Research On Spatiotemporal Characteristics And Influencing Factors Of Beijing-Tianjin-Hebei Visibility

With the rapid development of social economy and the continuous advancement of urbanization,air pollution has become a major problem that people have to face.Among them,visibility,as an important indicator reflecting atmospheric transparency and characterizing air quality,is closely related to people's production and life,transportation,physical and mental health,etc.Therefore,research on visibility is of great significance.Based on the meteorological monitoring data of the Beijing-Tianjin-Hebei region from January 1 to December 31,2019 and the monitoring data of air pollutants in the same period,this paper analyzes the characteristics of the spatial and temporal changes of atmospheric visibility in the Beijing-Tianjin-Hebei region.The nonlinear relationship can be explained in previous studies.For the lack of problems,a model combining integrated learning and SHAP interpretation framework was proposed to study the contribution of visibility influencing factors,and the differences in the contribution of urban and suburban factors were compared.Finally,the quantitative analysis of the compound influence relationship between PM_2.5 and relative humidity on visibility was focused on.The main conclusions are as follows:(1)In terms of time,the visibility of the Beijing-Tianjin-Hebei region has obvious characteristics of inter-day and inter-month cycles.In a year,the spring and summer seasons are better than the autumn and winter seasons.Within a day,the visibility increases first.The downward trend reaches the highest value at 3 and 4o'clock in the afternoon,and there is no significant difference between working days and non-working days;spatially,the overall visibility is decreasing from south to north,with slight differences in the four seasons,and there is an obvious autocorrelation relationship.The degree of autocorrelation is the highest.From the perspective of the aggregation mode,the visibility is mostly high-value aggregation and low-value aggregation.On the whole,the visibility of Beijing is followed by Tianjin and Hebei is the worst.(2)This article uses multiple linear regression and ensemble learning(e Xtreme Gradient Boosting,XGBoost)and random forest regression methods to build models to analyze the overall interpretation of the visibility of influencing factors,among which the random forest regression model is The combined coefficient R² and the explanatory variance reached 0.8973 and 0.8978,respectively,and the fitting effect was the best;then the SHAP explanatory framework combined with random forest was introduced to specifically analyze the magnitude and direction of the contribution of the impact factors,and clearly reveal that the characteristic variables are different The specific contribution of the interval to visibility.The change rate of the contribution of PM2.5 changes from rapid to gentle with 100?g/m3 as the turning point;comparative analysis of the difference in the contribution of urban and suburban factors can be found.The impact of urban air pollutants on visibility is higher than that In the suburbs,there are differences in the univariate contributions of SO2,CO,wind speed and temperature.When SO2 is within 20?g/m3,the visibility change rate in the urban area is significantly greater than that in the suburbs.(3)There is a complex compound influence relationship between PM_2.5concentration and relative humidity and visibility.Under different relative humidity,PM_2.5 has different effects on visibility.In an air environment with RH<60%,the overall visibility is better.And the sensitivity to PM_2.5 concentration changes is low.With the increase of relative humidity,the sensitivity of visibility to PM_2.5 increases,and the influence of PM_2.5becomes stronger and stronger.