Climate-Induced Vulnerability Assessment of Small Watersheds Using a CRITIC–GMDH Hybrid Model: A Methodology Tailored for Data-Scarce Regions

Small watersheds play a crucial role in sustaining river hydrology, ecological flows and local water security. However, they are increasingly threatened by climate change, rapid transformations of land use and escalation of anthropogenic pressures. These problems are worse in areas with little data, where few hydrological observations, sparse monitoring networks, and inconsistent long-term datasets make it hard to accurately assess vulnerability and make plans. To address this critical gap, this study introduces a unique and data-efficient Criteria Importance Through Intercriteria Correlation- Group Method of Data Handling (CRITIC-GMDH) hybrid framework, specifically developed to accurately assess watershed vulnerability in regions where large, continuous, or high-resolution datasets are unavailable. This interpretable decision-support approach integrates CRITIC for objective indicator weighting with the nonlinear modelling capability of the GMDH, enabling robust vulnerability prediction under constrained data conditions, overcoming key limitations of conventional hydrological models and black-box machine learning techniques. The framework incorporates eleven hydro-meteorological, geomorphological, and socio-economic parameters, including rainfall, temperature, runoff, watershed area, watershed length, water quality index, average slope, forest area, impervious area, population density, and highest flood level. The approach is demonstrated across four major river basins in Northeast India, such as Gomati, Haora, Khowai, and Manu, which represent highly sensitive and partially transboundary catchments. Future climate projections from CMIP6 SSP1-2.6 and SSP5-8.5 scenarios were used to compute the Vulnerability Index across decadal periods (2005–2065). Results show a significant escalation in vulnerability, particularly under SSP5-8.5, with Haora and Gomati exhibiting Vulnerability Index > 0.85, indicating extreme exposure to climate extremes, and urbanization stress. Sensitivity analysis identifies rainfall, runoff, and temperature as dominant controlling parameters, and validation through the Falkenmark indicator and green-blue water stress indices confirms emerging scarcity risks. The study provides a scientifically grounded pathway for watershed prioritization and climate-resilient planning, offering an adaptable methodological foundation for sustainable management of small river systems in data-scarce regions.