Historical and future contributions of inland  waters to the Congo Basin carbon balance

Hastie, Adam; Lauerwald, Ronny; Ciais, Philippe; Papa, Fabrice; Regnier, Pierre

doi:https://doi.org/10.5194/esd-12-37-2021

Articles | Volume 12, issue 1

https://doi.org/10.5194/esd-12-37-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/esd-12-37-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 12, issue 1

Research article

|

07 Jan 2021

Research article |

| 07 Jan 2021

Historical and future contributions of inland waters to the Congo Basin carbon balance

Adam Hastie, Ronny Lauerwald, Philippe Ciais, Fabrice Papa, and Pierre Regnier

Download

Final revised paper (published on 07 Jan 2021)
Preprint (discussion started on 11 Mar 2020)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'review', Anonymous Referee #1, 25 Mar 2020
- AC1: 'Response to Reviewer 1', Adam Hastie, 25 May 2020
RC2: 'Hastie et al. Historical and future contributions of inland waters to the Congo basin carbon balance', Anonymous Referee #2, 18 Apr 2020
- AC2: 'Response to Reviewer 2', Adam Hastie, 25 May 2020

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

ED: Reconsider after major revisions (10 Jun 2020) by Rui A. P. Perdigão

AR by Anna Wenzel on behalf of the Authors (20 Aug 2020) Author's response

ED: Referee Nomination & Report Request started (04 Sep 2020) by Rui A. P. Perdigão

RR by Anonymous Referee #1 (07 Sep 2020)

Suggestions for revision or reasons for rejection

GENERAL COMMENTS

The authors have addressed most of my previous comments. In the discussion of the differences between model results and data-driven estimates there could be a clarification in the comparison of numbers (see below). Also I do not agree with the way the authors dismiss differences in k to explain the divergence of results. Changes in k will lead to changes in emissions if the water is substantially over-saturated in CO2 (the change of k will have a small effect of the flux, in the water is very close to saturation, which is not the case of the Congo). I suggest that the authors provide a sensitivity analysis and increase the k by 50% and see how much the flux changes in the Congo, rather than refer to a study in the Amazon.

MAJOR COMMENTS

L78 : Battin et al (2009) report on aquatic heterotrophy (respiration) not C transfer to LOAC

L 391 : The comparison of pCO2 in the main-stem Congo at Kisangani and Kinshasa does allow to conclude that the model reproduces the “broad spatial pattern of pCO2 measured in Borges et al. (2019)” as stated. There are other important features of “spatial pattern” of pCO2 observations such as much higher pCO2 values in the small and major tributaries in some cases with pCO2 values up to 18,000 ppm. I suggest to rephrase to “broad spatial pattern of pCO2 measured in the main-stem Congo reported by Borges et al. (2019)”

L409 : Borges et al. (2019) (Figure 20) also reports a 2 yr time series (2017-2018) of pCO2 at Kisangani. Wang et al. (2013) report a 1 yr time series of pCO2 at Kinshasa. It would make a convincing case if the time-series comparison was extended to these 2 other sites were complete annual cycles are available rather than only to Bangui on the basis that is the “most complete” as stated (whatever “complete” means in this context).

L 584 : The value of k of 3.5 m/d is ok for high order streams (for instance minstem) and k used by Borges et al (2019) strongly increases in low order streams up to 39 m/d in order 1 streams. The largest fraction of CO2 emission reported by Borges et al. (2019) from the river network is related to low order streams.

L 587 : The statement that CO2 flux in orchileak is not sensitive to the k value is really strange and might require a few words of explanation. For a water pCO2 of 2800 ppm (seems to be sort of simulated pCO2 by the model in the Congo river) and a k of 3.5 m/d, the emission is more or less 287 mmol/m2/d. If the k is increased by 50% (=5.3 m/d) the resulting flux is more or less 430 mmol/m2/d, so also 50% higher.

So whether the fluxes are derived from ORCHILEAK or “data-driven models”, a substantial increase of k leads to a substantial increase of k. I suggest that the authors provide a sensitivity analysis and increase the k by 50% in Orchileak and see how much the flux changes in the Congo, rather than refer to a study in the Amazon.

Based on the reported values of flux, k and surface area, I compute that the spatially integrated average of river pCO2 from ORCHILEAK is approximately 2800 ppm. The spatially integrated average of river pCO2 from Borges et al (2019) is approximately 5560 ppm based on numbers in Table 1. The average k Borges et al (2019) is 8 m/d, also based on data from Table 1,which is higher than the k value in ORCHILEAK of 3.5 m/d. So the difference between both estimates is related to both lower pCO2 and lower k in Orchileak. The authors should clarify how K is computed in Orchileak and discuss in light of the k computation scheme of Borges et al. (2019) that is transparently explained in the methods and supplements. This might shed some light on why the k values are so different in both studies. The higher pCO2 values in Borges et al. (2019) might result from not representing macrophytes and/or because the model under-estimates pCO2 in low order streams. This is not possible to check since the model results are presented as overall means and not information is given as function of stream size. I suggest that the authors present in more detail the pCO2 values simulated to tributaries and in particular low order streams.

L 578 comparison with Lauerwald et al. (2015) also suggests that you « substantially underestimate total riverine CO2 evasion”.

Regarding historical changes in LOAC fluxes, the authors might consider including in the discussion the recent paper of Moukandi N’kaya et al. (2020) that attribute decadal changes in DOC export from the Congo to changes in hydrology and inundation patterns.

REFS

Moukandi N’kaya et al. (2020) Temporal Variability of Sediments, Dissolved Solids and Dissolved Organic Matter Fluxes in the Congo River at Brazzaville/Kinshasa, Geosciences 2020, 10, 341; doi:10.3390/geosciences10090341

Wang, Z. A., D. J. Bienvenu, P. J. Mann, K. A. Hoering, J. R. Poulsen, R. G. M. Spencer, and R. M. Holmes (2013), Inorganic carbon speciation and fluxes in the Congo River, Geophys. Res. Lett., 40, doi:10.1002/grl.50160

Hide

RR by John Melack (26 Sep 2020)

Suggestions for revision or reasons for rejection

Hastie and co-authors have responded well to comments on their initial submission. Though their projections of changes through this century are more uncertain and speculative than as suggested in the abstract, the main text does acknowledge the considerable problems with the projections. The following specific comments are minor or can be readily clarified.
Introduction
L49: Change ‘sparcity’ to ‘paucity’.
Methods
L140-142: ‘Fixed gas exchange velocities of 3.5 m d-1 and 0.65 m d-1 respectively. are used for rivers (including open floodplains) and forested floodplains.’
A couple of references for these gas exchange velocities could be added. In particular, MacIntyre et al (2019) provides results for forested floodplains.
MacIntyre et al. 2019. Turbulence and gas transfer velocities in sheltered flooded forests of the Amazon basin. Geophysical Research Letters. doi.org/10.1029/2019GL083948
Figure 3a: Are fluxes from L. Tanganyika and other lakes included? If so, what is the source of the gas concentrations and fluxes.
L241-242: ‘The best performing climate forcing dataset was ISIMIP2b followed by Princeton GPCC with root mean square errors (RMSE) of 29% and 40% and Nash Sutcliffe efficiencies (NSE) of 0.20 and -0.25, respectively.’
Are RMSE and NSE values considered fair or good?
248-250: ‘water residence times - 0.5 (days) for floodplain reservoirs’
What is the basis for residence times of 0.5 days on the floodplains, as it seems too fast?
L255-257:’ As in previous studies on the Amazon basin (Lauerwald et al. 2017, Hastie et al., 2019) we defined bank-full discharge, i.e. the threshold discharge at which floodplain inundation starts (i.e. overtopping of banks), as the median discharge (50th percentile) of the present-day climate forcing period (1990 to 2005).’

The response to the comment on the initial submission (‘The concept of bank-full discharge as a threshold for initiation of inundation of floodplains is questionable as applied to tropical floodplain such as those in the Amazon or Congo. Studies inundation dynamics in the Amazon with detailed measurements or modeling indicate that inundation occurs more or less continuously as the rivers rise and that the water comes from both the rivers and uplands.’) was mis-interpreted. The issue being raised was with regard to the proportion of water from different sources. The issue concerns the observation that natural floodplains, such as those in the Amazon and Congo, have channels that connect the rivers to the floodplains such that waters rise on the floodplains in concert with the river rise, not just after bank-full discharge is reached.

Table 1: Is there snowfall in basin?
How does the river area used compare to the recent estimate by Allen and Pavelsky (2018. Global extend of rivers and streams. Science 361: 585–588) ?
Results
Table 2: State ‘observed flooded area’ is from GIEMS. As noted correctly in the text (L364-379), the GIEMS data under-estimate the flooded area.
L448-449: ‘We simulate a mean annual flux of DOC throughfall from the canopy of 27 ±1 Tg C yr-1. How does this compare to measured fluxes (e.g. Filoso et al. 1999. Composition and deposition of throughfall in a flooded forest archipeligo (Anavilhanas, Negro River, Brazil). Biogeochemistry 45:169-195).

Hide

ED: Publish subject to minor revisions (review by editor) (12 Oct 2020) by Rui A. P. Perdigão

AR by Adam Hastie on behalf of the Authors (29 Oct 2020) Author's response Manuscript

ED: Publish as is (16 Nov 2020) by Rui A. P. Perdigão

Short summary

We used a model of the Congo Basin to investigate the transfer of carbon (C) from land (vegetation and soils) to inland waters. We estimate that leaching of C to inland waters, emissions of CO₂ from the water surface, and the export of C to the coast have all increased over the last century, driven by increasing atmospheric CO₂ levels and climate change. We predict that these trends may continue through the 21st century and call for long-term monitoring of these fluxes.