the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
12 Mar 2021
12 Mar 2021
Spatial distributions of XCO2 seasonal cycle amplitude and phase over northern high latitude regions
- 1Department of Chemistry and the Geophysical Institute, University of Alaska Fairbanks, Fairbanks, AK, USA
- 2Department of Earth, Ocean, and Atmospheric Science, Florida State University, Tallahassee, FL, USA
- 3Karlsruhe Institute of Technology (KIT), Institute of Meteorology and Climate Research, Karlsruhe, Germany
- 4National Institute for Environmental Studies, Tsukuba, Japan
- 5Earth and Environmental Sciences, Los Alamos National Laboratory, Los Alamos, NM, USA
- 6California Institute of Technology, Pasadena, CA, USA
- 7Department of Physics, University of Toronto, Toronto, Canada
- 8Finnish Meteorological Institute, Sodankylä, Finland
- 9Institute of Environmental Physics, University of Bremen, Germany
- 1Department of Chemistry and the Geophysical Institute, University of Alaska Fairbanks, Fairbanks, AK, USA
- 2Department of Earth, Ocean, and Atmospheric Science, Florida State University, Tallahassee, FL, USA
- 3Karlsruhe Institute of Technology (KIT), Institute of Meteorology and Climate Research, Karlsruhe, Germany
- 4National Institute for Environmental Studies, Tsukuba, Japan
- 5Earth and Environmental Sciences, Los Alamos National Laboratory, Los Alamos, NM, USA
- 6California Institute of Technology, Pasadena, CA, USA
- 7Department of Physics, University of Toronto, Toronto, Canada
- 8Finnish Meteorological Institute, Sodankylä, Finland
- 9Institute of Environmental Physics, University of Bremen, Germany
Abstract. Satellite-based observations of atmospheric carbon dioxide (CO2) provide measurements in remote regions, such as the biologically sensitive but under sampled northern high latitudes, and are progressing toward true global data coverage. Recent improvements in satellite retrievals of total column-averaged dry air mole fractions of CO2 (XCO2) from the NASA Orbiting Carbon Observatory 2 (OCO-2) have allowed for unprecedented data coverage of northern high latitude regions, while maintaining acceptable accuracy and consistency relative to ground-based observations, and finally providing sufficient data in spring and autumn for analysis of the satellite-observed XCO2 seasonal cycles across a majority of terrestrial northern high latitude regions. Here, we present an analysis of XCO2 seasonal cycles calculated from OCO-2 data for temperate, boreal, and tundra regions, subdivided into 5° latitude by 20° longitude zones. We quantify the seasonal cycle amplitudes (SCA) and the annual half drawdown day (HDD). OCO-2 SCA is in good agreement with ground-based observations at five high latitude sites and OCO-2 SCA show very close agreement with SCA calculated for model estimates of XCO2 from the Copernicus Atmospheric Monitoring Services (CAMS) global inversion-optimized greenhouse gas flux model v19r1. Model estimates of XCO2 from the GEOS-Chem CO2 simulation version 12.7.2 with underlying biospheric fluxes from CarbonTracker2019 yield SCA of larger magnitude and spread over a larger range than those from CAMS and OCO-2; however, GEOS-Chem SCA still exhibit a very similar spatial distribution across northern high latitude regions to that from CAMS and OCO-2. Zones in the Asian Boreal Forest were found to have exceptionally large SCA and early HDD, and both OCO-2 data and model estimates yield a distinct longitudinal gradient of increasing SCA from west to east across the Eurasian continent. Longitudinal gradients in both SCA and HDD are at least as pronounced as meridional gradients (with respect to latitude), suggesting an essential role for global atmospheric transport patterns in defining XCO2 seasonality. GEOS-Chem surface contact tracers show that the largest XCO2 SCA occurs in areas with the greatest contact with land surfaces, integrated over 15–30 days. The correlation of XCO2 SCA with these land contact tracers are stronger than the correlation of XCO2 SCA with the SCA of CO2 fluxes within each 5° latitude by 20° longitude zone. This indicates that accumulation of terrestrial CO2 flux during atmospheric transport is a major driver of regional variations in XCO2 SCA.
- Preprint
(9152 KB) -
Supplement
(25745 KB) - BibTeX
- EndNote
Nicole Jacobs et al.
Status: final response (author comments only)
-
RC1: 'Comment on acp-2021-185', Anonymous Referee #1, 11 Apr 2021
This study examines spatial variability in the seasonal cycle of column-averaged dry-air mole fractions of CO2 (XCO2) across the arctic-boreal zone. This is analysis is performed with XCO2 retrievals from ground-based instruments and OCO-2, as-well as simulated XCO2 using two chemical transport models. The authors find that the amplitude of the seasonal cycle is largest and half drawdown day is earliest over eastern Eurasia due to a combination of surface fluxes and transport. The carbon cycle of the northern latitudes is an important area of research, and this analysis furthers our understanding of this region. However, there are a number of major issues with the analysis and manuscript that need to addressed before I can recommend publication. I have outlined my concerns below.
General comments
- Many of the seasonal cycle fits shown in Fig. S4-S11 appear to be quite unphysical. Thus, it is unclear if the analysis is really capturing accurate SCA estimates. There should be uncertainty quantification in the SCA fits, perhaps using bootstrap resampling or another technique. Ideally, the analysis could also be performed fitting truncated Fourier series, to test the impact of the functional form on the results.
- I was not able to understand if the SCA analysis accounts for temporal sampling differences between OCO-2 and the model simulations. It appears that the models are sampled daily throughout the year. I think that it is quite likely that the SCA fits will be quite sensitive to the observational sampling. The sensitivity of SCA estimates to temporal sampling should be quantified.
- The contact tracer analysis was insufficiently described. I could not understand what this analysis was telling us. When are the tracers released? And what amount? How long is the simulation run? What precisely is being shown in Figure 10 (tracer was plotted at what time? for simulation starting on what day? that released what quantity of tracer? And released it over what spatiotemporal window?)?
- The manuscript is quite hard to follow in places. It would help to explicitly describe the subpanels in the figures, for example, “(a) Quantity Y versus quantity X with Z processing”. Please also ensure that the main results from the figures are described in the text when the figure is referenced.
- I think that the impact of this study would be improved if the OCO-2 retrievals with the standard bias correction and filtering were presented throughout the main text in addition to the high-latitude focused bias correction and filtering. As a potential user of these data, I am very interested in understanding the impact of differences in bias correction and filtering. From Fig. S1, it appears that differences are substantial. Furthermore, it would be of interest to determine if differences between different QC/bias-correction result in larger SCA differences than between the models.
Specific comments
P1L12: It is a little confusing to refer to a GEOS-Chem run with CT2019 fluxes as “GEOS-Chem”. It would be better to use a specific acronym such as “GC-CT2019” to make clear that it is GEOS-Chem run with CT2019.
P1L16-17: This is only for >50N that is examined here. The meridional gradient is still greater from ~0 ppm at the equator to >10ppm at the North pole.
P1L16-17: Reads strange to use “Longitudinal” and “meridional”. I suggest using “zonal” instead of “longitudinal”.
Sec 2.4: I do not see how XCO2 is calculated. Is an averaging kernel applied or is it the true XCO2?
Sec 3.1: I found the main points of Sec 3.1 quite unclear. It would be helpful to walk the reader through the results. The section starts by stating that differences in spatial sampling may impact the results, but from reading the rest of the section I am not clear on the impact of spatial sampling on the results. It would be helpful to explicitly state the results of this comparison, and what are the implications for the analysis that follows. In particular, I’m having a hard time understanding what Fig. 2 and Fig. 3 are telling us (please explicitly state what the sub-panels are showing). What does Fig. 3 show us that Fig. 2 does not? And why are there TCCON symbols for the OCO-2 vs model comparison?
P10L9-10: “Results in the supplement (see Fig. S30) indicate that SCA derived from clipped time-series of OCO-2 and CAMS (restricted to 2014-2016) were only marginally different from SCA derived for the full time (2014-2019)”. Figure S30 does not really support this claim. The figure shows difference in SCA of ~0.5 ppm and up to 10 days in HDD for TCCON sites. These do not seem marginal.
P10L20: What does “systematic distribution” mean?
P10L31: What does it mean for a gradient in North America to be “consistent” with a gradient in Eurasia?
Sec. 4.1 & Sec 4.2: These are both results sections, and the methods for this analysis need to be fully described. I suggest moving Sec 4.1 and Fig. 8 to supplementary materials, as it is well known that terrestrial biosphere fluxes drive seasonal variations in XCO2.
Fig 4.: What exactly is being plotted? Is this an instantaneous field? And after a simulation of what length?
P13L32: I do not understand what is being correlated, and what is the correlation coefficient?
P14L12-18: This seems out of place; this should be in the methods section.
P14L20: Why are the CT2019 fluxes being referred to as “GEOS-Chem”. This analysis is only looking at CT2019 fluxes, GEOS-Chem is not used here.
P15L20-21: “The dominance of terrestrial biospheric exchange in the GEOS-Chem model is likely an intentional quality built into the model” – What does this mean?
P15L24: “15 or 30 lifetimes”?
- Many of the seasonal cycle fits shown in Fig. S4-S11 appear to be quite unphysical. Thus, it is unclear if the analysis is really capturing accurate SCA estimates. There should be uncertainty quantification in the SCA fits, perhaps using bootstrap resampling or another technique. Ideally, the analysis could also be performed fitting truncated Fourier series, to test the impact of the functional form on the results.
-
RC2: 'Comment on acp-2021-185', Anonymous Referee #2, 07 Jun 2021
Spatial distributions of XCO2 seasonal cycle amplitude and phase over northern high latitude regions by Jacobs et al., is an ambitious analysis of the CO2 dry air column-average mole fraction measured by the OCO-2 satellite. The authors analyze satellite data that has been bias-corrected based on previous work by Jacobs and complement this analysis with simulated XCO2 from GEOS-Chem and a surface contact-tracing approach. The analysis is interesting and wrangling the OCO-2 observations at high latitudes for meaningful science is an important advance. Before publication, however, there are a few major points for the authors to address.
My largest concern with the paper is that the authors are underutilized their contact tracing simulations. The contact tracing approach was not clearly explained, and results from these simulations were introduced only in the discussion section. This needs to be fleshed out more in the results section, with a mind to new or refined supporting figures. Fig. 10 could be modified to better guide the reader as to the similarities noted with Fig. 4. For example, could boxes be added in Fig. 10 around the zones whose amplitude is in the top Nth percentile in Fig. 4? Related to the contact tracer approach, it was unclear why the CarbonTracker posterior fluxes were run through GEOS-Chem rather than simply using the CarbonTracker posterior CO2 fields. Was this so the transport model was consistent for the CO2 simulation and the contact tracer simulation? If so, this should be explained in the text. However, the use of a single transport model is a major limitation to the authors’ conclusion that accumulation of transported CO2 controls the seasonal cycle amplitude spatial patterns. It may be onerous to repeat the surface contact tracer analysis in another transport model, but could the authors conbine information from the CarbonTracker (TM5) posterior and their own CarbonTracker/GEOS-Chem simulations to substantiate their conclusion?
A second major issue is that the key insights from the research could be better laid out. Is it that the column seems to show earlier drawdown than inversions constrained by surface observations? The different behavior in Siberia compared to other zones? What are the implications for a correlation between amplitude and HDD in terms of processes that affect the seasonality of net exchange? Perhaps reorganization of the discussion section, with the surface tracer analysis presented in the results, will make it easier for these points to come through in the discussion.
I struggled with Section 3.2, since the section begins with a listing of material in the supplement. The authors should organize this information to provide a clear summary in the first few sentences of the paragraph/section, and then later refer to the supplementary figures for more details. In particular, the “unrealistic trop in wintertime values” might warrant a figure in the main text since the realism of the seasonal cycle fits is crucial for interpretation of the rest of the paper.
The term HDD was used before the concept was introduced or defined (L29 p7)
The authors cite a meridional gradient in Fig. 12, which I didn’t see convincingly in any panel. This is brought up in the discussion (p14), and I don’t think removing this sentence would affect the authors’ underlying argument. If it is left in the document, more support is required.
Nicole Jacobs et al.
Data sets
Fairbanks, AK EM27/Sun observations of XCO2, XCH4, and XCO with GGG2014 N. Jacobs, W. R. Simpson, F. Hase, T. Blumenstock, Q. Tu, M. Frey, M. K. Dubey, and H. A. Parker https://doi.org/10.3334/ORNLDAAC/1831
Model code and software
GEOS-Chem 12.7.2 The international GEOS-Chem Community https://doi.org/10.5281/zenodo.3701669
Nicole Jacobs et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 366 | 133 | 8 | 507 | 38 | 2 | 3 |
- HTML: 366
- PDF: 133
- XML: 8
- Total: 507
- Supplement: 38
- BibTeX: 2
- EndNote: 3
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1