Investigation of 3D-effects for UV/vis satellite and ground based observations of volcanic plumes

Abstract

Abstract. We investigate effects of the 3-dimensional (3D) structure of volcanic plumes on the retrieval results of satellite and ground based UV-vis observations. For the analysis of such measurements usually 1D scenarios are assumed (the atmospheric properties only depend on altitude). While 1D assumptions are well suited for the analysis of many atmospheric phenomena, they are usually less appropriate for narrow trace gas plumes. For UV/vis satellite instruments with large ground pixel sizes like GOME-2, SCIAMACHY, or OMI, 3D effects are of minor importance, but usually these observations are not sensitive to small volcanic plumes. In contrast, observations of TROPOMI aboard Sentinel-5P have a much smaller ground pixel size (3.5 × 5.5 km2). Thus on the one hand, TROPOMI can detect much smaller plumes than previous instruments. On the other hand 3D effects become more important, because the TROPOMI ground pixel size is smaller than the height of the troposphere and also smaller than horizontal atmospheric photon path lengths in the UV/vis spectral range. In this study we investigate the following 3D-effects using Monte-Carlo radiative transfer simulations: 1. the light mixing effect caused by horizontal photon paths, 2. the saturation effect for strong SO2 absorption, 3. geometric effects related to slant illumination and viewing angles, and 4. Plume side effects related to slant illumination angles and photons reaching the sensor from the sides of volcanic plumes. Especially the first two effects can lead to a strong and systematic underestimation if 1D retrievals are applied (more than 50 % for the light mixing effect, and up to 100 % for the saturation effect). Besides the atmospheric radiative transfer, the saturation effect also affects the the spectral retrievals. Geometric effects have a weaker influence on the quantitative analyses, but can lead to a spatial smearing of elevated plumes or even to virtual double plumes. Plume side effects are small for short wavelengths, but can become large for longer wavelengths (up to 100 % for slant viewing and illumination angles). For ground based observations, most of the above mentioned 3D effects are not important, because of the narrow FOV and the closer distance between the instrument and the volcanic plume. However, the light mixing effect shows a similar strong dependence on the horizontal plume extension as for satellite observations and should be taken into account for the analysis of ground based observations.

Automatic Estimation of Daily Volcanic Sulfur Dioxide Gas Flux From TROPOMI Satellite Observations: Application to Etna and Piton de la Fournaise

“…The quality of the retrieval mainly depends on variations in surface reflectivity, solar zenith angle, atmospheric conditions, proximity to swath edge, 3D effects, while also depending on the retrieval algorithm itself (Fioletov et al., 2020; McCormick et al., 2013; Wagner et al., 2023). In the literature, the “uncertainty of the retrieval” is generally reported in the form of a single aggregated standard deviation “ σ ” (after appropriate unit conversion, σ = κ · σ CA , where pixel mass standard deviation σ is expressed in kton and pixel column amount standard deviation σ CA is expressed in Dobson Units).…”

Improved catalog of NOx point source emissions (version 2)

“…With TROPOMI, however, pixel size becomes so small that 3D effects of radiative transfer matters. As shown in Wagner et al (2023), horizontal light paths lead to a smearing out of the satellite observations of a confined plume: TVCDs of plume pixels are generally biased low when derived with a 1D AMF, while neighboring pixels are biased high. For a plume of 1 km × 1 km × 1 km, Wagner et al (2023) report a low bias of up to 30 % for NO 2 for the TROPOMI pixel covering the plume.…”

“…As shown in Wagner et al (2023), horizontal light paths lead to a smearing out of the satellite observations of a confined plume: TVCDs of plume pixels are generally biased low when derived with a 1D AMF, while neighboring pixels are biased high. For a plume of 1 km × 1 km × 1 km, Wagner et al (2023) report a low bias of up to 30 % for NO 2 for the TROPOMI pixel covering the plume. Note, however, that this effect is slightly dampened by the spatial integration within 15 km applied in the v2 catalog.…”

Automatic estimation of daily volcanic sulfur dioxide gas flux from TROPOMI satellite observations: application to Etna and Piton de la Fournaise

“…The quality of the retrieval mainly depends on variations in surface reflectivity, solar zenith angle, atmospheric conditions, proximity to swath edge, 3D effects, while also depending on the retrieval algorithm itself (McCormick et al, 2013;Fioletov et al, 2020;Wagner et al, 2023). In the literature, the "uncertainty of the retrieval" is generally reported in the form of a single aggregated standard deviation "σ" (after appropriate unit conversion, σ = κ.σ CA , where pixel mass standard deviation σ is expressed in kton and pixel column amount standard deviation σ CA is expressed in Dobson Units).…”