Total column ozone comparisons
The time series of the satellite and reanalysis data, along with individual measurements at the EMA in Quito (see the “Methods” section), is depicted in Figure 1. The main features of the time series are consistent across the different data sets, mostly in terms of annual shape. profiles and the time of year when TCO reaches its maximum (mid-September, Fig. S1). However, there are differences in the magnitude of TCO between ozonesondes and datasets that have been quantified and discussed below. Table 1 summarizes the spatial and temporal resolution of the space measurements and the comparison with ozonesondes.
The TCO of TROPOMI/S5P overestimates the sounding measurements by a mean positive bias of + 8.8 DU, corresponding to a mean difference of 3.7% (Figs. 2a,b). Although the spatial resolution of TROPOMI/S5P is much better than that of GOME-2/MetOP-B, the latter performs similarly as its average bias (Fig. 2c,d) over the soundings is + 7.7 DU (3.2% difference). Similar deviations in the measurements of these sensors have been documented at other tropical stations17. In contrast, the difference between OMI/Aura and the soundings (+ 2.7 DU or 1.2% mean difference; Fig. 2e,f) is a third of the difference observed for TROPOMI/5SP and GOME-2/MetOP-B. At the same time, the TCO of OMPS/Finland’s nuclear power plant was found to have performed the best of all products (Figure 2g,h), as the average deviation to soundings is −0.6 DU (−0.2% average difference). In all cases, the slope of the linear regression is 0.7 and the correlation coefficient (R2) is 0.6. From an observation point of view, the TCO of OMPS/Finland NPP and OMI/Aura exceeds TROPOMI/S5P and GOME-2/MetOP-B in the tropical Andes. Previous work shows that OMI/Aura and OMPS/Finland agree within 2% with measurements from most other tropical stations.1.
As shown, the product with the best spatial resolution (TROPOMI/S5P) does not provide the best comparison. This is counterintuitive, as a finer resolution would seem more likely to resolve a significant elevational gradient at the study site. However, an important aspect to consider is the structure and magnitude of tropospheric ozone, which has been identified as a factor causing bias in TCO retrievals.18. Satellite algorithms use “a priori” information from ozone profiles to retrieve TCO from backscattered UV measurements19. Due to the large longitudinal variability of TrCO, the improvement of the TROPOMI/S5P algorithm includes Ziemke et al.20 improves the regulation of the troposphere11. However, ozone climatology uses profiles mostly at sea level in the Atlantic and Pacific basins, while profiles high above sea level in the tropical Andes deviate from these locations mainly in the troposphere. Thus, previous studies show significant differences in profile structure and TrCO (lower) magnitude from Galapagos (Ecuador) and Natal (Brazil) stations, while stratospheric ozone is similar and consistent with Microwave Limb Sounder (MLS/Aura) measurements.12,13. We therefore compared mean differences in TCO, TrCO, and stratospheric column ozone (SCO) between the Ziemke climate and the EMA Quito collection of observations (Table S1). While the SCO is practically the same (+ 1.3 DU or 0.6% difference), the TCO by climate shows a mean deviation of + 9.3 DU (3.8% mean difference) due to an overprediction of tropospheric ozone (+ 8 DU or 42.5%). This result is similar to the previously discussed comparison of TROPOMI/S5P individual measurements with ozonesondes. Although further research and more data are needed, our findings point out that profiles in the Andean tropics need to be included in satellite climatology, especially to better describe TrCO in a region that would have an overall positive effect on TCO.
Regarding the reanalysis products, MERRA-2 provides a more accurate comparison (Fig. 2i,j) with the observations than CAMS (Fig. 2k,l), as the average bias is +4.5 DU compared to +8 DU (less than 2% vs. 3.3 % mean difference). Data assimilation in MERRA-2 includes TCO from OMI/Aura and stratospheric mixing ratios from MLS/Aura21. Meanwhile, CAMS adds the TCO of GOME-2/MetOP-B to former ozone sources22. From an observational perspective, MERRA-2 outperforms CAMS in estimating TCO, but further studies should be conducted to better pinpoint specific differences between both models and the data sources they use.
Finally, we quantify the differences of all products in relation to OMPS/Finland’s nuclear power plant, as it is the data set that compares the best in situ observations. As shown in Fig. S2, the average differences for TROPOMI/S5P, GOME-2/MetOP-B, and OMI/Aura are 3.8%, 3.5%, and 1.4%, respectively. As for the reanalysis, the mean differences between MERRA-2 and CAMS are 1.1% and 3%, respectively. That’s why OMI/Aura and MERRA-2 are best compatible with OMPS/Finland’s nuclear power plant.
Tropospheric Ozone Comparison
The time series of TrCO for data sets with daily observations is shown in Figure 3, albeit at pressure levels suitable for making comparisons. For example, TROPOMI/S5P is available from ground level (760 hPa) down to 270 hPa, which corresponds to an altitude of about 10 km. However, from the high-resolution profiles, the Andean tropical troposphere is located at 96 hPa (17 ± 0.7 km) (see “Methods,” Table S2). In addition, previous work shows that the tropopause level identified by the chemical definition (“Methods” section) mostly coincides with the coldest point of the temperature profile13. This shows that unlike the Pacific and Atlantic23,24, ozone in the Andean tropics is generally well mixed up to the level of the tropopause. Although 270 hPa in TROPOMI/S5P is suitable for determining TrCO at midlatitudes, it neglects the rest of the tropospheric column at the study site. We have not found a specific explanation in the literature as to why the nominal pressure of the ready-to-use TrCO product is 270 hPa all over the globe.2.11. We believe this may be somewhat misleading especially for end users living in the Andean tropics. Similarly, TrCO from GOME-2/MetOP-B is available up to 200 hPa, albeit as monthly averages (Fig. S3). In contrast, reanalysis products are available as mass mixing ratios at pressure level intervals from the surface throughout the atmospheric column. Thus, comparisons that capture the entire Andean tropical troposphere can be performed by integrating data down to 100 hPa.
Comparing TrCO from TROPOMI/S5P to ozone sondes (integrated up to 270 hPa) gives a positive bias of + 3.6 DU or a mean difference of 32.5% (Figs. 4a,b). Previous studies indicate that tropospheric ozone in the tropical Andes is low because altitude subtracts 5–7 DU of TrCO, while boundary layer ozone is also low.13. However, TROPOMI/S5P generally measures higher values, even though the observations correspond to a fraction of the column. Recent studies evaluating the quality of TROPOMI/S5P TrCO against ozonesondes in the tropics also found a bias. For example, differences at several sea level stations were found to be + 4 DU (up to 22% higher) when the data were averaged over two years, while the overall positive bias was + 2.3 DU (or 11%) when the data were smoothed across longitudes2. The reason for this positive bias was reported to be not fully understood and was partly due to possible systematic differences in the time of measurement, assuming that TrCO follows a diurnal pattern. We also report on a positive bias in the Andean tropics, which also requires further investigation. In part, this overestimation is likely due to satellite climatology overpredicting ozone in the Andean troposphere, as discussed in the previous section. However, further comparisons with additional data are needed in the future to better understand the nature and persistence of these differences.
MERRA-2 TrCO compared to soundings down to 100 hPa provides the best comparison in the troposphere (Figs. 4c,d). The deviation with respect to the ozone probes is + 1.5 DU (11.5% difference), which we consider low considering that it is difficult to capture TrCO correctly at this very complex site where validations have not been done before. In the case of CAMS, the integration was also done up to 100 hPa. The average deviation (Fig. 4e,f) is + 3.5 DU (23%), which doubles MERRA-2, but the linear regression correlation coefficient is higher (R2= 0.8 vs 0.6). In contrast, the TrCO of GOME-2/MetOP-B is only available to the end user as monthly averages. Thus, this product was only qualitatively compared to EMA in the time series (Fig. S3), but the data are insufficient to draw quantitative conclusions.
Finally, from an end-user perspective, currently MERRA-2 TrCO is best suited for the Andean tropics. First, because data integration can be done for a pressure level that captures the entire Andean tropospheric column. Second, because the differences in observations are the least.