Environmental factors and sources influencing Cryptosporidium river concentrations: A data-rich spatio-temporal analysis using Explainable AI

Cryptosporidium is a waterborne parasite which poses significant risks to drinking water safety. Despite its importance, considerable uncertainty remains regarding the effect of catchment conditions on the scale and timing of Cryptosporidium concentrations in source waters. This paper performs a spatio-temporal analysis of the environmental factors influencing Cryptosporidium in rivers by applying eXplainable AI (XAI) tools to a large-scale nationwide dataset. Comprising ≈ 50,000 measurements taken at 52 diverse sites across the British mainland over a 24-year period, the study is unique in bringing together extensive land use, topographical, meteorological, hydrological and human/animal density data to analyse the seasonality and magnitude of Cryptosporidium river concentrations, and to identify the conditions most strongly associated with elevated values. A machine learning (ML) model – XGBoost – was trained with environmental inputs at different spatial scales. Interpretation of feature effects using SHapley Additive exPlanations (SHAP) found that humidity and rainfall had a strong positive effect on Cryptosporidium concentrations. Human population density was shown to be more highly predictive of Cryptosporidium concentrations than either grassland (grazing) areas or livestock density, suggesting that human sources of Cryptosporidium may be more important than animal sources in many rivers. Density of combined sewer overflows (CSOs) and sewage treatment works (STWs) were found to be more effective predictors at the local (≤5 and ≤25 km upstream) scale, pointing the possible impact of local point sources. The study revealed strong seasonality in Cryptosporidium at many sites, with the timing of peak concentrations varying significantly with catchment characteristics, such as human population, STW density, urban area and prevailing meteorological conditions.