基于AOD数据的新疆大型露天煤炭开采区PM2.5和PM10反演

MODIS气溶胶产品AOD与PM2.5、PM10浓度高度相关,已广泛应用在PM2.5、PM10浓度模拟.该研究以新疆维吾尔自治区大型露天煤炭开采区准东矿区为研究对象,结合实测的2014年5月、7月、9月、12月PM2.5、PM10质量浓度数据与经过垂直湿度订正的MODIS气溶胶产品AOD,利用多元回归进行拟合建模,从建立的40个模型中选取最优模型并据此对研究区PM2.5、PM10的质量浓度进行定量估算.结果表明:AOD与PM2.5、PM10呈极显著正相关;4个月AOD与PM2.5、PM10质量浓度估算模型最优模型均为多项式模型;其中7月AOD与PM2.5质量浓度拟合模型较好(R2=0.6258),实测值与预测值拟合趋势线R2为0.8057;9月PM10拟合模型效果理想(R2=0.7329),实测值与预测值拟合趋势线R2为0.8077;将AOD代入最优模型反演PM2.5,从空间层面上反映出各区域PM2.5浓度差异明显.研究结果可为AOD的深度利用与PM2.5、PM10浓度的遥感估算提供参考,在大气污染物空间分布、监测大气环境质量、污染预测等方面都具有重要意义.

Inversion of PM2.5 and PM10 content based on AOD data in large opencast coal mining area of Xinjiang

Due to the high efficiency, large scale, low cost, and some other advantages, satellite remote sensing technology can cover the shortage of traditional ground-based observations, which can reflect the distribution, transmission path and diffusion dynamic of atmospheric pollutants in large scale. The MODIS aerosol product i.e. aerosol optical depth (AOD) and PM2.5 and PM10 (aerosol particulate with the diameter of less than 2.5 and 10 μm, respectively) had a high correlation, and AOD has been applied into the quantitative simulation of PM2.5 and PM10 concentration in existing researches. However, it is hard to estimate the PM2.5 and PM10 concentration with high precision, because of the temporal and spatial differences of AOD. The pretreatment of the vertical humidity correction for MODIS aerosol products can eliminate the influence of uncertainties in the atmosphere to a certain extent, and improve the precision and robustness of the quantitative estimation. Therefore, this study aimed to bring the vertical humidity correction into the preprocessing of MODIS aerosol product AOD. With 52 atmospheric dust samples collected from the Zhundong Industrial Park in Xinjiang Uighur Autonomous Region, China, the AOD and the concentration of PM2.5 and PM10 obtained in May, July, September, and December of 2014 were combined to establish the multiple regression fitting model. A total of 40 quantitative models were established, and the model based on polynomial was more robust and accurate than the others, which was applied to predict the concentration of PM2.5 and PM10 of Zhundong Industrial Park. Finally, the optimal fitting models were applied in the prediction of local inhalable particulate matter concentration in May, July, September, and December of 2014. Taking the case of PM2.5, multiple regression model and AOD were used to estimate the local PM2.5 mass concentration, the spatial representation of which was conducted by ArcGIS 10.0. The results showed that: The mass concentrations of PM2.5 and PM10 in the study area were inhomogeneous, and the concentration level of PM10 was much higher than that of PM2.5; and the variations of them were significant. AOD was significantly related with PM2.5 and PM10, separately (P<0.01). The optimal predicting models between AOD and the concentration of PM2.5, PM10 in each month (May, July, September, and December) were the polynomial models. The R2 of the estimation model between AOD and the concentration of PM2.5 reached 0.6258 in July and the R2 of the trend line fitted between measured value and predicting value was 0.8057; the R2 of the estimation model between AOD and the concentration of PM10 was 0.7329 in September, and the R2 of the trend line fitted between measured value and predictive value was 0.8077. The optimal model was applied with AOD to invert the concentration of PM2.5, which could reflect the spatial distribution characteristics and variations of PM2.5 mass concentration in the Zhundong Industrial Park. This research can provide reference for the deep utilization of AOD and the estimation of PM2.5 and PM10 concentrations by means of remote sensing method, which has important significance in spatial distribution, remote sensing monitoring, and the forecasting of local atmospheric pollutants.