基于卫星遥感的氮氧化物排放快速反演方法比较

Comparison of Satellite-Based Fast Inversion Methods for Nitrogen Oxides Emissions

基于对流层二氧化氮(NO_(2))垂直柱浓度卫星遥感数据,实现快速、高空间水平分辨率(5 km或更高)的氮氧化物(NO_(x)=NO+NO_(2))排放反演,可为空气污染精准治理提供及时、细致的排放数据。现有多种低计算成本的快速反演方法,如指数修正高斯模型、散度模型和PHLET算法,但其反演效果尚未得到充分对比分析。以2019年夏季京津冀地区为研究对象,对比了上述3种方法的反演效果,研究发现,指数修正高斯模型主要适用于点源排放,但在京津冀等排放源密集地区的反演效果较差;散度模型考虑了在预定NO_(x)大气寿命情况下的水平输送,能快速识别主要排放源位置,但存在排放低估和负排放等问题;PHLET算法考虑了水平输送、NO_(2)垂直柱浓度和NO_(x)大气寿命的非线性关系以及卫星像元不规则等因素,对排放的估计较为准确。改善风场数据、填补卫星数据缺失和改善NO_(x)化学损失估计是进一步提升排放反演质量的关键。

Satellite-based fast inversion for nitrogen oxides(NO_(x)=NO+NO_(2))emissions at low computational costs and high resolutions(≤5 km or finer)can provide timely,detailed data to support targeted pollution control.To date,a variety of low-cost fast inversion methods have been developed,such as the Exponentially Modified Gaussian(EMG),Divergence(DIV),and the PHLET(Peking University High-resolution Lifetime-Emission-Transport)models.However,quantitative comparisons of these methods and their emission results are lacking.This study compares the above three inversion methods for the Beijing-Tianjin-Hebei region during the summer of 2019.We found that the EMG model,which was designed for point source emission inversion,performs poorly in Beijing-Tianjin-Hebei due to dense emission sources even within each city.The DIV considers the horizontal transport of NO_(x) with a predetermined(fixed)lifetime and can quickly identify the locations of emission sources;however,it tends to underestimate the emission amounts and even leads to negative emissions in many places.PHLET algorithm considers the horizontal transport of NO_(2),the nonlinear relationship between local NO_(2) concentrations and lifetimes,and the twoway matching between irregular satellite pixels and regular model grid cells,resulting in more reliable emission estimates.Filling in missing satellite data through data fusion,improving wind data resolution and accuracy,and improving NO_(x) chemical loss estimation will significantly enhance the quality of emission inversion.