the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Aug 2024
Future land-use pattern projections and their differences within the ISIMIP3b framework
Abstract. Land use is a key human driver affecting Earth’s biogeochemical cycles, hydrology, and biodiversity. Therefore, projecting future land use is crucial for global change impact analyses. This study compares harmonized land-use and management trends, analyzing uncertainties through a three-factor variance analysis involving socioeconomic-climate scenarios, land-use models, and climate models. The projected patterns are used as human-forcing inputs for the Intersectoral Impact Model Intercomparison Project phase 3b (ISIMIP3b) and multiple impact modeling teams. We employ two models (IMAGE and MAgPIE) to project future land use and management under three socioeconomic-climate scenarios (SSP1-RCP2.6, SSP3-RCP7.0, and SSP5-RCP8.5), driven by impact data like yields, water demand, and carbon stocks from updated climate projections of five global models, considering CO2 fertilization effects. On the global level, in the SSP1-RCP2.6 scenario (low adaptation and mitigation challenges), there is high agreement among land-use models on land-use trends. However, significant differences exist in management-related variables, such as the area allocated for second-generation bioenergy crops. Uncertainty in land-use variables increases with higher spatial resolution, particularly concerning the locations where cropland and grassland shrinkage could occur under this scenario. In SSP5-RCP8.5 and SSP3-RCP7.0, differences among land use models in global and regional trends are primarily associated with grassland area demand. Concerning the variance analysis, the selection of climate models minimally affects the variance in projections at different scales. However, the influence of the socioeconomic-climate scenarios, the land-use model, and interactions among the underlying factors on projected uncertainty varies for the different land-use and management variables. Our results highlight the need for more intercomparison exercises focusing on future spatially explicit projections to enhance understanding of the intricate interplay between human activities, climate, socioeconomic dynamics, land responses, and their associated uncertainties on the high-resolution level as models evolve.
- Preprint
(19288 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-2441', Anonymous Referee #1, 13 Sep 2024
This study utilizes two models (IMAGE and MAgPIE) to project future land use and management under three socioeconomic-climate scenarios. It compares harmonized land-use and management trends, and analyzes uncertainties arising from socioeconomic-climate scenarios, land-use models, climate models, harmonization processes, and the interactions between these factors. The findings indicate that uncertainty in land-use variables increases with higher spatial resolution, while the choice of climate models has minimal impacts on projection variance across scales. The study highlights the need for more intercomparison exercises focused on spatially explicit projections to improve understanding of the complex interactions between human activities, climate, socioeconomic dynamics, land responses, and associated uncertainties at high-resolution levels as models continue to evolve.However, there are some explanations that need further elaboration in the manuscript. Please refer to my detailed comments below.Major Comments:
- Clarification of Figures 7 and 8: Could the author clarify why the gridded differences are primarily influenced by the specific factors highlighted in Figures 7 and 8, rather than other potential variables? On Lines 438–441, the manuscript states: “One of the primary explanations for the effect of harmonization on forests is the different inputs regarding forests among the LUMs and LUH2 historical maps used in harmonization, especially in areas with intermediate tree cover. For example, global forest areas in 2000 range among different satellite sources and FAO between 3600 and 4300 million hectares (Ma et al., 2020)." While this provides some explanations for the effect of harmonization on forests, more detailed explanations of the dominance of the factors shown in Figures 7 and 8 would be helpful for better understanding.
- Comparison of ISIMIP3b and LUH2 Datasets: The manuscript compares the ISIMIP3b LUC with CMIP6 LUH2 data in various instances, such as on Lines 287–289: “This drop in demand for second-generation bioenergy crops is related to changes in the mitigation assumptions of SSP1-RCP2.6, which involves updated impacts on yields.” This is informative, but could the authors provide a more detailed explanation of the core differences between the ISIMIP3b and LUH2 datasets, and explain how these fundamental differences contribute to the observed discrepancies? This additional context would help the reader better understand the significance of ISIMIP3b LUC and understand why it differs from CMIP6 LUH2.
- Explanation of Equation 2: Please explain how interaction is defined and how the interaction calculation is conducted.
- Uncertainty from Land Use Downscaling: The land-use downscaling process could introduce uncertainty into the gridded LUC. I suggest the authors could discuss this uncertainty in the discussion section
Minor Comments:
- Lines 447-448: “However, we found some differences regarding the regional and local distribution of land-use change, specifically in cropland for the LAM region.” Please explain why this difference in cropland occurs.
- Lines 69-70: “... which has commonly been used for impact analyses in global and regional studies. (Yu et al., 2019; Qiu et al., 2023; Hoffmann et al., 2023).” Please check if the period before the parentheses needs to be removed.
- Lines 448-450: “For SSP5-RCP8.5 and SSP3-RCP7.0, global and regional trends disagree regarding the direction of change in grassland area, which leads to differences in forests and natural vegetation.” Please explain the potential reasons behind this.
- Figure B2: Did the study consider changes in pasture and forest yield in addition to crop yield?
- Line 460: “On the one hand, for example, LUMs have been used to conduct studies focused on China, India, or the European Union, which has involved further development and validation of the models’ outputs for these countries/regions (Singh et al., 2023; Wang et al., 2023; Veerkamp et al., 2020) on different resolutions.” Are these popularly studied regions showing better consistency among LUMs?
- Lines 465-467: “Second-generation bioenergy crops (Figures B7, B10-B13) are generally allocated in concentrated and highly fertile areas across all scenarios. These areas primarily include the west coast of Australia, southern Brazil, the Eastern European Plain (especially in SSP1-RCP2.6), Southeast Asia, southern China, and West Africa.” Please explain why these differences in bioenergy crop allocation occur.
- Lines 319-321: “More specifically, MAgPIE’s cropland allocation is based on minimizing production costs and local biophysical constraints, while IMAGE’s approach relies on a constant elasticity of transformation function, which associates land supply responsiveness with changes in yields and prices (Schmitz et al., 2014).” Could the author elaborate on how these model differences contribute to variations in the LUC results?
- Region Division (Figure B1): The globe is divided into five regions in the manuscript (Figure B1). Please explain the criteria for this division.
- Figure B2 Placement: Given the importance of this modeling protocol, I suggest moving Figure B2 into the main text.
Citation: https://doi.org/10.5194/egusphere-2024-2441-RC1 -
RC2: 'Comment on egusphere-2024-2441', Anonymous Referee #2, 14 Nov 2024
The study compares the results of two land use models under different scenarios. The study covers a wide range of sections and provides information that demonstrates the robustness of the application. However, some minor and major improvements can still be made:
- Line 46-48. Include more recent studies to support the frameworks and models to project and compare future land-use and land-use-related variables.
- Lines 61-75 should be part of the methods section.
- Line 100. Not clear how the demand for bioenergy production aligned with climate policies was determined.
- Given that the identification of uncertainties was one of the purposes of the study, authors should discuss more in detail the implications of them when these models are being used, in particular for policy making. More examples as the one presented in lines 454-456 are missing.
- Authors provide a rich set of results, however a summary key messages across land uses and global regions are missing, that is, messages that contextualise the value of the findings for decision making based on modelling outputs. This could be done in the abstract or in a conclusions section
Citation: https://doi.org/10.5194/egusphere-2024-2441-RC2
Data sets
ISIMIP 3b, Land Use Models outputs intercomparison Edna J. Molina Bacca https://zenodo.org/records/12964394
Model code and software
MAgPIE - An Open Source land-use modeling framework version 4.4.0 Jan Philipp Dietrich, Benjamin Leon Bodirsky, Isabelle Weindl, Florian Humpenöder, Miodrag Stevanovic, Ulrich Kreidenweis, Xiaoxi Wang, Kristine Karstens, Abhijeet Mishra, Felicitas Dorothea Beier, Edna Johanna Molina Bacca, Patrick von Jeetze, Michael Windisch, Michael Scott Crawford, David Klein, Vartika Singh, Geanderson Ambrósio, Ewerton Araujo, Anne Biewald, Hermann Lotze-Campen, and Alexander Popp https://github.com/magpiemodel/magpie/releases/tag/v4.4.0
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 212 | 60 | 14 | 286 | 13 | 7 |
- HTML: 212
- PDF: 60
- XML: 14
- Total: 286
- BibTeX: 13
- EndNote: 7
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1