摘要
构建武汉市NO_2浓度的土地利用回归(Land use regression,LUR)模型,可用于个体NO_2长期暴露水平的估计。收集了武汉市10个空气质量监测站2015年的日均NO_2监测数据作为因变量,以武汉市土地利用、海拔高度、人口密度和道路总长度数据作为预测变量,采用逐步回归方法构建LUR模型,并采用留一交叉验证法对模型的精度进行评价。结果显示,武汉市NO_2浓度主要与所在地半径2千米缓冲区内的植被地面积和半径5千米缓冲区内的农用地面积相关。LUR模型的调整R^2大小为0.85,表明模型能解释大部分的变异;LOOCV检验的调整R^2大小为0.63,表明模型具有较好的拟合精度。
The paper aims at establishing the land use regression(LUR) model of NO2 concentration in Wuhan for estimating the individual long-term NO2 exposure level. The average daily NO2 monitoring concentrations of 10 air quality monitoring stations during 2015 in Wuhan were collected and used as dependent variable, and the information of land use, altitude, population density and length of road in Wuhan were collected and used as predictive variables, the multiple stepwise regression method was used to build the LUR model and leave-one-out-cross-validation(LOOCV) method is used to evaluate the accuracy of LUR model. The LUR model showed that NO2 concentration in Wuhan was mainly related to the area of vegetation with 2 km buffer and the area of cropland land with 5 km buffer. The adjusted R2 of the LUR model was 0.85, suggesting that the model could explain most of the variation. The adjustment R2 of LOOCV was 0.63, indicating that the model may fitted well.
作者
刘阳红
李浪姣
王伟业
LIU Yang-hong;LI Lang-jiao;WANG Wei-ye(Jinggangshan University,Ji'an 343009,China;Wuhan University,Wuhan 430079,China)
出处
《廊坊师范学院学报（自然科学版）》
2018年第4期66-69,共4页
Journal of Langfang Normal University(Natural Science Edition)
基金
2017年度江西省教育厅科学技术研究项目“孕期睡眠及肥胖在妊娠期糖尿病发生中交互作用的研究”(GJJ170647),课题组成员:王伟业、黄涛、彭清妹、宋波、李小平、姚昭
