the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Jul 2025
Characterization of the UV radiometric calibration for the TROPOMI operational ozone profile retrieval algorithm
Abstract. The European Space Agency (ESA) Sentinel-5 Precursor (S5P) is a low Earth orbit polar satellite carrying the single payload instrument TROPOspheric Monitoring Instrument (TROPOMI). Since its launch on 13 October 2017, the S5P mission has been acquiring almost 7 years of nadir ozone profile data, retrieved from the UV spectral bands 1–2 (270–330 nm). The retrieval algorithm of the ozone profile can be strongly affected by systematic effects in the measured radiance, and absolute calibration of the input spectra is necessary. In this study, we characterize the TROPOMI bands 1–2 radiometric bias in comparison with simulations obtained with the Determining Instrument Specifications and Analysing Methods for Atmospheric Retrieval (DISAMAR) radiative transfer model. To account for these systematic effects, a radiometric correction (soft-calibration) is applied to the input measurements, which results in the reduction of the spectral reflectance fit residuals of 20–30 %, improved precision of the integrated total and tropospheric ozone columns of 10–15 % and a reduction of along-track orbit artifacts. Together with the analysis of the in-flight calibration measurements, the soft-calibration correction spectra provide also useful insights for the improvement of the instrument radiometric calibration. Therefore, bands 1–2 measurements have been reprocessed specifically for this study, with improvements regarding the detector straylight and background signal correction algorithms, to investigate the impact of the updates into the TROPOMI bands 1–2 radiometric bias. The new soft-calibration correction spectra, obtained with the reprocessed bands 1–2 measurements, show significantly reduced magnitude (around 15–20 %, especially in band 1) and also less across-track position and spectral/temporal biases. The new soft-calibration will be part of ESA’s next official ozone profile algorithm version 2.9.0, coordinated with L1b update to processor version 3.0.0, and used for the second TROPOMI mission reprocessing.
- Preprint
(35311 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2025-2167', Juseon Bak, 19 Jul 2025
-
RC2: 'Comment on egusphere-2025-2167', Glen Jaross, 07 Aug 2025
In general, I find this manuscript version improved over the initial submission. The authors have done a better job of organizing results to make clear when corrections are and are not applied. More explicit statements for each figure would still be appreciated. It is clear that the new instrument characterization described in this paper is improving the accuracy of the radiance data (thereby reducing the magnitude of soft calibration corrections), but some of the details regarding how those improvements were achieved is lacking. It is only by providing such details that readers can evaluate the quality of the characterizations and learn something about instrument performance.
Section 2.2, Lines 135-138
This text describes the contents of Figure 2, which presents a compelling case that the Mg line signals have increased over time relative to background signals. Unfortunately, the reader has no idea from this figure what the time scales are, since the paper provides no assignment of dates for S5P orbit numbers. Two common reasons these Fraunhofer line changes occur is solar activity and improper correction for additive signal errors. The most likely additive errors are detector dark current and spectral stray light. All three phenomena are valid explanations for why the Mg line depths shown in Fig. 2 are decreasing. The authors state that the Fig. 2 results are independent of solar activity effects, but offer no details supporting this statement. Was a correction applied, and if so what was it? They do not discuss their background signal corrections, which must also increase as the instrument ages. These are glaring omissions if they expect the reader to accept the most surprising conclusion, that internally-scattered stray light is increasing within TropOMI. There are few plausible mechanisms for postlaunch increases in spectrometer scattering (primarily caused by the grating), so the authors need a convincing argument why the least likely of the 3 explanations is in fact the actual cause of the observed signal changes. One convincing metric would be to compare the Mg II core-to-wing ratios from solar irradiance and Earth radiance spectra at different times in the mission. In the absence of additive errors these two ratios should remain in lock step with each other. If, however, spectral stray light is increasing, the effect on Earth radiances should be much greater than on solar irradiances.Section 2.2.1, Line 153
The statement that the SL correction algorithm is a 2D correction that only addresses in-band stray light is confusing. I suspect the authors mean the correction address both spatial and spectral stray light, but the latter is limited to only photons from within the same band. If so, a clarification of this point would be helpful.Section 2.2.1, Lines 179-189
The authors should provide a clearer explanation of how the SL convolution kernels are adjusted. What criteria were used to assess the correct spectral and spatial distributions? Scattering occurring after the entrance slit typically has no preferential spatial or spectral direction. The 2D scattering PSF is usually symmetric. Therefore, the large difference in the two orthogonal cross sections of that PSF (shown in Fig. 3) is rather surprising and the authors are encouraged to provide more detail about why they feel the shapes should behave so differently. I presume that telescope stray light (i.e. pre-slit scattering) has been excluded from the plots in Fig. 3a. Please state so explicitly.Section 2.4
The authors mention the introduction of time-varying SL correction to account for the increasing Mg II signals and the SL row signals, but they provide no details about how this is implemented. Are the kernel shapes altered in time? If so, how? Are the spatial and spectral components of the kernel locked relative to each other or allowed to vary independently? Whatever the approach, it will be helpful if the authors can demonstrate the effect of these dynamic corrections on the metrics shown in Figures 2 and 4. Figure 10 does indicate the effect on RTM residuals of these instrument corrections, but it does not address the effect on irradiance or stray light rows.Section 4.3
Regarding the observed residuals (as represented by soft calibration corrections) presented in Figures 7 & 8 the authors do not discuss whether or not a solar activity correction has been applied. The increase seen in Figure 7a is consistent with the activity increase over the mission timeline. Figure 8 also appears to be consistent with the increase. An accurate solar activity correction must be applied prior to residual analysis, therefore the authors should provide details about any correction that was applied. What data was it derived from? What is the magnitude of the changes in time at key band 1 wavelengths?Citation: https://doi.org/10.5194/egusphere-2025-2167-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 389 | 25 | 15 | 429 | 12 | 18 |
- HTML: 389
- PDF: 25
- XML: 15
- Total: 429
- BibTeX: 12
- EndNote: 18
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Attached