对低能见度的变化进行预测研究。研究结果表明：随机森林模型预测值序列与真实值序列相关系数较高，表明随机森林算法在能见度变化趋势上预测效果较好。从随机森林算法输出的因素重要性发现环境污染物的贡献较为重要，进一步研究了各环境污染物的日变化特征和月变化特征。利用HYSPLIT模式确定了机场近地面气团的来源，这使得机场气象要素以及环境污染物的来源地得以确定。

Abstract: Based on the hourly data of routine observation of Guiyang Longdongbao International Airport in 2017 and the hourly data of environmental pollutants at Xinhua Road station (1446A) in Guiyang during the same period, the random forest algorithm was used to predict the change of low visibility under the joint action of routine meteorological elements such as temperature, humidity, wind speed and environmental pollutants. The results show that the correlation coefficient between the predicted value sequence and the real value sequence of the stochastic forest model is high, which indicates that the stochastic forest algorithm is effective in predicting the change trend of visibility. According to the importance of factors output by random forest algorithm, it is found that the contribution of environmental pollutants is more important, and the daily and monthly variation characteristics of each environmental pollutant are further studied. The HYSPLIT model is used to determine the source of the air mass near the surface of the airport, which makes the meteorological elements of the airport and the source of environmental pollutants can be determined.