Abstract. We present a 14-year 12-km hourly air quality dataset created by assimilating satellite observations of aerosol optical depth (AOD) and carbon monoxide (CO) in an air quality model to fill gaps in the contiguous United States (CONUS) air quality monitoring network and help air quality managers understand long-term changes in county level air quality. Specifically, we assimilate the Moderate Resolution Imaging Spectroradiometer (MODIS) AOD and the Measurement of Pollution in the Troposphere (MOPITT) CO observations in the Community Multiscale Air Quality Model (CMAQ) every day from 01 Jan 2005 to 31 Dec 2018 to produce this dataset. The Weather Research and Forecasting (WRF) model simulated meteorological fields are used to drive CMAQ offline and to generate meteorology dependent anthropogenic emissions. Both the weather and air quality (surface fine particulate matter (PM_2.5) and ozone) simulations are subjected to a comprehensive evaluation against multi-platform observations to establish the credibility of our dataset and characterize its uncertainties. We show that our dataset captures regional hourly, seasonal, and interannual variability in meteorology very well across the CONUS. The correlation coefficient between the observed and simulated surface ozone and PM_2.5 concentrations for different Environmental Protection Agency (EPA) defined regions across CONUS are 0.77–0.91 and 0.49–0.79, respectively. The mean bias and root mean squared error for modeled ozone are 3.7–6.8 ppbv and 7–9 ppbv, respectively, while the corresponding values for PM_2.5 are -0.9–5.6 µg/m³ and 3.0–8.3 µg/m³, respectively. We estimate that annual CONUS averaged maximum daily 8-hour average (MDA8) ozone and PM_2.5 trends are -0.30 ppb/year and -0.24 μg/m³/year, respectively. Wintertime MDA8 ozone shows an increasing but statistically insignificant trend at several sites. We also found a decreasing trend in the 95^th percentile of MDA8 ozone but an increasing trend in the 5^th percentile. Most of the sites in the Pacific Northwest show an increasing but statistically insignificant trend during summer. An ArcGIS air quality dashboard has been developed to enable easy visualization and interpretation of county level air quality measures and trends by stakeholders, and a Python-based Streamlit application has been developed to allow the download of the air quality data in simplified text and graphic formats.

中文翻译：

---

通过美国本土 (CONUS) 上空面向公众的空气质量仪表板进行长期高分辨率空气质量再分析


摘要。我们提出了一个 14 年 12 公里每小时的空气质量数据集，通过在空气质量模型中同化气溶胶光学深度 (AOD) 和一氧化碳 (CO) 的卫星观测结果来创建，以填补美国本土 (CONUS) 空气质量监测的空白网络并帮助空气质量管理者了解县级空气质量的长期变化。具体来说，我们从2005年1月1日至2018年12月31日每天吸收社区多尺度空气质量模型（CMAQ）中的中分辨率成像光谱辐射计（MODIS）AOD和对流层污染测量（MOPITT）CO观测来生成此数据集。天气研究和预报 (WRF) 模型模拟气象场用于驱动 CMAQ 离线并生成与气象相关的人为排放。天气和空气质量（表面细颗粒物（PM _2.5 ）和臭氧）模拟均根据多平台观测进行综合评估，以建立我们数据集的可信度并表征其不确定性。我们表明，我们的数据集很好地捕捉了整个美国大陆气象学的区域每小时、季节和年际变化。对于美国大陆不同环境保护局 (EPA) 定义的区域，观测和模拟的表面臭氧和 PM _2.5 浓度之间的相关系数分别为 0.77–0.91 和 0.49–0.79。模拟臭氧的平均偏差和均方根误差分别为 3.7–6.8 ppbv 和 7–9 ppbv，而 PM _2.5 的相应值为 -0.9–5.6 µg/m ³ 分别。我们估计每年 CONUS 平均最大日 8 小时平均值 (MDA8) 臭氧和 PM _2.5 趋势为 -0。分别为 30 ppb/年和 -0.24 µg/m ³ /年。冬季 MDA8 臭氧在几个地点显示出增加但统计上不显着的趋势。我们还发现 MDA8 臭氧在 95 ^th 百分位处呈下降趋势，但在 5 ^th 百分位处呈上升趋势。太平洋西北地区的大多数地点在夏季都呈现出增加的趋势，但统计上并不显着。开发了 ArcGIS 空气质量仪表板，以便利益相关者能够轻松可视化和解释县级空气质量测量值和趋势，并且开发了基于 Python 的 Streamlit 应用程序，可以以简化的文本和图形格式下载空气质量数据。