Keywords:Atmosphere, Satellite observation, Nitrogen dioxide
TROPOspheric Monitoring Instrument (TROPOMI) aboard the Sentinel-5 Precursor satellite has an unprecedented fine horizontal resolution of 7 km×3.5 km (at nadir) for tropospheric nitrogen dioxide (NO2) column observations and is expected to improve estimation of emissions, including localized kilometer-scale emissions. However, previous studies reported that underestimation could occur in tropospheric NO2 column data from satellite observations, although its causes are still under discussion. The present study attempts to validate TROPOMI tropospheric NO2 column data to confirm that the underestimate occurs and clarify the causes. For this purpose, we used ground-based 4-different-azimuth-viewing Multi-Axis Differential Optical Absorption Spectroscopy (4AZ-MAXDOAS) installed at Chiba, Japan (35.63ºN, 140.10ºE, 21 m asl). The 4AZ-MAXDOAS observed tropospheric NO2 simultaneously in 4 different azimuth directions, enabling the evaluation of spatial inhomogeneity of NO2, which has been considered to be the major cause of the underestimate. From 4AZ-MAXDOAS data, we found that differences in tropospheric NO2 column data among 4 different azimuth directions reached up to 40%, indicating the existence of significant horizontal spatial inhomogeneity in NO2 around the observation site. Then, we compared 4AZ-MAXDOAS data with coincident TROPOMI data. TROPOMI data showed an underestimation by up to about 50% compared to 4AZ-MAXDOAS data, confirming the underestimate in TROPOMI data. However, the correlation was not clear between the magnitude of the underestimate and the coefficient of variance in 4-azimuth data from 4AZ-MAXDOAS observations. This suggests that the observed underestimate cannot be explained only by the effect of the NO2 spatial inhomogeneity. Instead, the underestimate should be attributed more significantly to the assumption made in the air mass factor calculation.