Insights into summertime surface ozone formation from diurnal variations in formaldehyde and nitrogen dioxide along a transect through New York City

Estimating tropospheric ozone (O3) production from observations is challenging but possible given the close coupling of O3 with formaldehyde (HCHO) and nitrogen dioxide (NO2), two remotely sensed air pollutants. Previous reliance on once-daily satellite overpasses highlights the need to study diurnal changes and surface-column relationships. Using surface observations, Pandora spectrometer retrievals, and a high-resolution (1.33 km) air quality model (WRF-CMAQ), we characterize diurnal patterns of HCHO and NO2 at seven locations along an upwind-downwind pathway through NYC during June-August 2018. Diurnal patterns of the few available surface HCHO concentrations suggest biogenic emission influence, while a bimodal surface NO2 pattern implies local anthropogenic NOx emissions. Details of these patterns vary by site: an afternoon NO2 spike at New Haven (CT) indicates traffic emissions, while a delayed daily HCHO peak at Westport (CT) relative to other sites likely reflects sea breeze dynamics. Peak column concentrations generally lag surface peaks by about four hours, occurring at 9-10 AM for morning NO2 (from Pandora and WRF-CMAQ) and around 4 PM for midday HCHO (from WRF-CMAQ). TROPOMI overpass time at 1:30 PM misses peak column HCHO and NO2 concentrations. A box model (F0AM) constrained with site-level observations and WRF-CMAQ fields indicates 1-9 ppb hr-1 higher noontime local O3 production rates on three sets of paired high- versus mid-to-low-O3 days. F0AM sensitivity analyses on these six days suggest a predominantly transitional O3 formation regime at urban and downwind sites, differing at some sites from the NOx-saturated regime diagnosed for summertime average conditions via the weekday-weekend effect.

Methods
We use version 5.3.1 Community Multiscale Air Quality Modeling System coupled online with version 4.1.1 Weather Research and Forecasting model (WRF-CMAQ) at 1.33 km by 1.33 km horizontal resolution. This simulation provided hourly preceding-hour-average estimates of surface HCHO, NO2, NO, CO, and O3 at the near-surface layer and instantaneous at-the-hour estimates of HCHO and NO2 in 36 vertical sigma pressure levels. WRF-CMAQ simulations are based on model specifications, emissions, and meteorology for the 1.33 km × 1.33 km simulation described in Torres-Vazquez et al. (2022). Emissions are sourced from the 2016 modeling platform version 7.2, based on the 2014 U.S. National Emissions Inventory (NEI) and the 2017 NEI with some additional sector-specific updates. Namely, on-road and non-road emissions were processed down to the county level for 21 counties in and around NYC for 2018 using the Motor Vehicle Emissions Simulator (MOVES) version 2014b. The inventory also incorporates updates from 2018 Continuous Emission Monitoring data for EGU emissions, as well as inline biogenic emissions computed using the Biogenic Emission Inventory System (BEIS) version 3.61. We approximate the values at each site using the nearest land-based pixel based on the geographic coordinates of the site selected from the original model simulations.

基于观测数据估算对流层臭氧（O₃）生成难度较大，但鉴于O₃与甲醛（HCHO）、二氧化氮（NO₂）这两种遥感空气污染物存在紧密耦合关系，该估算具备可行性。过往研究多依赖单日单次卫星过境观测，这凸显了开展日变化特征与柱-地面浓度对应关系研究的必要性。本研究结合地面观测数据、Pandora光谱仪（Pandora spectrometer）反演结果，以及1.33km高分辨率的空气质量模型（WRF-CMAQ），对2018年6-8月沿纽约市（NYC）上风向-下风向传输路径的7个监测点的HCHO与NO₂日变化特征进行了表征。
现有少量地面HCHO浓度的日变化特征显示其受生物源排放影响，而双峰型地面NO₂浓度日变化则表明存在局地人为NOₓ排放。不同监测点的日变化特征存在差异：康涅狄格州纽黑文市出现午后NO₂浓度峰值，反映了交通排放特征；而康涅狄格州韦斯特波特的HCHO日峰值相较于其他站点出现延迟，这大概率与海风环流动力学过程有关。柱浓度峰值通常较地面浓度峰值滞后约4小时：早晨NO₂（来自Pandora观测与WRF-CMAQ模拟）的柱浓度峰值出现在9-10时，而午间HCHO（来自WRF-CMAQ模拟）的柱浓度峰值则出现在16时左右。13:30过境的对流层监测仪（Tropospheric Monitoring Instrument, TROPOMI）未能捕捉到HCHO与NO₂的柱浓度峰值。
基于站点观测与WRF-CMAQ场约束的箱式模型（box model, F0AM）分析显示，在三组共6个高臭氧日与中低臭氧日配对样本中，正午时段局地O₃生成速率高出1-9 ppb·h⁻¹。针对这6天的F0AM敏感性试验表明，城市及下游站点的O₃生成机制以过渡型为主，部分站点的生成机制与通过工作日-周末效应诊断得到的夏季平均氮氧化物饱和机制存在差异。

研究方法
本研究采用水平分辨率为1.33km×1.33km的、版本4.1.1的天气研究与预报模型（Weather Research and Forecasting model, WRF）与版本5.3.1的多尺度空气质量模型（Community Multiscale Air Quality Modeling System, CMAQ）在线耦合的WRF-CMAQ模式。该模拟可提供近地层逐小时前一小时平均的地面HCHO、NO₂、NO、CO及O₃估算值，以及36层垂直σ气压坐标层内的整点瞬时HCHO与NO₂估算值。WRF-CMAQ模拟基于Torres-Vazquez等人（2022）所描述的1.33km分辨率模拟的模型设置、排放清单与气象场。
排放清单源自7.2版2016年建模平台，该平台基于2014年美国国家排放清单（U.S. National Emissions Inventory, NEI）与2017年NEI，并针对部分行业进行了额外更新。具体而言，2018年纽约市及周边21个县的道路与非道路机动车排放，是通过2014b版机动车排放模拟器（Motor Vehicle Emissions Simulator, MOVES）细化至县级尺度得到的。该排放清单还纳入了2018年发电单元（Electric Generating Unit, EGU）的连续排放监测数据更新结果，以及采用3.61版生物源排放清单系统（Biogenic Emission Inventory System, BEIS）计算得到的在线生物源排放。本研究通过选取原始模拟结果中与监测点地理坐标最匹配的陆基网格像元，来近似得到各监测点的模型估算值。