2.1. Observational analysis

Remotely sensed daily surface air temperatures were accessed for 1 January – 1 March 2003–2020 from the Atmospheric Infrared Sounder (AIRS) instrument onboard the NASA Aqua satellite. Launched in May 2002, Aqua follows a global, sun-synchronous orbit with AIRS temperatures designed to achieve a 1 K accuracy in the lower troposphere under clear and partly cloudy conditions (^{Aumann et al., 2003}). Because surface temperatures increase seasonally during the two-month retrieval period, daily anomalies were computed for each pixel using the 18-yr period of record. The analysis was separated into two periods, 1–22 January 2020 and 23 January – 1 March 2020 to characterize the evolution of aerosol concentrations and temperature during the implementation of the quarantines. The latter period represents the timespan when mobility restrictions were strictest (^{Sulaymon et al., 2021}), and will hereafter be referenced as the C19Q period, whereas the pre-C19Q period (1–22 January 2020) will be termed PRE. Though mobility remained restricted in some locations beyond 1 March 2020 (^{Sulaymon et al., 2021}), this study focuses on the period when containment was most prevalent. Meanwhile, the Global Historical Climatology Network (GHCN) (^{Menne et al., 2012a; Menne et al., 2012b}) supplied in-situ daily mean temperature (T_AVG) observations for 153 active GHCN sites in China east of 90°E with at least 50 years of records. The same PRE and C19Q anomalies were calculated for the GHCN T_AVG as the AIRS observations. Large-scale lower-tropospheric patterns were visualized with the ERA5 global atmospheric reanalysis (^{Hersbach et al., 2020}).

Aerosol optical depth (AOD) measurements were retrieved from Terra's Moderate Resolution Imaging Spectroradiometer (MODIS) (2003−2020) and NOAA-20's Visible Infrared Imaging Radiometer Suite (VIIRS). While Terra MODIS AOD possesses a longer period of record, Level 2 NOAA-20 VIIRS AOD data contain the diagnosis of an intermediate aerosol model that classifies aerosols in each pixel as either predominantly oceanic, dust, generic, urban, or heavy smoke (^{Laszlo and Liu, 2016}). For each 0.10° pixel over east China, the fraction of all PRE retrievals classified as “urban” was computed, and this process was repeated for C19Q. For instance, if a VIIRS pixel had 10 quality retrievals during PRE, and they were composed of four urban-, two smoke-, two dust-, and two oceanic-dominant days, then that pixel would receive a 40% PRE urban aerosol fraction. The change in urban aerosol fraction between the C19Q and PRE periods was determined and expressed as a percent change (i.e., decrease from 30% PRE urban aerosol retrievals to 15% during C19Q would correspond to a 50% reduction). Because some areas of the domain experience a low frequency of urban aerosols under normal circumstances or experienced a low number of quality retrievals (i.e., cloud contamination) in early 2020, percent increases/decreases could vary wildly (i.e., increase from 2% PRE urban aerosol retrievals to 6% during C19Q would correspond to a 200% increase). Thus, the analysis was coarsened to 1.0° resolution, and the percent change in urban aerosol fraction was only computed if the PRE fraction was at least 13.9%, the median value across the domain.