A conceptual Bayesian Network model to characterise faecal pollution risk from pit latrines to groundwater in a rural tropical catchment

Groundwater is the primary source of drinking water for many rural communities in sub-Saharan Africa, yet its vulnerability to faecal contamination from pit latrines remains poorly understood, particularly in tropical, data-limited catchments. This study developed a conceptual Bayesian Network (BN) model to assess the risk of faecal indicator organism (FIO) contamination of groundwater from pit latrines in the Upper Zambezi catchment, Zambia. Model variables were identified through a literature review. The BN structure was developed in GeNIe Modeler, Academic Version 5.0.5310.0, and refined through structured expert elicitation and stakeholder consultations. Preliminary model parameterisation was based on literature-derived information. Pit FIO concentration and pit latrine density were identified as the primary determinants of groundwater contamination risk, while temperature and sludge moisture influence FIO survival. Rainfall and soil hydraulic conductivity govern contaminant transport by affecting groundwater recharge and water table depth. A rapid assessment categorised the risk of FIO contamination of groundwater from pit latrines in the Upper Zambezi catchment as low to moderate. The study demonstrates the potential of BNs to support groundwater FIO risk assessment in data-limited environments, where conventional process-based modelling is constrained by limited data availability and uncertainty. The proposed framework provides a foundation for targeted data collection, model validation and the development of decision-support tools for climate-resilient groundwater protection and rural sanitation management. Future work should focus on empirical parameterisation and validation of the model using groundwater microbiological monitoring, sanitary surveys, and targeted investigations of FIO transport from pit latrines to groundwater.