@@ -153,35 +172,47 @@ Extension to the product guide
------------------------------
Several publications report extended data criteria beyond the PUMs quality_flags that should be used to ensure data quality.
Examples of such are the emission source detection paper by Fioletov et al., 2020 and the volcanic so2 monitoring paper by Theys et al., 2019.
Both publications mention the poor quality of observations at the edges of TROPOMI observation swath as well as the reduced quality at high Solar Zenith Angles.
The integration time of pixels toward the edge of the swath has been reduced to decrease the pixel size, however this also reduces the overall quality of the observation (SNR).
Examples of such are the emission source detection paper by *(Fioletov et al., 2020)*,
and the volcanic SO\ :sub:`2` monitoring paper by Theys et al., 2019.
Both publications mention the poor quality of observations at the edges of TROPOMI observation swath
as well as the reduced quality at high Solar Zenith Angles.
The integration time of pixels toward the edge of the swath has been reduced to decrease the pixel size,
however this also reduces the overall quality of the observation (SNR).
Therefor we advise to only select the pixels with id's 25-425, and for really strict cases only 50-400.
Examples of both the edge pixels and SZA effects are shown in the figures below, with 2 figures from Fioletov et al.,2020 and an example for the SZA based on the CSO results.
Examples of both the edge pixels and SZA effects are shown in the figures below,
with 2 figures from Fioletov et al., 2020 and an example for the SZA based on the CSO results.
:alt: Examples of column, error, and quality number for two orbits.
*Examples of retrieved SO2 columns, retrieval errors, and quality flags.*
Furthermore, Fioletov et al., 2020 (Fig below) report large-scale biases in the current TROPOMI product, with TROPOMI showing significantly higher total columns, which can be expected to introduce a potential high bias throughout the domain.
A solution advised by Fioletov et al is to remove the bias by comparing up- and down-wind values around an SO2 emissions source, but that will not be included in this algorithm.
A solution advised by Fioletov et al is to remove the bias by comparing up- and down-wind values around an SO\ :sub:`2` emissions source, but that will not be included in this algorithm.
