In-stream <i>Escherichia coli</i> modeling using high-temporal-resolution data with deep learning and process-based models

Abstract. Contamination of surface waters with microbiological pollutants is a major concern to public health. Although long-term and high-frequency Escherichia coli (E. coli) monitoring can help prevent diseases from fecal pathogenic microorganisms, such time-consuming expensive. Process-driven models are an alternative means for estimating concentrations pathogens. However, process-based modeling still has limitations in improving the model accuracy because complexity relationships among hydrological environmental variables. With rise data availability computation power, use data-driven increasing. In this study, we simulated fate transport E. 0.6 km2 tropical headwater catchment located Lao People's Democratic Republic (Lao PDR) using deep-learning model. The deep learning was built long short-term memory (LSTM) methodology, whereas constructed Hydrological Simulation Program–FORTRAN (HSPF). First, calibrated both as well subsurface flow. Then, 6 min time steps HSPF LSTM models. provided accurate results flow 0.51 0.64 Nash–Sutcliffe efficiency (NSE) values, respectively. contrast, NSE values yielded by were ?0.7 0.59 0.35, gave unacceptable performance value ?3.01 due capturing dynamics land-use change. concentration showed drop patterns corresponding annual changes land use. This study showcases application deep-learning-based efficient simulation at scale.