Rahman et al. (2026) Spatiotemporal dynamics of flood susceptibility under future precipitation variability, population growth, and land cover change

Identification

• Journal: International Journal of Applied Earth Observation and Geoinformation
• Year: 2026
• Date: 2026-02-23
• Authors: Zahid Ur Rahman, Meimei Zhang, Fang Chen, Safi Ullah, Lei Wang, Zahoor Ahmad, Muhammad Fahad Baqa
• DOI: 10.1016/j.jag.2026.105193

Research Groups

• Key Laboratory of Digital Earth Science, Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing, China
• International Research Center of Big Data for Sustainable Development Goals, Beijing, China
• University of Chinese Academy of Sciences, Beijing, China
• Hamad Bin Khalifa University (HBKU), Qatar Foundation, Doha, Qatar

Short Summary

This study assesses the spatiotemporal dynamics of flood susceptibility in the Kabul River Basin (KRB) from 2020 to 2100 under future precipitation variability, population growth, and land cover change. Findings indicate a significant increase in areas with "Very High" flood susceptibility and a decline in "Very Low" susceptibility zones, particularly in the central and southern regions, driven by these interacting dynamic factors.

Objective

• To assess the spatiotemporal patterns of projected flood susceptibility in the Kabul River Basin (KRB) from 2020 to 2100 under different future scenarios, integrating the combined effects of precipitation variability, population growth, and land cover changes.

Study Configuration

• Spatial Scale: Kabul River Basin (KRB), approximately 57,600 square kilometers, with flood susceptibility maps generated at 30 meter resolution.
• Temporal Scale: Projected from 2020 (baseline) to 2100, with specific assessment periods for 2040, 2060, 2080, and 2100.

Methodology and Data

• Models used:
  • eXtreme Gradient Boosting (XGBoost) machine learning model for flood susceptibility prediction and data downscaling.
  • Cellular Automata (CA) Markov chain for future land cover projection.
  • SHapley Additive exPlanations (SHAP) for model interpretability.
  • Taylor diagram for prediction comparison and consistency analysis.
  • Bootstrap ensemble approach for uncertainty estimation.
• Data sources:
  • Static Predictors: NASA ASTER Digital Elevation Model (DEM) (30 m) for elevation, slope, curvature, drainage density, topographic diversity (TD), topographic position index (TPI), and topographic wetness index (TWI). Landsat-9 imagery (30 m) for Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water Index (NDWI).
  • Dynamic Predictors:
    • Population density: Global 1 km-Grid Population Distributions (SSP2-4.5 scenario) and WorldPop 1 km dataset, downscaled to 30 m.
    • Precipitation: WorldClim CMIP6 Global Climate Models (1 km) under SSP2-4.5, complemented by NASA GPM observed precipitation data (1 km), downscaled to 30 m.
    • Land cover: MODIS dataset (500 m) for 2000 and 2020, with future projections using CA-Markov chain, downscaled to 30 m.
  • Flood Inventory: Flood point data compiled from media reports, scientific literature, field data, and satellite imagery. Non-flood samples obtained using a stratified random approach.

Main Results

• Areas classified as "Very Highly" susceptible to floods are projected to increase from 11.78% in 2020 to 12.17% in 2040, 14.44% in 2060, 13.32% in 2080, and 13.51% by 2100.
• Conversely, areas classified as "Very Low" susceptibility are projected to steadily decline from 66.17% in 2020 to 56.43% by 2100.
• The XGBoost model demonstrated strong predictive accuracy (Area Under the Curve (AUC): 0.961–0.962) and high cross-temporal consistency across future scenarios (Pearson Correlation Coefficient: 0.75–0.85).
• Bootstrap uncertainty analysis confirmed model robustness with mean AUCs ranging from 0.9817 to 0.9834, very low standard errors (0.0003), and narrow 95% confidence intervals (0.9719–0.9887).
• SHAP analysis identified slope (20.5%), topographic position index (TPI) (16.6%), and population density (15.3%) as the most influential predictors. Population density showed the largest positive contribution to flood susceptibility (SHAP value of +3.71).
• Projected population density increased significantly, with urban hotspots reaching 35,029 persons/km² by 2100 from 12,922 persons/km² in 2020.
• Precipitation showed an increasing trend, with maximum values rising from 1108 mm in 2020 to 1215 mm in 2060.
• Built-up areas are projected to increase by 21.61% from 2020 to 2100, primarily displacing cropland and shrubland.

Contributions

• Provides a comprehensive spatiotemporal assessment of future flood susceptibility (2020–2100) in a vulnerable transboundary region (Kabul River Basin) by integrating dynamic predictors (precipitation variability, population growth, land cover change) with static factors.
• Introduces a novel integrated multi-drivers modeling framework that combines downscaled CMIP6 climate projections, CA-Markov-based land-cover simulations, and an advanced supervised machine learning model (XGBoost) at a high spatial resolution (30 m).
• Enhances model transparency and reliability through the use of SHAP for predictor interpretability, Taylor-diagram-based cross-temporal consistency analysis, and bootstrap ensemble uncertainty estimation.
• Offers a transferable framework for flood assessment in other climate-sensitive mountainous regions (e.g., Indus, Brahmaputra, Mekong basins).
• Generates actionable insights for land use planning, disaster preparedness, and climate adaptation policies to mitigate future flood impacts in the KRB.

Funding

• National Natural Science Foundation of China (No. 42425103)
• Central Guiding Local Science and Technology Development Fund of Shandong-Yellow River Basin (No. YDZX2023019)
• Joint HKU-CAS Laboratory for iEarth (No. 313GJHZ2022074MI, E4F3050300)
• CAS-TWAS Centre of Excellence on Space Technology for Disaster Mitigation

Citation

@article{Rahman2026Spatiotemporal,
  author = {Rahman, Zahid Ur and Zhang, Meimei and Chen, Fang and Ullah, Safi and Wang, Lei and Ahmad, Zahoor and Baqa, Muhammad Fahad},
  title = {Spatiotemporal dynamics of flood susceptibility under future precipitation variability, population growth, and land cover change},
  journal = {International Journal of Applied Earth Observation and Geoinformation},
  year = {2026},
  doi = {10.1016/j.jag.2026.105193},
  url = {https://doi.org/10.1016/j.jag.2026.105193}
}