the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 30 Sep 2021
Additional soil organic carbon storage potential in global croplands
Abstract. Soil organic carbon sequestration (SOCseq) is considered the most attractive carbon capture technology to partially mitigate climate change. However, there is conflicting evidence regarding the potential of SOCseq. The additional storage potential on existing global cropland is missing. SOCseq is region-specific and conditioned by management but most global estimates use fixed accumulation rates or time frames. Here, we show how the SOC storage potential and its steady state varies globally depending on climate, land use and soil. Using 83,416 soil observations, we developed a quantile regression neural network that quantifies the SOC variation within soils with similar characteristics. This allows us to identify similar areas that present higher SOC with the difference representing an additional storage potential. The estimated additional SOC storage potential of 29 to 67 Pg C in the topsoil of global croplands equates to only 2 to 5 years of emissions offsetting and 32 % of agriculture's 92 Pg historical carbon debt estimate due to conversion from natural ecosystems. Since SOC is temperature-dependent, this potential is likely to reduce by 18 % by 2040 due to climate change.
- Preprint
(4908 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on soil-2021-73', Philippe C. Baveye, 02 Nov 2021
-
AC1: 'Reply on RC1', José Padarian, 11 Nov 2021
The comment was uploaded in the form of a supplement: https://soil.copernicus.org/preprints/soil-2021-73/soil-2021-73-AC1-supplement.pdf
-
AC1: 'Reply on RC1', José Padarian, 11 Nov 2021
-
RC2: 'Comment on soil-2021-73', Anonymous Referee #2, 05 Nov 2021
Using a quantile CNN model, the authors assessed the potential of soil carbon accrual in global soils.
Frankly, I have carefully read the comments by the other reviewer. I have to say I stand with her/him. Here, I only focused on some technical aspects which should be addressed before publication of the manuscript.
The quantile approach relies on a strong assumption that SOC will be same under similar soil forming factors. If SOC values are different, it much be induced by management. But the problem is that we would never find two soils with the same forming factors. Numerous factors (e.g., climate seasonality) regulate SOC dynamics and thus SOC stock at a typical site. At the same site, SOC would also experience temporal changes. In this study, only very few potential predictor variables of SOC were considered. Other variables such as soil parent material, land use history, climate variability are not included. More importantly, the approach adopts another strong assumption of steady state. If the soil is not at the steady state, the approach will be invalid, because a mediate soil (50% quantile) may experience SOC loss. The SOC would be an overestimation of the real 50% quantile, and vise versa.
The manuscript paid little attention to the potential uncertainties in the relevant estimations. Two major uncertainties I think should be explicitly tested are: the approach used to estimate BD and prediction uncertainty by the quantile CNN model.
I found that the method section is very fragmentary. Here, I just gave some examples. To my knowledge, BD has been reported for some soil profiles, why was BD estimated using a pedotransfer function? Could you please test the credibility of the BD estimates which are vital for estimating SOC stock? In page 3, a bootstrapping routine was mentioned. The reader cannot find anything descriptions on the purpose of this routine. To predict SOC stock? The author very briefly desribed future climate projections. As the SOC estimates were conducted at the global scale, I believe historical climate records are required to run the GCMs. How were the climate projections used in their models for predicting SOC stocks?
Citation: https://doi.org/10.5194/soil-2021-73-RC2 -
AC2: 'Reply on RC2', José Padarian, 11 Nov 2021
The comment was uploaded in the form of a supplement: https://soil.copernicus.org/preprints/soil-2021-73/soil-2021-73-AC2-supplement.pdf
-
AC2: 'Reply on RC2', José Padarian, 11 Nov 2021
-
RC3: 'Comment on soil-2021-73', Anonymous Referee #3, 08 Nov 2021
soil-2021-73
The manuscript addresses the potential of organic carbon storage in soils with respect to climate change mitigation. In particular, the effects of management practices are examined in a global study.I have fundamental problems with the approach taken and the lack of any kind of validation.
Carbon sequestration in agroecosystems appears to be a significant way to offset some anthropogenic CO2 emissions, and no-till is generally considered an efficient and essential component for sequestering SOC. However, data comparing no-till and full tillage show large uncertainties, and not all studies found that SOC levels increased following a change in management to no-till. While there may be a significant change in C distribution in the soil profile, this does not necessarily translate into an increase in total SOC.
Since the most important management factor appears to have a limited impact, the hypothesis of this study is generally in question. So the main question is what management practices are we talking about that would result in significant SOC storage. I am assuming that what we are seeing are the effects of potential natural land cover, not the effects of human land use.
In this regard, land use history is also a very important factor. This is probably the most difficult part of the equation. It is likely to have a greater influence compared to changes in analytical methods over time. A common problem with global studies and modeling is spatial resolution. Land use and its history often vary on very fine scales, which cannot be accounted for with low resolution spatial data.
One factor controlling SOC distribution is soil erosion. Countermeasures may well cause SOC to accumulate in the soil. Colluvial soil can also store a lot of SOC. Estimates put the resulting global storage at 78 Pg C. Such effects are not considered in this study because neither terrain characteristics, soil properties, nor parent material are accounted for in the models.
That said, the results of the SHAP analysis become clearer at the 75th and 90th percentiles. This may indeed indicate some effect of management practices, but also the general potential to develop higher SOC levels in some terrain positions, as evidenced by the increase in importance of low elevations. Again, this may be an effect of small-scale (lower elevation) terrain and soil variability - rather than management practices.
All analyses and results are relatively worthless if they are not validated. And here, no validation of the hypothesis and no validation statistics for the modeling are presented. Therefore, the result is relatively meaningless.
The consideration of climate change impacts is generally a good point and the results (based on the hypothesis) are interesting. However, the steady state assumptions used as the basis for the space-time substitution are problematic, especially for cropland. The authors should consider discussing this in a separate publication when the hypothesis is better justified, explained, and validated.
Since the hypothesis is overly simplistic and the results obtained are most likely as uncertain as previous approaches, I am not sure it adds much to the discussion, which is thus rather counterproductive given the urgency of the problem.
References:
https://doi.org/10.1038/s41598-019-47861-7
https://doi.org/10.1016/j.still.2008.05.010
https://doi.org/10.1038/nclimate3263
https://doi.org/10.1016/j.agee.2010.08.006
Specific questions and comments:
The title and the abstract are not very clear.
Is it a "Quantile CNN model" as in the heading of section 2.2 or a "Quantile DL model"?
If I understand correctly, the global predictions are made based on cropland/pasture data only. The calculated SOC totals seem to be based on the global models. Here, however, at least the current forest areas would have to be removed, because otherwise the global storage capacity would be overestimated.Citation: https://doi.org/10.5194/soil-2021-73-RC3 -
AC3: 'Reply on RC3', José Padarian, 11 Nov 2021
The comment was uploaded in the form of a supplement: https://soil.copernicus.org/preprints/soil-2021-73/soil-2021-73-AC3-supplement.pdf
-
AC3: 'Reply on RC3', José Padarian, 11 Nov 2021
Status: closed
-
RC1: 'Comment on soil-2021-73', Philippe C. Baveye, 02 Nov 2021
-
AC1: 'Reply on RC1', José Padarian, 11 Nov 2021
The comment was uploaded in the form of a supplement: https://soil.copernicus.org/preprints/soil-2021-73/soil-2021-73-AC1-supplement.pdf
-
AC1: 'Reply on RC1', José Padarian, 11 Nov 2021
-
RC2: 'Comment on soil-2021-73', Anonymous Referee #2, 05 Nov 2021
Using a quantile CNN model, the authors assessed the potential of soil carbon accrual in global soils.
Frankly, I have carefully read the comments by the other reviewer. I have to say I stand with her/him. Here, I only focused on some technical aspects which should be addressed before publication of the manuscript.
The quantile approach relies on a strong assumption that SOC will be same under similar soil forming factors. If SOC values are different, it much be induced by management. But the problem is that we would never find two soils with the same forming factors. Numerous factors (e.g., climate seasonality) regulate SOC dynamics and thus SOC stock at a typical site. At the same site, SOC would also experience temporal changes. In this study, only very few potential predictor variables of SOC were considered. Other variables such as soil parent material, land use history, climate variability are not included. More importantly, the approach adopts another strong assumption of steady state. If the soil is not at the steady state, the approach will be invalid, because a mediate soil (50% quantile) may experience SOC loss. The SOC would be an overestimation of the real 50% quantile, and vise versa.
The manuscript paid little attention to the potential uncertainties in the relevant estimations. Two major uncertainties I think should be explicitly tested are: the approach used to estimate BD and prediction uncertainty by the quantile CNN model.
I found that the method section is very fragmentary. Here, I just gave some examples. To my knowledge, BD has been reported for some soil profiles, why was BD estimated using a pedotransfer function? Could you please test the credibility of the BD estimates which are vital for estimating SOC stock? In page 3, a bootstrapping routine was mentioned. The reader cannot find anything descriptions on the purpose of this routine. To predict SOC stock? The author very briefly desribed future climate projections. As the SOC estimates were conducted at the global scale, I believe historical climate records are required to run the GCMs. How were the climate projections used in their models for predicting SOC stocks?
Citation: https://doi.org/10.5194/soil-2021-73-RC2 -
AC2: 'Reply on RC2', José Padarian, 11 Nov 2021
The comment was uploaded in the form of a supplement: https://soil.copernicus.org/preprints/soil-2021-73/soil-2021-73-AC2-supplement.pdf
-
AC2: 'Reply on RC2', José Padarian, 11 Nov 2021
-
RC3: 'Comment on soil-2021-73', Anonymous Referee #3, 08 Nov 2021
soil-2021-73
The manuscript addresses the potential of organic carbon storage in soils with respect to climate change mitigation. In particular, the effects of management practices are examined in a global study.I have fundamental problems with the approach taken and the lack of any kind of validation.
Carbon sequestration in agroecosystems appears to be a significant way to offset some anthropogenic CO2 emissions, and no-till is generally considered an efficient and essential component for sequestering SOC. However, data comparing no-till and full tillage show large uncertainties, and not all studies found that SOC levels increased following a change in management to no-till. While there may be a significant change in C distribution in the soil profile, this does not necessarily translate into an increase in total SOC.
Since the most important management factor appears to have a limited impact, the hypothesis of this study is generally in question. So the main question is what management practices are we talking about that would result in significant SOC storage. I am assuming that what we are seeing are the effects of potential natural land cover, not the effects of human land use.
In this regard, land use history is also a very important factor. This is probably the most difficult part of the equation. It is likely to have a greater influence compared to changes in analytical methods over time. A common problem with global studies and modeling is spatial resolution. Land use and its history often vary on very fine scales, which cannot be accounted for with low resolution spatial data.
One factor controlling SOC distribution is soil erosion. Countermeasures may well cause SOC to accumulate in the soil. Colluvial soil can also store a lot of SOC. Estimates put the resulting global storage at 78 Pg C. Such effects are not considered in this study because neither terrain characteristics, soil properties, nor parent material are accounted for in the models.
That said, the results of the SHAP analysis become clearer at the 75th and 90th percentiles. This may indeed indicate some effect of management practices, but also the general potential to develop higher SOC levels in some terrain positions, as evidenced by the increase in importance of low elevations. Again, this may be an effect of small-scale (lower elevation) terrain and soil variability - rather than management practices.
All analyses and results are relatively worthless if they are not validated. And here, no validation of the hypothesis and no validation statistics for the modeling are presented. Therefore, the result is relatively meaningless.
The consideration of climate change impacts is generally a good point and the results (based on the hypothesis) are interesting. However, the steady state assumptions used as the basis for the space-time substitution are problematic, especially for cropland. The authors should consider discussing this in a separate publication when the hypothesis is better justified, explained, and validated.
Since the hypothesis is overly simplistic and the results obtained are most likely as uncertain as previous approaches, I am not sure it adds much to the discussion, which is thus rather counterproductive given the urgency of the problem.
References:
https://doi.org/10.1038/s41598-019-47861-7
https://doi.org/10.1016/j.still.2008.05.010
https://doi.org/10.1038/nclimate3263
https://doi.org/10.1016/j.agee.2010.08.006
Specific questions and comments:
The title and the abstract are not very clear.
Is it a "Quantile CNN model" as in the heading of section 2.2 or a "Quantile DL model"?
If I understand correctly, the global predictions are made based on cropland/pasture data only. The calculated SOC totals seem to be based on the global models. Here, however, at least the current forest areas would have to be removed, because otherwise the global storage capacity would be overestimated.Citation: https://doi.org/10.5194/soil-2021-73-RC3 -
AC3: 'Reply on RC3', José Padarian, 11 Nov 2021
The comment was uploaded in the form of a supplement: https://soil.copernicus.org/preprints/soil-2021-73/soil-2021-73-AC3-supplement.pdf
-
AC3: 'Reply on RC3', José Padarian, 11 Nov 2021
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,839 | 593 | 58 | 2,490 | 39 | 32 |
- HTML: 1,839
- PDF: 593
- XML: 58
- Total: 2,490
- BibTeX: 39
- EndNote: 32
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Cited
2 citations as recorded by crossref.
- Crop intensification effects on soil quality and organic carbon stocks: a case study of Haramosh Valley in Central Karakorum, Pakistan M. Alam et al. 10.1080/13504509.2022.2116613
- Estimating the attainable soil organic carbon deficit in the soil fine fraction to inform feasible storage targets and de-risk carbon farming decisions S. Karunaratne et al. 10.1071/SR23096