首页 > , Vol. , Issue () : -
对流层二氧化氮（NO2）是一种重要的痕量污染气体。现有基于OMI卫星探测器、覆盖亚洲地区的NO2公开产品QA4ECV、OMNO2和POMINO受到广泛使用，然而对于这三个产品的差异的定量认识仍然不足。本研究将本团队开发的POMINO产品更新至v2.1，进行了产品的改进和优化，并将反演区域扩大至覆盖东亚、东南亚和南亚大部分地区，随后定量分析了QA4ECV、OMNO2 v4和POMINO v2.1对流层NO2垂直柱浓度在2015-2020年间在不同采样条件下的异同。结果显示，POMINO版本的更新对自身NO2柱浓度反演结果的整体影响较小（<10%）。当三个产品均基于POMINO v2.1的云辐射分数进行一致采样时，产品之间在整个亚洲区域的平均差异约为10%，在京津冀等污染地区的差异最高可达40%。三个产品均显示，京津冀地区NO2柱浓度在五年间下降了约30%，而长三角地区的变化趋势较小。当三个产品基于各自的云辐射分数进行采样时，POMINO v2.1的有效数据量比其他两个产品增加了11-44%，特别是更好地保留了重污染情形下的NO2数据，从而降低了采样引起的对NO2污染水平的系统性低估。本研究对于NO2卫星产品差异的定量分析有助于认识氮氧化物污染状况以及排放和影响评估。
Nitrogen dioxide (NO2) is an important trace gaseous pollutant. There exist three widely used, publically available tropospheric NO2 vertical column density (VCD) products based on OMI over Asia, including QA4ECV, OMNO2 and POMINO. However, quantitative knowledge of the differences between the three products is still inadequate. This research firstly updates the POMINO product developed by our group to version 2.1, including bug fixes and algorithm improvement, and expanding the spatial domain to East Asia, much of Southeast Asia and most of South Asia. Then we quantitatively compare the NO2 data of QA4ECV, OMNO2 v4 and POMINO v2.1 in 2015-2020 under different sampling criteria. Results show that updates of POMINO do not significantly affect the retrieved NO2 VCDs (within 10% averaged over the spatial domain, dependent on seasons). When valid satellite pixels of three products are sampled consistently based on cloud radiation fraction of POMINO v2.1, the relative differences between the three products are about 10% averaged over Asia, although the maximum difference can reach 40% or more in polluted areas like Beijing-Tianjin-Hebei. All the three products show a 30% decrease in annual average NO2 VCDs in Beijing-Tianjin-Hebei over 2015–2020, in contrast to relatively small VCD changes over the Yangtze River Delta. When valid satellite pixels are sampled based on each product’s own cloud screening, POMINO v2.1 provides much more valid pixels in polluted situations by 11-44% and reduces the sampling bias, as a result of its explicit representation of aerosol optical effects in the NO2 and prerequisite cloud retrieval process. Our results provide a basis for using and interpreting the three products, including their differences, effects of sampling and impacts of aerosol representation.