基于MAIAC的汾渭平原典型区域气溶胶光学厚度时空变化研究

Research on spatial and temporal distribution of AOD in a typical region of the Fenwei Plain based on MAIAC

关键词:  MAIAC  汾渭平原  气溶胶光学厚度  时空变化

Key words:  MAIAC  the Fenwei Plain  aerosol optical depth (AOD)  spatial and temporal distribution

摘要:

汾渭平原作为大气污染防治的重点区域，其空气质量变化一直被重点关注。文章选取汾渭平原的典型区域陕西省铜川市为例，以气溶胶光学厚度 （AOD） 作为表征空气质量的重要指标，分析区域内 AOD的时空变化，为有效防控大气污染提供决策支持。研究选用基于多角度大气校正 （MAIAC） 算法的 MCD 19A2 遥感影像数据，分析获取 2011 年 1 月到 2020 年 12 月逐月的 AOD 空间分布，同时基于 ARIMA模型进行预测分析，并通过灰色关联分析 （GRA） 方法研究AOD的影响因素，为进一步提高其预测精度提供辅助支持。研究结果显示，区域内AOD均值从月度变化特征上看，6—8月相对较高，而在10—12月相对较低；从年度变化特征上看呈波动状，在2013年AOD均值达到最高，之后呈现整体下降趋势， 2020年达到最低。基于2011—2019年逐月AOD均值采用ARIMA （1, 2, 5） 模型预测2020年1 —12月的AOD均值，其预测均方根误差 （RMSE） 为0.117，预测精度良好，但仍有待进一步提高。通过AOD与区域内自然和社会因子指标的灰色关联分析可以看出，工业废气排放总量、人口总数和工业二氧化硫排放总量是影响研究区域AOD的重要社会因子指标，而自然因子由于气溶胶本身生成机制的复杂性，均对AOD有不同程度的影响，并且根据年月研究单元尺度的差异，相同因子的影响关联度也有所不同。由此得出，基于多时相的MCD 19A2产品研究AOD的时空变化是可行的，通过引入有效的影响因子作为辅助变量可进一步提高AOD的预测精度。为推动区域生态绿色可持续发展，不仅要减少废气排放、 加强政策调控、优化产业结构、发展绿色低碳经济、减少对能源和矿产的依赖，还要加快现有污染源的治理效力，并且要加强遥感等新技术的应用，提高环境质量的监测时效。

Abstract:

Background, aim, and scope The Fenwei Plain is a critical area for air pollution prevention and control, its air quality dynamics have always been the focus of attention. AOD (Aerosol Optical Depth) is an important indicator of air quality, so studying the spatial and temporal distribution of AOD is an effective method to access the change of air quality. In this paper, Tongchuan City located in Shaanxi Province within the Fenwei Plain was selected as a typical area to study the spatial and temporal variations of AOD, which could provide decision support for effective prevention and control of air pollution. Materials and methods The MCD（MODIS Combined Dataset）19A2 images based on MAIAC（Multi-Angle Implementation of Atmospheric Correction） algorithm were selected to study the spatial and temporal distribution of AOD in each month from January 2011 to December 2020. An ARIMA (Autoregressive Integrated Moving Average) model was applied for prediction analysis of AOD. By collecting the natural and social factors of the study region, the influencing factors of AOD were studied by GRA (Grey Relation Analysis), which could provide auxiliary roles for improving the prediction accuracy of AOD. Results The results indicated, the AOD monthly mean in the study area was relatively high from June to August, and relatively low from October to December. The AOD annual mean exhibited fluctuations, it reached a maximum in 2013, and then it showed an overall downward trend, the minimum appeared in 2020. The AOD monthly mean of the twelve months in 2020 were predicted with ARIMA（1, 2, 5） model based on the monthly mean from January 2011 to December 2019, the RMSE of prediction was 0.117, which indicated satisfactory prediction accuracy. Discussion Based on the long time series variation characteristics of AOD, it could be seen that the air quality in the study area has been improved since 2013, but the prediction accuracy still needs to be further improved. The GRA results between AOD and factors indicated that industrial emission, gross population and SO2 emissions were important social factors influencing AOD, however, all natural factors could affect AOD more or less because of the complexity of AOD generation mechanism, and according to the difference of research scale, the contribution of the same factors are different. Conclusions It was feasible to study the spatial and temporal variation of AOD based on multi-temporal MCD 19A2 images. The influencing factors should be introduced as auxiliary variables in order to improve the prediction accuracy of AOD. Recommendations and perspectives In order to promote ecological sustainable development, some efforts should be made. On the one hand, it is necessary to reduce emissions by strengthening policy management and optimizing industrial structure and trying to reduce dependence on energy and minerals; on the other hand, we should pay attention to the treatment of existing pollution sources; at the same time, the application of new technologies such as remote sensing should be further applied to improve the monitoring efficiency of environmental quality.