Predictive understanding of socioeconomic flood impact in data-scarce regions based on channel properties and storm characteristics: application in High Mountain Asia (HMA)

Abstract

High Mountain Asia (HMA) faces heightened vulnerability to natural disasters due to its extreme conditions and the escalating impacts of climate change. Understanding the long-term response of this landscape to hydroclimatic fluctuations is imperative, given the profound effects these changes have on millions of people annually. Heavy rain and the monsoon season bring forth floods and debris flows, resulting in significant damage to crops, infrastructure, and communities and having widespread human impacts. Despite efforts to estimate flood risk locally, traditional techniques often fall short due to the scarcity of high-quality, consistent data, especially in ungauged basins. To overcome this challenge, we propose a novel approach: a geomorphologically guided machine learning (ML) method for mapping flood effects across HMA. Central to our methodology is the life year index (LYI), a systematic measure that quantifies both the financial and the human losses incurred by disasters, specifically for this study fluvial and pluvial flooding. Our model was trained using a dataset comprising over 6000 flood events spanning 1980 to 2020, along with their corresponding 5- and 10-year LYI. Key predictors included (1) 5-year rainfall concentrations derived from ERA5 daily data, (2) a geomorphic classifier based on hydraulic scaling functions derived from high-resolution digital elevation models (DEM), and (3) population density. Results demonstrate the model’s effectiveness in identifying flood susceptibility hotspots at a national scale and delineating their evolution from 1980 to 2020. Moreover, the study underscores the severity of hydroclimatic extremes across the entire HMA region. Importantly, the proposed framework is versatile and can be adapted to generate various pluvial and fluvial flood vulnerability and risk maps in ungauged regions.