<?tk 4?>High-resolution, long-term land use datasets under multiple scenario frameworks are essential for understanding and simulating environmental changes (climate, hydrology, and ecosystem services), particularly in data-scarce regions like Ethiopia. This study develops a 1 km resolution land-use/cover (LULC) projection for Ethiopia, encompassing six LULC types under eight coupled socioeconomic and climatic scenarios (SSP-RCPs). We employ the Cellular Automata (CA)-based land use simulation model (FLUS) to simulate future LULC dynamics from 2020 to 2100 at decadal intervals, driven by land demand from the Land Use Harmonization 2 (LUH2) dataset. The simulations were informed by relevant spatial driving factors, including socioeconomic, distance-related, and natural forcings. The model achieved strong validation performance (kappa 0.89, overall accuracy (OA) 0.9, figure of merit (FoM) 0.17). Projections reveal substantial LULC changes across Ethiopia by 2100, characterized by declines in forest/shrubland, cropland and urban expansion, and shifts in the extent of grassland and barren land. Under high-pressure scenarios (SSP3-7.0, SSP4-3.4), cropland area is projected to more than double (+ 109%) while forest/shrubland declines by ~ 32% relative to the simulated 2015 baseline; in contrast, sustainability-oriented pathways (SSP1) limit cropland expansion to ~ 12–20% with forest losses below 8%. The observed changes in the magnitude and spatial pattern of land-use dynamics across the eight SSP-RCP scenarios indicate the coupled impacts of socioeconomic pathways and climate forcing on Ethiopia’s future land system. By offering finer spatial resolution and incorporating the latest coupled SSP-RCP scenarios, this dataset advances land-use planning, hydrological and climate modeling, supports sustainable development policies, and addresses future environmental challenges in Ethiopia. Graphical Abstract <?tk 4?>The graphical abstract illustrates the methodology used to project Ethiopia’s future LULC dynamics from 2020 to 2100 under coupled SSP-RCP scenarios, and highlights the study’s key results. The study employs a robust two-step framework that effectively integrates top-down land demand projections with bottom-up spatial allocation. First, future land demand for Ethiopia is derived from the LUH2 dataset, which provides global gridded LULC projections (2015–2100) under eight SSP-RCP scenarios at ~ 25 km resolution. Land demand is then spatially allocated at 1 km resolution using the FLUS model, which combines an ANN module for suitability analysis—trained with historical LULC data (1992–2015) and multiple spatial driving factors (climate, topography, socioeconomic, distance-related, etc.)—with a cellular automata (CA) allocation module that captures fine-scale landscape heterogeneity. The model’s reliability was confirmed through rigorous historical validation, achieving high accuracy (Kappa = 0.875, OA = 0.89). The resulting dataset offers: (i) temporal resolution at 10-year intervals, (ii) 1 km spatial resolution across Ethiopia, (iii) eight SSP-RCP scenario pathways, and (iv) six LULC categories. Projections consistently show cropland and urban expansion at the expense of natural vegetation. However, the magnitude and spatial pattern of this shift vary substantially across scenarios: high-pressure pathways (e.g., SSP3) drive severe forest and shrubland loss. In contrast, sustainability-oriented pathways (e.g., SSP1) substantially preserve natural ecosystems.

Future Land-Use/Land-Cover Projections under Coupled Socioeconomic and Climate Scenario (SSP-RCP) in Ethiopia

期刊论文
发表日期：

2026年5月5日

查看全文

Ermias Sisay Brhane,
Koji Dairaku

摘要

High-resolution, long-term land use datasets under multiple scenario frameworks are essential for understanding and simulating environmental changes (climate, hydrology, and ecosystem services), particularly in data-scarce regions like Ethiopia. This study develops a 1 km resolution land-use/cover (LULC) projection for Ethiopia, encompassing six LULC types under eight coupled socioeconomic and climatic scenarios (SSP-RCPs). We employ the Cellular Automata (CA)-based land use simulation model (FLUS) to simulate future LULC dynamics from 2020 to 2100 at decadal intervals, driven by land demand from the Land Use Harmonization 2 (LUH2) dataset. The simulations were informed by relevant spatial driving factors, including socioeconomic, distance-related, and natural forcings. The model achieved strong validation performance (kappa 0.89, overall accuracy (OA) 0.9, figure of merit (FoM) 0.17). Projections reveal substantial LULC changes across Ethiopia by 2100, characterized by declines in forest/shrubland, cropland and urban expansion, and shifts in the extent of grassland and barren land. Under high-pressure scenarios (SSP3-7.0, SSP4-3.4), cropland area is projected to more than double (+ 109%) while forest/shrubland declines by ~ 32% relative to the simulated 2015 baseline; in contrast, sustainability-oriented pathways (SSP1) limit cropland expansion to ~ 12–20% with forest losses below 8%. The observed changes in the magnitude and spatial pattern of land-use dynamics across the eight SSP-RCP scenarios indicate the coupled impacts of socioeconomic pathways and climate forcing on Ethiopia’s future land system. By offering finer spatial resolution and incorporating the latest coupled SSP-RCP scenarios, this dataset advances land-use planning, hydrological and climate modeling, supports sustainable development policies, and addresses future environmental challenges in Ethiopia.

Graphical Abstract

The graphical abstract illustrates the methodology used to project Ethiopia’s future LULC dynamics from 2020 to 2100 under coupled SSP-RCP scenarios, and highlights the study’s key results. The study employs a robust two-step framework that effectively integrates top-down land demand projections with bottom-up spatial allocation. First, future land demand for Ethiopia is derived from the LUH2 dataset, which provides global gridded LULC projections (2015–2100) under eight SSP-RCP scenarios at ~ 25 km resolution. Land demand is then spatially allocated at 1 km resolution using the FLUS model, which combines an ANN module for suitability analysis—trained with historical LULC data (1992–2015) and multiple spatial driving factors (climate, topography, socioeconomic, distance-related, etc.)—with a cellular automata (CA) allocation module that captures fine-scale landscape heterogeneity. The model’s reliability was confirmed through rigorous historical validation, achieving high accuracy (Kappa = 0.875, OA = 0.89). The resulting dataset offers: (i) temporal resolution at 10-year intervals, (ii) 1 km spatial resolution across Ethiopia, (iii) eight SSP-RCP scenario pathways, and (iv) six LULC categories. Projections consistently show cropland and urban expansion at the expense of natural vegetation. However, the magnitude and spatial pattern of this shift vary substantially across scenarios: high-pressure pathways (e.g., SSP3) drive severe forest and shrubland loss. In contrast, sustainability-oriented pathways (e.g., SSP1) substantially preserve natural ecosystems.