A novel hyperspectral remote sensing technique for horizontal distribution of atmospheric gaseous pollutant: to fill the spatio-temporal resolution limitations of satellite and in-situ observations

Abstract

High spatial-temporal resolution distribution of atmospheric gaseous pollutant is an important basis for tracing its emission, transport and transformation. At present, methods commonly used to obtain the regional horizontal distribution of trace gas are based on satellite remote sensing or numerous in-situ observation. However, typical trace gas monitoring satellites only have a few fixed overpassing times with a spatial resolution limited to several kilometers, which make it hard to locate minor emission sources. Limited in-situ observations have limited coverage, and can only obtain trace gas concentration information near the observation point. In this study, we propose a method for the long-term detection of the horizontal distribution of trace gas. The spatial resolution in the direction of rotation was up to 0.1°, and the spatial resolution in the optical path direction, reached the kilometer level. Meanwhile, the temporal resolution of the results reached the hourly level during the daytime. The obtained trace gas horizontal distribution was consistent with the in-situ and mobile measurements. Compared with satellite remote sensing, this method achieved horizontal distribution results with higher spatial and temporal resolutions, and located several small high-value areas in Hefei, China. The satellite NO2 vertical column density (VCD) distribution results were evaluated via the NO2 horizontal distribution obtained from the hyperspectral NO2 horizontal distribution at 13:30 (local time) (UTC+8:00) on April 2, 2022. The Tropospheric NO2 VCD results of the satellite at transit time (13:30) were consistent with the hyperspectral NO2 horizontal distribution results at 13:00–14:00 on the same day but were not consistent with the daily average NO2 results. The hourly NO2 concentration in each area was 10–40% lower than the daytime average obtained by the hyperspectral remote sensing result. Based on these results, we approximated the errors associated with the calculation of NO2 emissions based on the satellite results, with a maximum bias of approximately 69.45–83.34%.