the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Implementation and initial calibration of carbon-13 soil organic matter decomposition in Yasso model
- 1Finnish Meteorological Institute, P.O. Box 503, FI-00101 Helsinki, Finland
- 2Finnish Museum of Natural History LUOMUS, P.O. Box 64, 00014, University of Helsinki, Helsinki, Finland
- 3Natural Resources Institute Finland, P.O. Box 18, FI-01301, Vantaa, Finland
- 4Department of Microbiology and Institute for atmospheric research INAR, Faculty of Agriculture and Forestry, P.O. Box 56, 00014 University of Helsinki, Helsinki, Finland
- 5Natural Resources Institute Finland, P.O. Box 2, 00791, Helsinki, Finland
- 1Finnish Meteorological Institute, P.O. Box 503, FI-00101 Helsinki, Finland
- 2Finnish Museum of Natural History LUOMUS, P.O. Box 64, 00014, University of Helsinki, Helsinki, Finland
- 3Natural Resources Institute Finland, P.O. Box 18, FI-01301, Vantaa, Finland
- 4Department of Microbiology and Institute for atmospheric research INAR, Faculty of Agriculture and Forestry, P.O. Box 56, 00014 University of Helsinki, Helsinki, Finland
- 5Natural Resources Institute Finland, P.O. Box 2, 00791, Helsinki, Finland
Abstract. Soil carbon sequestration has gained traction as a mean to mitigate rising atmospheric carbon dioxide concentrations. Verification of different methods’ efficiency to increase soil carbon sink requires, in addition to good quality measurements, reliable models capable of simulating the effect of the sequestration practices. One way to get insight of the methods’ effects on carbon cycling processes is to analyse different carbon isotope concentrations in soil organic matter. In this paper we introduce a carbon-13 isotope specific soil organic matter decomposition add-on into the Yasso soil carbon model and assess its functionality. The new 13C-dedicated decomposition is straightforward to implement and depends linearly on the default Yasso model parameters and the relative carbon isotope (13C/12C) concentration. Despite of their simplicity, the modifications considerably improve the model behaviour in a 50-year long simulation.
Jarmo Mäkelä et al.
Status: final response (author comments only)
-
RC1: 'Comment on bg-2021-327', Anonymous Referee #1, 16 Feb 2022
This manuscript describes new stable carbon isotope capabilities added to the Yasso model. The new model capabilities are described clearly. The model updates were parameterized and evaluated using measured datasets in a way that was well described and justified. Overall, I though the manuscript was a clear and concise description of a valuable new model capability. 13C measurements are a common metric for understanding soil organic matter decomposition processes and adding this capability to a SOM model is a valuable advance.
I did think that in some areas the introduction and conclusions went beyond the scope of the actual results. Specifically, the model developments and testing were entirely focused on 13C fractionation and did not include changes to or evaluation of overall soil C decomposition rates. Therefore, the hypothesis in the introduction about “significant improvements in SOM decomposition predictions” seems broader than is justified. The study does yield improvements in predictions of 13C dynamics, but this was not used to improve overall SOM predictions.
The first two paragraphs of the introduction (lines 10-20) provides a good justification for improving SOM models. However, the focus in these paragraphs on agricultural soils and carbon monitoring is not well related to the actual model structure and evaluation which only includes litter decomposition and peat systems. Carbon sequestration in mineral soils is sensitive to mineral-organic interactions and mineral-associated organic matter accounts for a large fraction of SOM (e.g., Lugato et al., 2021). However, Yasso does not include mineral interactions and treats humus as a passive pool and was only evaluated using litter and peat decomposition. Therefore, it does not seem justified to introduce the model in the context of agriculture soils. Since the model seems intended to simulate peat systems, I think it would be more reasonable to introduce it in the context of better understanding and predicting carbon dynamics in peatland or organic soils.
Reference: Lugato, E., Lavallee, J. M., Haddix, M. L., Panagos, P., & Cotrufo, M. F. (2021). Different climate sensitivity of particulate and mineral-associated soil organic matter. Nature Geoscience, 14(5), 295–300. https://doi.org/10.1038/s41561-021-00744-x
Other comments:
Section 2.1: The peat depth profile measurements that were used to validate the model should also be described in this section.
Figure 1: It would be helpful if the figure axes used the L notation that is used in the text so it is clearer what is being plotted. Is marginal likelihood in these plots the same as L?
Figure 2: Consider using different symbols for the branch and needle data to accommodate red-green colorblindness (which is common) or in the case of printing the paper in grayscale.
Line 138: It was not immediately clear to me how relative 13C content can change over time in the default model without any fractionation included. I think this occurs because the initial pools have different isotope ratios and are mixing over time which causes the isotope ratios to change. But a more specific explanation of this would be helpful. It might also be helpful to show a diagram (perhaps in the appendix) of transfers among the different pools so it is more clear what kind of mixing over time can occur.
Line 145: The actual depths should be included. And I suggest including a more detailed explanation of why the depth sampling was consistent with the 10 year age assumption. Was there evidence from that site that the age difference was actually close to 10 years across depths?
Figure 3: I suggest splitting this figure into separate panels as in Figure 2. The large number of lines and colors makes the figure difficult to interpret. Also, can bulk 13C in the model be calculated to compare with the bulk 13C measurement from peat?
Line 162: The negative parameter values are consistent with the theoretical expectation of slower 13C decomposition rate (as described in the introduction) which is a good result for the model and would be valuable to point out more explicitly.
Line 167: “This situation is not ideal” – why not? Is it inconsistent with measurements or theoretical expectations? It doesn’t seem particularly unreasonable to me.
Line 179-180: It’s not clear to me how the results demonstrate improvement to SOM model accuracy and predictability since they were not used to inform any changes to the overall C decomposition rate or structure. Improvements were limited to 13C dynamics.
Line 189: Similarly, it’s not clear that the study made improvements to SOM decomposition in general outside the direct comparisons to 13C content of organic matter pools.
- AC1: 'Reply on RC1', Jarmo Mäkelä, 30 Mar 2022
-
RC2: 'Comment on bg-2021-327', Anonymous Referee #2, 09 Mar 2022
The preprint manuscript “Implementation and initial calibration of carbon-13 soil organic matter decomposition in Yasso model” describes calibration of the Yasso model to 13C data collected from a litterbag decomposition experiment. The model was calibrated using 13C values measured on sequential extracts of pine litter and branch samples from a 4-year litterbag experiment. The decomposition parameter matrix of the Yasso model was modified to account for 13C using simple scalars. After optimization, three out of 4 scalars were negative, which was consistent with the hypothesis that 13C is preferentially retained in decomposing organic matter. The optimized model was applied to data from a peat core and produced more realistic predictions than the default model.
This manuscript is clear and concise. However, I think this manuscript should be framed differently to better showcase the results. The manuscript is framed narrowly in terms of soil carbon sequestration as a climate mitigation tool. However, the analyses and results are not directly relevant to soil carbon sequestration efforts. Specifically:
- The study system is unmanaged and focused on C cycling in litter and organic soils, and has no obvious connection to the agricultural soil carbon management strategies listed in the introduction.
- The 13C calibrated model performs no better at predicting changes in bulk C, hence its relevance to soil carbon measurement and verification efforts are unclear or at the very least indirect.
Later in the manuscript the significance of the 13C calibrated Yasso model is described differently, in terms of integration with 13C enabled ESMs. This seems like a much clearer justification for the calibration effort. Taken at face value, the results presented here are nearly trivial: calibrated the Yasso model to 13C data results in a better fit to 13C data. As a technical result, this is to be expected. What is the concrete significance of this incremental advance for our understanding of soil carbon cycling? What can the calibrated model eventually tell us about the cycling of the bulk C pool or the broader functioning of soil beyond fractionation of 13C?
If the 13C modifiers are generalizable to other systems (which may or may not be the case), I can see how they might enable the Yasso model so that it could be calibrated based on tracer experiments or in cases where the d13C of vegetation has shifted, or how it might be useful for interpreting time series of 13CO2 data to attribute fluxes to different soil C pools. These sorts of application are alluded to, but perhaps the manuscript would stand on its own more clearly if it was framed more clearly as an intermediate step towards these larger scientific goals.
Detailed comments:
Abstract: Details of the calibration dataset are not given in the abstract – consider including them.
Line 1; Line 10: I agree that strategies for increasing soil carbon as a climate mitigation strategy have received increasing attention over the years. However, I think this initial framing is an innapropriate place to start this manuscript (see broader comments above). Carbon cycling in soil is a fundamental aspect of terrestrial ecosystem function. Soil carbon influences the climate system and a whole range of global biogeochemical cycles regardless of how we try to manage it.
Line 7: I suggest deleting “despite of their simplicity”, as it implies that we expect that simple modifications will not generate improvements.
Lines 21-32: This paragraph begins by addressing the challenge of deciding which processes to include in models, but the application for 13C seems to mostly relate to parametrization. Is 13C useful for both determining model structure and fitting parameters? Are these distinct challenges?
Line 28: Writing edit -- delete “By” before “estimating”.
Lines 114-115: In other words, the precipitation and temperature dependence was the same for both isotopes? These factors are included in the original “alpha” term?
Line 126: how were the parameter “grid” and increment refocused? Was this done in a systematic way?
Figure 1: What do the color gradients represent? Likelihoods, presumably? In the panels situated along the diagonal, does the vertical axis on each panel show the likelihood? What do the vertical lines represent – parameter values at maximum likelihood? This caption needs to be expanded to clarify.
Figures 2-3: Why does d13C change over time in the default case? The default parameters are identical for 12C and 13C, correct? In this case, shouldn’t the 12C:13C ratio be preserved in all transformations, and the d13C value remain the same over time?
Methods section: Please include details about the computing methods. How were these procedures implemented? What computing environment was used (e.g., Python, R, Matlab)? Were any R packages used to assist with fitting?
Lines 131-132: I believe there are formal methods for evaluating collinearity between parameters. Computing a “collinearity index” might be useful for determining whether the parameters are identifiable (although such indices still reduce to qualitative rules of thumb). There are methods in R for this sort of analysis (package “FME” might be useful).
Lines 151 – 152: Here the emphasis is on incorporation into ESMs, not MRV for soil carbon sequestration.
Lines 145-146: So depth and time have been exchanged? Is this based on an assumption that the peat is accreting linearly? How was the conversion between depth and time parametrized? Why 10 year intervals, why not 20 or 50 years? More justification/expanation is needed here.
Lines 167-169: I do not follow this reasoning. Is the non-ideal finding that the parameter for the N pool is positive? How does the lack of depth resolution explain this?
Lines 179 – 180: The results presented here indicate that calibration of 13C parameters to 13C data improves accuracy and predictive power for 13C. However, they do not show how this improves the skill of the model with respect to bulk C pools or fluxes. What can these results tell us beyond 13C fractionation?
- AC2: 'Reply on RC2', Jarmo Mäkelä, 30 Mar 2022
-
EC1: 'Comment on bg-2021-327', Ben Bond-Lamberty, 09 Apr 2022
Dear authors: in my decision I used the stock language "please consider carefully and respond to all their thoughtful suggestions" but should have said "I have revised your responses and am convinced there's a solid path forward to address the referees' comments". Sorry for the mistake.
- EC2: 'Reply on EC1', Ben Bond-Lamberty, 09 Apr 2022
Jarmo Mäkelä et al.
Jarmo Mäkelä et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 395 | 92 | 21 | 508 | 7 | 10 |
- HTML: 395
- PDF: 92
- XML: 21
- Total: 508
- BibTeX: 7
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1