Greenhouse gas column observations from a portable spectrometer in Uganda

Humpage, Neil; Boesch, Hartmut; Okello, William; Chen, Jia; Dietrich, Florian; Lunt, Mark F.; Feng, Liang; Palmer, Paul I.; Hase, Frank

doi:https://doi.org/10.5194/amt-2023-234

Preprints

https://doi.org/10.5194/amt-2023-234

Preprints

28 Nov 2023

| 28 Nov 2023

Status: this preprint is currently under review for the journal AMT.

Greenhouse gas column observations from a portable spectrometer in Uganda

Neil Humpage, Hartmut Boesch, William Okello, Jia Chen, Florian Dietrich, Mark F. Lunt, Liang Feng, Paul I. Palmer, and Frank Hase

Abstract. The extensive terrestrial ecosystems of tropical Africa are a significant store of carbon, and play a key but uncertain role in the atmospheric budgets of carbon dioxide and methane. As ground-based observations in the tropics are scarce compared with other parts of the world, recent studies have instead made use of satellite observations assimilated into atmospheric chemistry and transport models to conclude that methane emissions from this geographical region have increased since 2010 as a result of increased wetland extent, accounting for up to a third of global methane growth, and that the tropical Africa region dominates net carbon emission across the tropics. These studies rely critically on the accuracy of satellite datasets such as those from OCO-2, GOSAT, and Sentinel-5P TROPOMI, along with results from atmospheric transport models, over a geographical region where there are few independent data to test the robustness of published results.

In this paper we present the first ground-based observations of greenhouse gas column concentrations over East Africa, obtained using a portable EM27/SUN FTIR spectrometer during a deployment covering the first few months of 2020 in Jinja, Uganda. We operated the instrument near-autonomously by way of an automated weatherproof enclosure, and observed total atmospheric column concentrations of the greenhouse gases carbon dioxide and methane, as well as carbon monoxide, a useful proxy for emissions from incomplete combustion processes in the region. We discuss the performance of the combined enclosure and spectrometer system that we deployed in Jinja to obtain this data, and show comparisons of our ground-based observations with satellite datasets from OCO-2 and OCO-3 for carbon dioxide, and Sentinel-5P TROPOMI for methane and carbon monoxide, whilst also comparing our results with concentration data from the GEOS-Chem and CAMS atmospheric inversions that provide a means of increasing spatial and temporal coverage where satellite data are not available. For our measurement period, we find statistically significant differences at the 95 % confidence level between the EM27/SUN and OCO-2 X_CO₂ (OCO-2 lower by a mean of 1.20 ppm, standard deviation 1.05 ppm), and between the EM27/SUN and Sentinel-5P X_CO (Sentinel-5P lower by a mean of 3.68 ppb, standard deviation 7.00 ppb), whilst we found that the differences between the EM27/SUN and OCO-3 X_CO₂ (OCO-3 lower by a mean of 1.15 ppm, standard deviation 1.61 ppm), and between the EM27/SUN and Sentinel-5P X_CH₄ (Sentinel-5P lower by a mean of 8.33 ppb, standard deviation 10.5 ppb), were not statistically significant. With regards to the model comparisons, we also see statistically significant differences between the EM27/SUN and a global GEOS-Chem inversion for X_CO₂ (GEOS-Chem lower by a mean of 0.35 ppm, standard deviation 1.08 ppm), between the EM27/SUN and a high-resolution GEOS-Chem inversion for X_CH₄ (GEOS-Chem lower by a mean of 3.80 ppb, standard deviation 12.5 ppb), and between the EM27/SUN and CAMS global analysis X_CO (CAMS lower by a mean of 11.7 ppb, standard deviation 8.94 ppb). Our results demonstrate the value of ground-based observations of total column concentrations, and show that the combined EM27/SUN and enclosure system employed would be suitable for acquisition of the longer-term observations needed to rigorously evaluate satellite observations and model calculations over tropical Africa.

Received: 10 Nov 2023 – Discussion started: 28 Nov 2023

Download & links

Neil Humpage, Hartmut Boesch, William Okello, Jia Chen, Florian Dietrich, Mark F. Lunt, Liang Feng, Paul I. Palmer, and Frank Hase

Status: final response (author comments only)

RC1:
'Comment on amt-2023-234', Anonymous Referee #1, 24 Jan 2024
In their study, the authors present ground-based observations of carbon dioxide, methane, and carbon monoxide performed by a Bruker EM27/SUN Fourier transform infrared spectrometer in Jinja (Uganda) during the first months of 2020 (23rd January to April 2020). As introduced by the authors, ground-based observations are scarce in tropical Africa, while being needed to evaluate satellite and model datasets. The instrumental setup, retrieval method, and QA/QC are thoroughly described. The authors then include and describe several spaceborne datasets (OCO-2, OCO-3, and TROPOMI) as well as model and reanalysis datasets (GEOSChem and CAMS). They finally propose a comparison of the ground-based time series with the other included datasets. The manuscript is well-written, and the sections are appropriate. The methods are thoroughly described and robust. Results are well discussed and concluded. Therefore, I recommend publication in AMT after my minor comments below are addressed.

Minor comments:
My only comment on the methods is on the choice to perform t-tests to statistically test the mean difference between the ground-based retrievals and the other datasets. For CO2 and CH4, there are only 8 days that meet the coincidence criteria (which is already quite "loose"). Many ground-based and satellite retrievals are considered in the daily medians, but, in the end, I understood that the test is performed on the 8 median values available. My question is: is the t-test robust when used with such small sample sizes? Reading tables 1, 2, 3, one could easily conclude that the mean differences are relatively small and most of the time within their respective standard deviations. Would it not be more interesting and easier for the reader to only have these two statistics with their relative values (%) in addition?

Sometimes the justifications for not having statistical differences are weak to me. Page 16, lines 360-362, and Page 20 lines 388-389, the authors state that the noted differences are not statistically significant because of the small sample sizes. Was something used to assess the effect of the sample size or is it just an interpretation of the results from the authors? For the specific case of CH4 in Table 2, in the GEOSChem including TROPOMI data comparison, there are 27 and 31 compared days for the dry and rainy seasons, respectively. While these sample sizes are still relatively small, they are substantially larger than that of satellite comparisons (OCO-2, OCO-3, and TROPOMI for CO2 and CH4). So, how can the authors attribute the non-statistically significant differences to the sample sizes in that case?

At the end of Section 2.3, it could be interesting to have statistics on the number of retrievals discarded by the QA/QC.

Typo/editing comments/suggestions:
Page 7 lines 186-187: I would replace the occurrences of "radiance spectra" with "solar absorption spectra."

Subsections 3.1 and 3.2: It could be clearer for the reader if the subsection titles included the species retrieved by these satellites. For example, 3.1 could become "Orbiting Carbon Observatory (OCO-2 and OCO-3) XCO2 retrievals," and 3.2 "Sentinel-5 Precursor XCH4 and XCO retrievals." This is only a suggestion.

CAMS: CAMS should be referred to as a (atmospheric composition) reanalysis as it is not strictly a model or an inversion. This should be harmonized through the manuscript.

TROPOMI: I would replace all occurrences of "Sentinel-5P TROPOMI" (and similar occurrences, e.g., "S5P Copernicus" in Table 2) with "TROPOMI." Whatever is the author's choice, it should be harmonized throughout the manuscript.

Page 12, line 285, aren’t -> are not

Tables 1, 2, and 3: The total number of soundings could be replaced by "Total number of satellite soundings" or "Total number of satellite retrievals."
Citation: https://doi.org/10.5194/amt-2023-234-RC1
- AC1:
  'Reply on RC1', Neil Humpage, 08 Mar 2024
  We would firstly like to thank the Anonymous Referee for their constructive feedback on the manuscript. As a result of their comments we propose to make the following changes to the final version of the manuscript:
  On the choice of t-tests to assess statistical significance given the small sample size, there is literature concluding that "there are no principal objections to using a t-test with Ns as small as 2" (de Winter, 2019), although they do qualify this with "as long as the effect size is expected to be large". In these comparisons the relative differences between the EM27/SUN measurements and the satellite/model/reanalysis datasets are small, so we think that especially where we only have a handful of samples we are overstating the statistical significance of some of these differences. We agree with the referee that it would be better and clearer to present the comparisons in terms of mean difference, standard deviation and relative difference -- we will update the tables, and the results section to reflect this. The second comment regarding the justifications for not having statistically significant differences is related to this, so we will revisit and update these parts of the results sections to avoid overinterpreting statistical significance (or lack thereof).
  Regarding the suggestion to include statistics on the number of retrievals discarded by the QA/QC, we agree and will add a summary to the end of Section 2.3.
  We will also address the following typo/editing comments/suggestions:
  Page 7 lines 186-187: I would replace the occurrences of "radiance spectra" with "solar absorption spectra." -- we will replace these as suggested.
  
  Subsections 3.1 and 3.2: It could be clearer for the reader if the subsection titles included the species retrieved by these satellites. For example, 3.1 could become "Orbiting Carbon Observatory (OCO-2 and OCO-3) XCO2 retrievals," and 3.2 "Sentinel-5 Precursor XCH4 and XCO retrievals." This is only a suggestion -- we agree with this suggestion, and will rename the subsections to make them clearer for the reader.
  
  CAMS: CAMS should be referred to as a (atmospheric composition) reanalysis as it is not strictly a model or an inversion. This should be harmonized through the manuscript -- we will consistently refer to CAMS as a reanalysis throughout the manuscript, as suggested.
  
  TROPOMI: I would replace all occurrences of "Sentinel-5P TROPOMI" (and similar occurrences, e.g., "S5P Copernicus" in Table 2) with "TROPOMI." Whatever is the author's choice, it should be harmonized throughout the manuscript -- other than when it is first mentioned, where we make it clear the TROPOMI is on board Sentinel-5P, we will consistently refer to TROPOMI throughout the manuscript (including the Tables and in legends in Figures).
  
  Page 12, line 285, aren’t -> are not -- we will replace as suggested.
  
  Tables 1, 2, and 3: The total number of soundings could be replaced by "Total number of satellite soundings" or "Total number of satellite retrievals." -- in Tables 1, 2, and 3, we will replace this column header with "Total number of satellite retrievals"
  
  Reference:
  de Winter, J.C.F. (2019) "Using the Student's t-test with extremely small sample sizes," Practical Assessment, Research, and Evaluation: Vol. 18, Article 10. DOI: https://doi.org/10.7275/e4r6-dj05
  
  Citation: https://doi.org/10.5194/amt-2023-234-AC1
RC2:
'Comment on amt-2023-234', Shima Bahramvash Shams, 25 Jan 2024
This manuscript presents the retrieval results of methane, CO2, and CO from the ground based portable FTIR spectrometer in East Africa, Jinja-Uganda. There is a great value in fairly automated portable FTIR in remote region and this work presents the first of its kind in the East Africa. They compared the retrieval results with satellite data including OCO-2 and OCO-3 for carbon dioxide, and Sentinel-5P TROPOMI for methane and carbon monoxide as well as atmospheric chemistry models including GEOS-Chem and CAMS. They applied a dispersion model to provide a thorough characterization of emission sources that influence the retrieved values. They reported that retrieved CO2 from FTIR was in better agreement with OCO-3 compared to OCO-2. CO retrievals were significantly higher than TROPOMI. Methane retrieval where consistent with satellite driven values.

The paper is well-written, and the result discussion is comprehensive. However, there are a few major comments that need to be addressed. The paper lacks a detailed description of the FTIR and retrieval characteristics. Additionally, it is crucial to include sensitivity analysis for comparison with OCO-2 and OCO-3, considering the large smoothing effect created by averaging over a radius of 300 km. This is especially important as spatial asymmetry seems to exist in available OCO-2 and OCO-3 pixels, potentially impacting the paper's comparisons for CO2. Detailed comments are provided below, and as a result, my final decision is a minor revision.

While the main goal of the paper is to highlight the importance of portable FTIR in atmospheric research there is insufficient detail provided about the characteristics of the FTIR instrument. I suggest that the authors include details such as detector spectral ranges, optical system design, optical path difference, spectral resolution, temporal resolution, and etc. Additionally, figures for average kernels and DOF are needed to help readers become familiar with this particular FTIR and its sensitivity. Also, statistics of signal to noise ratio would help justify whether temporal averaging of the spectra is necessary or not.

In comparison with OCO2 and OCO3 for CO2 assessment, authors applied co-location averaging. Their argument is the gap in the satellite data. Does it mean that OCO2 never exactly overpasses the FTIR site as shown in Fig.4? It seems that there is an exact pixel that overlaps with the FTIR site from OCO3. Why not using the exact pixel that matches for OCO3 and then for OCO2, estimating the grid point by weighted averaging or similar methods? Averaging over a circle of a 300 km radius, presents a significant spatial smoothing effect. Also as shown in Fig. 4, available pixels are mostly on the east side of FTIR site, which could shift the average toward an unrealistic value compared to what is the estimation for the exact overlay grid. A sensitivity analysis is required to make sure this spatial asymmetry is not affecting their assessment results.

Minor comments:

-P4, L116: please add some references for the validation works using TCCON site.

-P8, L189: Please be specific about the sources of A priori for the retrieval, which model or climatology they are made off.

-Figure 3_and Table 1: I would change the left-hand side label from number of soundings to number of scan or successful retrieval as sounding is mostly referred to in situ measurement and it is confusing if there are in situ observation available.
Citation: https://doi.org/10.5194/amt-2023-234-RC2
- AC2: 'Reply on RC2', Neil Humpage, 08 Mar 2024
  
  We would firstly like to thank Shima Bahramvash Shams for their constructive feedback on the manuscript. As a result of their comments we propose to make the following changes to the final version of the manuscript:
  1. The referee suggests that there is insufficient detail provided about the instrument characteristics and retrieval. Whilst we feel this material is well covered in the references provided, we are happy to add an appendix covering the characteristics mentioned (example spectra, information on spectral range, resolution, and signal-to-noise ratio, and plots of averaging kernels to illustrate vertical sensitivity) to help a reader unfamiliar with the instrument to understand the measurements.
  2. We accept that there is a spatial smoothing effect that comes with co-location averaging. It is a common problem when comparing ground-based observations with satellite data that the satellite sampling pattern often does not fall exactly on the ground-based spectrometer location, which is why previous studies of a similar nature have taken the spatial co-location averaging approach (Inoue et al 2016, Wunch et al 2017, Sha et al 2021). To demonstrate the validity of this approach and justify the selection of the co-location radii used, we will add an appendix showing the impact of varying the coincidence criteria on the mean and standard deviation of the satellite vs. ground-based column concentration difference.
  We will also address the minor comments, as follows:
  
  -P4, L116: please add some references for the validation works using TCCON site -- we will add appropriate references here (Inoue et al 2016, Wunch et al 2017, Sha et al 2021)
  
  -P8, L189: Please be specific about the sources of A priori for the retrieval, which model or climatology they are made off -- we will add further detail and references on how the a priori profiles are generated here, as suggested.
  
  -Figure 3_and Table 1: I would change the left-hand side label from number of soundings to number of scan or successful retrieval as sounding is mostly referred to in situ measurement and it is confusing if there are in situ observation available -- we will change both of these to 'Successful Retrievals' to make the Figure and Table clearer to the reader, as suggested.
  
  References:
  
  Inoue, M., Morino, I., Uchino, O., Nakatsuru, T., Yoshida, Y., Yokota, T., Wunch, D., Wennberg, P. O., Roehl, C. M., Griffith, D. W. T., Velazco, V. A., Deutscher, N. M., Warneke, T., Notholt, J., Robinson, J., Sherlock, V., Hase, F., Blumenstock, T., Rettinger, M., Sussmann, R., Kyrö, E., Kivi, R., Shiomi, K., Kawakami, S., De Mazière, M., Arnold, S. G., Feist, D. G., Barrow, E. A., Barney, J., Dubey, M., Schneider, M., Iraci, L. T., Podolske, J. R., Hillyard, P. W., Machida, T., Sawa, Y., Tsuboi, K., Matsueda, H., Sweeney, C., Tans, P. P., Andrews, A. E., Biraud, S. C., Fukuyama, Y., Pittman, J. V., Kort, E. A., and Tanaka, T.: Bias corrections of GOSAT SWIR XCO₂ and XCH₄ with TCCON data and their evaluation using aircraft measurement data, Atmos. Meas. Tech., 9, 3491–3512, https://doi.org/10.5194/amt-9-3491-2016, 2016.
  
  Wunch, D., Wennberg, P. O., Osterman, G., Fisher, B., Naylor, B., Roehl, C. M., O'Dell, C., Mandrake, L., Viatte, C., Kiel, M., Griffith, D. W. T., Deutscher, N. M., Velazco, V. A., Notholt, J., Warneke, T., Petri, C., De Maziere, M., Sha, M. K., Sussmann, R., Rettinger, M., Pollard, D., Robinson, J., Morino, I., Uchino, O., Hase, F., Blumenstock, T., Feist, D. G., Arnold, S. G., Strong, K., Mendonca, J., Kivi, R., Heikkinen, P., Iraci, L., Podolske, J., Hillyard, P. W., Kawakami, S., Dubey, M. K., Parker, H. A., Sepulveda, E., García, O. E., Te, Y., Jeseck, P., Gunson, M. R., Crisp, D., and Eldering, A.: Comparisons of the Orbiting Carbon Observatory-2 (OCO-2) XCO2 measurements with TCCON, Atmos. Meas. Tech., 10, 2209–2238, https://doi.org/10.5194/amt-10-2209-2017, 2017.
  
  Sha, M. K., Langerock, B., Blavier, J.-F. L., Blumenstock, T., Borsdorff, T., Buschmann, M., Dehn, A., De Mazière, M., Deutscher, N. M., Feist, D. G., García, O. E., Griffith, D. W. T., Grutter, M., Hannigan, J. W., Hase, F., Heikkinen, P., Hermans, C., Iraci, L. T., Jeseck, P., Jones, N., Kivi, R., Kumps, N., Landgraf, J., Lorente, A., Mahieu, E., Makarova, M. V., Mellqvist, J., Metzger, J.-M., Morino, I., Nagahama, T., Notholt, J., Ohyama, H., Ortega, I., Palm, M., Petri, C., Pollard, D. F., Rettinger, M., Robinson, J., Roche, S., Roehl, C. M., Röhling, A. N., Rousogenous, C., Schneider, M., Shiomi, K., Smale, D., Stremme, W., Strong, K., Sussmann, R., Té, Y., Uchino, O., Velazco, V. A., Vigouroux, C., Vrekoussis, M., Wang, P., Warneke, T., Wizenberg, T., Wunch, D., Yamanouchi, S., Yang, Y., and Zhou, M.: Validation of methane and carbon monoxide from Sentinel-5 Precursor using TCCON and NDACC-IRWG stations, Atmos. Meas. Tech., 14, 6249–6304, https://doi.org/10.5194/amt-14-6249-2021, 2021.
  
  Citation: https://doi.org/10.5194/amt-2023-234-AC2

Neil Humpage, Hartmut Boesch, William Okello, Jia Chen, Florian Dietrich, Mark F. Lunt, Liang Feng, Paul I. Palmer, and Frank Hase

Viewed

Total article views: 350 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
259	72	19	350	18	13

HTML: 259
PDF: 72
XML: 19
Total: 350
BibTeX: 18
EndNote: 13

Views and downloads (calculated since 28 Nov 2023)

Month	HTML	PDF	XML	Total
Dec 2023	66	25	3	94
Jan 2024	59	16	4	79
Feb 2024	39	8	2	49
Mar 2024	75	17	8	100
Apr 2024	20	6	2	28

Cumulative views and downloads (calculated since 28 Nov 2023)

Month	HTML	PDF	XML	Total
Dec 2023	66	25	3	94
Jan 2024	59	16	4	79
Feb 2024	39	8	2	49
Mar 2024	75	17	8	100
Apr 2024	20	6	2	28

Viewed (geographical distribution)

Total article views: 362 (including HTML, PDF, and XML) Thereof 362 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 15 Apr 2024

Short summary

We used a Bruker EM27/SUN spectrometer within an automated weatherproof enclosure to measure greenhouse gas column concentrations over a three-month period in Jinja, Uganda. The portability of the EM27/SUN allows us to evaluate satellite and model data in locations not covered by traditional validation networks. This is of particular value in tropical Africa, where extensive terrestrial ecosystems are a significant store of carbon and play a key role in the atmospheric budgets of CO₂ and CH₄.


Total:	0
HTML:	0
PDF:	0
XML:	0