Multicriteria analysis of water quality models for simulating Cryptosporidium oocysts at the watershed scale

Cryptosporidium is an environmentally robust pathogen responsible for severe waterborne disease. As a result, the emergence of Cryptosporidium in surface water has become an increasing concern for drinking water providers. However, simulating Cryptosporidium with available water quality models remains challenging due to its environmental persistence. With limited work available on modeling Cryptosporidium loading, a systematic review is warranted to examine the potential use of water quality models. This review considered twelve water quality models for simulating Cryptosporidium loading, including BIT, BSLC, GWLF, HSPF, PCB, PLOAD, SWAT, SWMM, WAM, WARMF, WASP, and WATNAT-PROMISE. Several screening criteria were used, including the ability to simulate Cryptosporidium loading, model accessibility (i.e., available, well-documented, transparency in the application, and ease of use), the inclusion of point and non-point sources, and the ability to predict spatiotemporal variations, along with simulations of pathogen transport and physical-chemical processes. Four models, SWAT, HSPF, GWLF, and WAM, met screening criteria and were studied for further assessment based on full set of criteria, including input parameters required for model development (i.e., soil characteristics, hydrology, climate, point and non-point sources, water quality, land use, water body, BMPs, correlated parameters, database availability, and processes related to Cryptosporidium survival and release) along with the model's capacity for calibration, validation and sensitivity with uncertainty analysis. This evaluation identified SWAT as the only model meeting all criteria, offering the greatest potential for simulating Cryptosporidium loading. However, the review also highlighted that other models (i.e., HSPF, GWLF, and WAM) may be more appropriate for specific applications depending on data availability, catchment size, modeling objectives, computational and calibration demands, and regulatory and screening-level requirements. The SWAT model has a microbial sub-model that provides flexible functions for bacterial dynamics and adaptable processes to model pathogen transport from various sources. Additionally, integrated tools include SWAT-CUP that improves the calibration process and addresses model uncertainty, making it a more robust and efficient model compared to others. Furthermore, SWAT incorporates the spatial heterogeneity of watershed features, including soil properties, supporting a comprehensive analysis of spatiotemporal trends in Cryptosporidium loading.