the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 23 Mar 2026
Impacts of land-use and land-cover change on blue–green water partitioning
Abstract. Land-use and land-cover change (LULCC) is a major driver of terrestrial water cycle changes, yet its effects on how precipitation is partitioned into blue (runoff) and green water (transpiration) flows remain unclear. Here we address this knowledge gap using Earth system model simulations from the Land Use Model Intercomparison Project (LUMIP) under contrasting socioeconomic pathways (SSP1-2.6 and SSP3-7.0). We find that future sustainable LULCC (i.e., predominantly avoided deforestation and preservation of natural non-forest ecosystems) significantly impacts blue-green water partitioning, with regions showing positive leaf area index (LAI) and gross primary productivity (GPP) responses generally corresponding to larger green water shares. These effects are strongest in the tropics and particularly during dry seasons, where LAI and GPP responses are largest. Regions with the strongest green water gains show the highest sensitivity of blue-green water partitioning to vegetation responses, with the largest partitioning shifts per unit change in LAI or GPP. Precipitation responses to LULCC further modulate the strength of blue–green water partitioning shifts. In some regions, higher transpiration is partly offset by increased rainfall, limiting reductions in blue water availability. While we find consistent ecohydrological responses to LULCC across the multi-model ensemble mean, substantial regional inter-model disagreement arises due to differences in model-specific plant functional types and their parametrisations. Our results underscore that the water cost or benefit of land management depends jointly on vegetation function, precipitation feedbacks, and model structural uncertainty.
- Preprint
(5300 KB) - Metadata XML
-
Supplement
(20265 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Reviewer comments on egusphere-2026-513', Anonymous Referee #1, 15 Apr 2026
-
RC2: 'Comment on egusphere-2026-513', Josephin Kroll & Rene Orth (co-review team), 18 May 2026
Review of Heselschwerdt et al., egusphere-2026-513
“Impact of land use and land cover change on blue-green water partitioning“
This study examines the role of land use and land cover change on the partitioning of precipitation to blue (runoff) and green (transpiration) water fluxes. The authors determine how a switch between two different land use and land cover change (LULCC) scenarios (SSP1-2.6 and SSP3-7.0) alters blue-green water partitioning across four Earth system models from the Land Use Model Intercomparison project. The authors relate changes in blue-green water partitioning to vegetation functioning, considering varying sensitivity among the data distribution. They strongly emphasize and discuss inter-model spread throughout the manuscript.
______________________
Recommendation:
The paper needs major revisions.
The paper is an important contribution to advance the understanding of how water fluxes will change in the future in response to land use and land cover change. The authors substantially widen the area across which the role of LULCC for water partitioning has been examined by using a so far only regionally applied methodology globally. The authors utilize data from CMIP6/LUMIP for their analysis and find substantial inter-model spread regarding the role of LULCC for future water cycle projections. This makes it hard to draw robust conclusions regarding the role of LULCC for blue-green water partitioning. However, it also enables the authors to provide an extensive discussion of possible reasons for model uncertainty and disagreement thereby granting insights for model development.Before publication of the manuscript, the following points need to be considered:
______________________
Major comments:
- We feel that the main focus of the paper on future changes in water and carbon cycling in response to land use and land cover changes (Figures 1-3) requires revision given the substantial differences between individual models in terms of the translation of land use/land cover changes into changes in LAI, GPP and BGWS. That said, a possible way forward could be to identify regions where the diagnosed signals in the four chosen models are relatively robust and representative, e.g. by determining regions where the four-model-ensemble signal agrees with that of the nine-model-ensemble in the historical period in terms of the sign (Figure S4). In other regions, however, results should not be shown (e.g. masking with grey color), and the discussion should focus rather on uncertainties (which it already does in many parts of the manuscript) than on diagnosed signals that can inform climate change adaptation and mitigation.
- Related to the above comment, (i) the title currently does not reflect the main finding of the study where model uncertainty does not actually allow for robust conclusions of the impacts of land use and land cover change, (ii) the first two research questions stated at the end of the introduction (lines 79-80) can not be answered given the uncertainty across models; and in fact these research questions are not revisited later in the manuscript because the authors acknowledge the uncertainty across models, thereby comprehensively addressing their third research question.
- The authors discuss in the text related to Figures 1-3 that/which land use/land cover changes translate to changes in LAI, GPP and BGWS. At the same time, the figures only show resulting changes in LAI, GPP and BGWS while omitting the information on underlying land use/land cover changes. More generally, the motivation for studying LAI and GPP changes in Figures 1 and 2 is not clear. It would be informative to also show land use/land cover changes in the main figures in order to relate the resulting LAI, GPP and BGWS changes to this cause. We acknowledge that some land use/land cover change information is provided in the supplementary material, but this should be related to the content of main figures 1-3 and moved to the main material.
- Some methodological choices made by the authors required more justification. For example, the interpretation of the BGWS results is complicated by the fact that transpiration, which is chosen as a green water flux here, may be small compared to evaporation in some (arid) regions (line 167). Also, the thresholds applied for the amount of the investigated water fluxes are very small (line 177) such that weak signals may be overinterpreted such as changes in water flux partitioning in arid regions with little precipitation.
- The manuscript provides comprehensive discussion around possible reasons for model disagreement. This is a strength of the current manuscript. However, the multifaceted reasons for model disagreement make the related discussion complex and hard to follow. Therefore, we would suggest introducing a schematic figure illustrating the sources of model uncertainties and disagreement. Then, the text related to Figure 4 and the first part of the discussion section could be linked with this figure.
- There is inconsistent terminology across the manuscript. For example, (i) change and anomaly are used interchangeably (e.g. in section 3.2), (ii) the difference between ΔLULCC and ΔsustainLULCC is not clear, (iii) IBGWS is not referred to in the results section, and (iv) vegetation water use is not defined while probably used as a description of transpiration.
We do not wish to remain anonymous - joint review by Rene Orth and Josephin Kroll.
We have notified the editor of our collaborations with Peter Greve.
__________________________
Minor comments:
- While the introduction section provides a useful background on the state-of-the-art of impacts of land use/land cover changes on water cycling, it does not describe the existing knowledge gap(s) which are addressed in this study. This makes it difficult to link this section with the actual analysis described in the further sections.
- Research question 2 develops around the role of vegetation water use (= transpiration?) or precipitation feedbacks for changes in BGWS. Currently, the findings for this are hidden in the supplement and not combined. This could be nicely illustrated by a main figure combining categories from the change in ET or P with the change in BGWS.
- We would suggest to combine the discussion with the results section rather than with the conclusion.
- Line 8: the term “green water gains” is not defined
- Lines 26 & 56: the paper of Hoek van Dijke et al. 2022 which is mentioned in other places in the manuscript, could be referenced here
- Usage of T instead of ET, reasoning given in Line 35ff. not sufficient
- Line 36: Orth & Destouni 2018 do not perform a model-based analysis but focus on observation-based datasets
- Lines 36-38: Either explain the BGWS metric more, or omit this here and introduce it only in the methods section.
- Line 39: signal at the coast of Africa varies quite a lot between models
- Line 58: depress → reduce
- Line 100: It would be good to check the contribution of both terms to make sure that not one is dominant about the other which would indicate limited representativeness of the findings with respect to background climate
- Line 113 Specify ‘For context’ ; maybe add on how land-use in this time period looked - e.g. in Europe land use /deforestation had its highest rates during 19th century
- Line 124: content from S1 might be more relevant for following the discussion than the technical details provided in T1
- Line 127: Explain the usage of surface soil moisture, instead of e.g. total column soil moisture. How representative is a 30-year mean considering the high variability in the upper 10cm of the soil?
- Line 132: Comparing the mean of several UK model runs with individual models is likely unfair as e.g. aspects as internal climate variability are merged out. To enhance comparability, we suggest touse an individual ensemble member, and compare the results with that of the other ensemble members to estimate the relevance of them.
- Line 144,317, 461: Explain more what you mean with “realized land cover response” and how the same prescribed land use (change) would lead to e.g. different tree cover fractions in different models
- Line 145: Why are some LULC types given as fractions and some as tile fractions? Is there a difference between both?
- In sec 2.3 maybe give an example between which values BGWS globally varies so that readers can later better understand ΔBGWS[%]
- Sec 2.4: Maybe use a heading that tells about the content rather than the method you are explaining
- Line 210: Do you observe less forest and more cropland somewhere else than in China? If yes, does the explanation based on bioenergy crops hold there as well?
- Line 219-221: Is this already interpretation/discussion? If so, maybe add a reference here
- Line 222: Check whether it is a hyphen or a minus between GPP and LAI
- Fig 3: caption: missing ‘in’ before %-sign; ‘In the MMM, regions with low ensemble…’ add ‘with dots’ in the end
- Lines 230-234: This is informative content which, however, should be discussed later in this subsection.
- Line 242: does ‘not consistently’ refer to across space or across models?
- Line 250-256: Link supplementary figures of changes in P, ET, R
- Check supplementary figures to remove data over the ocean
- Line 263-275: Maybe add boxes around the example regions you are referring to in the text to make it easier for readers to follow across the different figures
- Line 275: through a shrinking precipitation denominator - does this refer to eq. 3?
- Line 280: ‘suggesting a tighter local ET-precipitation co-variability’ specify in which of the previously mentioned regions
- Line 285: ‘Similarly’ - to what?
- Line 400: unclear what ‘direction of vegetation change (loss versus gain)’; forest and crops are both vegetation
- Lines 455 & 458: The difference between the first and second source of uncertainty is unclear.
Citation: https://doi.org/10.5194/egusphere-2026-513-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 486 | 180 | 44 | 710 | 522 | 41 | 56 |
- HTML: 486
- PDF: 180
- XML: 44
- Total: 710
- Supplement: 522
- BibTeX: 41
- EndNote: 56
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments
The paper provides a rich global-scale analysis of effects of land use and land cover change (LULCC) upon the partitioning of blue and green water flows, using a LUMIP model ensemble in different setups (focused on a ‘sustainable LULCC’ scenario). The analysis appears to be well done and plausibly interpreted, with a quite detailed examination of process interactions, regional features, and model uncertainties. That said, the Results and Discussion are rather dense with much information, parts of which could be portrayed in a somewhat more accessible way. Below I provide some comments on how to possibly achieve this. Overall, may comments are relatively minor and could be addressed mainly by some clarifications and restructurings.
Specific comments
The Abstract should clarify that you mainly analyze transpiration (as a representative of green water) rather than total evapotranspiration. While outputs for evaporation appear to have been analyzed also, I do miss a short discussion of whether a full inclusion of evapotranspiration (not just transpiration) in the ratio would produce relevantly different results.
Please state whether (and how) the direct effects of increasing carbon dioxide concentrations upon vegetation productivity / coverage and upon transpiration is considered in the models. No need to separately quantify the particular contributions of these effects here, but at least some words on this would be helpful (in the Discussion); i.e. is transpiration suppressed a lot under SSP3-7 / high CO2 concentrations. Relatedly, the UKESM model simulates “dynamic PFT competition” while in other models vegetation distribution is prescribed (according to Table 1). Does this have implications for results from that particular model?
Sections 3.1 and 3.2 contain a lot of information with many acronyms etc. Is it possible to provide a table listing the model experiments (that could also be used to remind the reader of the different setups analyzed) and key results for each, such as the correlations, global areas affected by change XY, or something similar? Then the table could be pointed to for these results rather than mentioning them all in the text.
Besides, if there is need for shortening the length of the text, section 3.3 is a candidate; it is a regional zoom of the analysis providing much detail, maybe not all of it required to get the main messages across.
At the beginning of the Discussion, the key results could be highlighted again as bullet points, and then – which I would definitely recommend – the Discussion should have some subtitles for each of these points (like green water shares, teleconnections, uncertainties, limitations). This would be another, easy-to-implement way to highlight once more the main results and to guide the reader through the large amount of information.
Technical corrections – none.