2013年以来，我国的清洁空气政策显著提高了空气质量。量化政策对于PM2.5降低的贡献是当前国内外研究者关注的热点，并开展了大量基于大气化学传输模式以及统计模型的研究。然而，大气化学传输模式的模拟难以反映环境政策的执行在空间上的差异，统计模型分析则依赖于监测站的时间连续观测，无法支持空间维度上的分析。目前关于环境政策对PM2.5下降贡献的空间差异的量化分析仍然十分匮乏。本研究将引入卫星遥感数据，通过对现有卫星数据驱动的PM2.5预测模型及补缺方法的系统性评估及优化，建立时空连续的高分辨率PM2.5预测。之后一方面在空间维度上扩展机器学习算法以控制气象条件对污染物浓度的影响，从而量化环境政策引起的PM2.5浓度下降的空间差异。另一方面采用因果推断的研究设计，根据各个区县开展PM2.5监测的差异对区县进行分组及匹配，并通过类似随机试验的回归分析控制气象场的影响，从而量化环境政策的贡献。

The severe PM2.5 pollution in China is among the best-known environmental problems in the past decade. To improve air quality, China implemented strict pollution control policies since 2013 that remarkably decreased PM2.5 pollution. Quantifying the contribution of clean air policies on PM2.5 reduction is challenging since meteorology also modifies the variations in PM2.5 concentration. Both chemical transport modelling and statistical approaches have been previously employed to assess the contribution of clean air policies on PM2.5 reduction. However, chemical transport models can hardly reflect the spatial heterogeneity in policy implementation and policy efficacy, and the statistical approaches generally relied on ground measurements that could hardly support spatial analyses. Studies on the quantification of spatial heterogeneity of clean air policies’ effects are limited. Here we plan to comprehensively evaluate the gap-filling methods in the PM2.5 prediction model driven by satellite data in order to provide spatiotemporally continuous prediction of PM2.5 at a high spatial resolution. Then with the complete-coverage PM2.5 predictions, we will apply the machine learning method that disparts the effect of meteorology on PM2.5 versions. By analyzing the trend of meteorology-adjusted PM2.5 variations, we will assess the spatial differences in policy induced PM2.5 reduction. Additionally, with the causal inference method, we will apply the difference-in-difference matching estimator and group China counties by their PM2.5 reduction policies. By matching counties with their neighbors, we could control the effects of meteorology and compare the effects of clean air policies across space. This study will provide assessment of the spatial heterogeneity of clean air policy efficacy in China in order to inform and support future pollution control policies.