To evaluate freshwater challenges in southwestern Bangladesh under both current and future climate conditions, we developed an agent-based water-scarcity model that integrates dynamically downscaled climate projections with detailed socioeconomic and water-resource data. This agent-based approach captures the heterogeneity of household behavior and its feedback with local water availability, enabling bottom-up assessment of water-scarcity risks. The model aims to identify fine-scale water-scarcity hotspots and to inform targeted adaptation strategies within the Jameel Observatory–CREWSnet framework.
Each household is represented as an autonomous agent characterized by income, household size, water demand, preferred water sources, and roof area relevant to rainwater harvesting (RWH). The model simulates household decision-making regarding the selection of drinking-water sources and associated expenditures under varying climatic and hydrological conditions.
To quantify vulnerability, the model computes a Water Poverty Index (WPI), defined as the ratio of monthly household freshwater expenditure to income, where higher values indicate greater water stress. By combining hydrological variability with socioeconomic heterogeneity, the model enables detailed assessment of present and future water-scarcity hotspots and supports the design of adaptive interventions, such as optimized placement of RWH systems or reverse-osmosis (RO) plants.
To support model development and calibration, we collected/produced a comprehensive suite of datasets encompassing socioeconomic, hydrological, and climatic dimensions. Household demographics, income, and infrastructure characteristics such as roof and tank sizes were obtained from the Bangladesh Bureau of Statistics (BBS), the Household Income and Expenditure Survey (HIES), and Karim (2010). Water-source availability, groundwater quality, rainfall, and water prices were compiled from BBS, the Bangladesh Water Development Board (BWDB), TRMM, MRCM, Y-RISE, and BRAC. High-resolution climate projections were derived from 10-km MRCM simulations for 1975–2050, dynamically downscaled from three CMIP6 models under multiple Shared Socioeconomic Pathway (SSP) scenarios. Outputs from our rainwater-harvesting (RWH) model were incorporated to represent spatial variations in rainwater potential, while both observed and projected groundwater-salinity fields are, or will be, used to refine patterns of groundwater vulnerability. Field data from BRAC’s household water-price campaign provide essential inputs for calibrating expenditure behavior and validating the model’s economic components.
Physical variables
Rainfall, GW level, GW aquifer, GW salinity level, GW arsenic level
Socio-economic variables
household income, price of water, # of household, household size, source of drinking water, roof size, water tank size
Learn more about how we’re using the agent-based water scarcity model to support the Water Security Flagship.