土地利用回归模型在大气NO_2浓度空间分布模拟中的应用

Application of land use regression model in simulation of spatial distribution of ambient NO_2 concentration

目的探讨土地利用回归(land use regression,LUR)模型在武汉市大气NO_2浓度空间分布模拟中的应用。方法从武汉市环保局获取2016年1月1日至12月31日市内20个空气质量监测站点的NO_2日均浓度数据,结合土地利用数据、人口密度数据、道路交通数据、工业污染源数据和地形地貌数据,建立土地利用回归模型。采用留一法交叉验证(leave-one out cross-validation,LOOCV)检验模型稳定性。结果土地利用回归模型最终纳入三个变量,分别是7 000 m内污染源数、3 000 m内所有道路长度和5 000 m内建设用地面积。模型调整R^2为0.91,LOOCV的R2为0.89,模型有很高的预测能力。采用LUR预测得到一年内武汉市高分辨率(2 km×2 km)的NO_2浓度分布图。模拟结果显示研究期间武汉市NO_2的平均浓度为24.59μg/m^3,最小值为20.63μg/m^3,最大值为56.77μg/m^3。浓度较高的区域为中心城区,较低的区域为周边远郊区。结论 LUR模型可应用于武汉市大气NO_2浓度空间分布模拟及空气污染暴露评估。

Objective Exploring the application of land use regression（LUR） model in the simulation of spatial distribution of ambient NO2 concentration in Wuhan. Methods Daily average concentrations of NO2 reported by 20 air quality monitoring stations were obtained from Wuhan Environmental Protection Bureau from January 1 to December 31 2016, and were fitted in land use regression model, considering land use, population density, road traffic, industrial pollution source and topography.A leave-one out cross-validation（LOOCV） procedure was used to evaluate the stability of the model. Results Three predictors were fitted in the final LUR model： the number of industrial sources within a 7-km buffer, the length of all roads within the 3-km buffer around monitoring sites, and the construction land area within a 5-km buffer. Adjusted R2 in LUR models and LOOCV R^2 were 0.91 and 0.89 respectively. During the investigated period, the average,minimum and maximum concentration of NO2 in Wuhan was 24.59 μg/m^3, 20.63 μg/m^3 and 56.77 μg/m^3 respectively. Higher concentration of NO2 occurred in the center of urban area, lower in rural area. Conclusion LUR model can be used for the simulation of spatial variance of ambient NO2 concentration and the assessment of air pollution exposure in Wuhan.