The risk of infection by enteric pathogens in bathing waters is generally monitored by using fecal indicator bacteria (FIB). Mechanistic models are efficient tools to predict FIB concentrations in bathing waters, both in near-future forecasting and in long-term climate change projections. However, most existing mechanistic FIB models are limited by the availability of observations for validation and incorporation of all relevant physical, biological, and chemical (physico-biochemical) processes. Therefore, the quantitative influence of different physio-biochemical processes and impact factors is missing. To enhance the understanding of FIB fate in different aquatic systems, we developed a comprehensive yet generically applicable physico-biochemical model, focused on Escherichia coli (E. coli). It includes a die-off module and a sediment interaction module. Separate validation of the two sub-modules demonstrated the reliability of our modeling approach. The die-off module shows a higher R² value (0.88) and lower RMSE value (1.1 day^-1) than the existing models (0.48–0.79, and 1.8–7.2 day ^-1). This demonstrated an improvement by adding Ultraviolet-A and Ultraviolet-B (UVB) inactivation and UV spectrum extinction due to colored dissolved organic matter (CDOM) absorption. According to our sediment module validation, considering the impact of sediment composition on E. coli attachment can improve the allocation of E. coli between waters and sediments. Sensitivity analysis showed that 1) photo-inactivation is important in low CDOM waters, but not in high CDOM waters, where the UV penetration is limited; 2) the impact of sediment interaction can extend the duration of a peak event in high turbid waters. This work demonstrated the dominant impact factors in different aquatic systems for E. coli prediction. The new generic model enables better simulation of bathing water quality across different types of aquatic environments, which can be a useful tool to inform management at bathing sites. Future applications can choose processes selectively from the new FIB physico-biochemical model and couple it with appropriate hydrological/hydrodynamic models to address specific environmental conditions and research purposes.

Accurate and reliable flow forecasting form an important basis for efficient real-time river management, including flood control, flood warning and so on. In order to improve the accuracy of flow forecasting, gain matrix of Kalman filter was applied to real-time correction of hydraulic model for spatial distributing the system deviation (called expected value of system noise in Kalman filter). That means Kalman gain matrix is used to distribute model system deviation from measurement cross sections to the entire state of the river system. State functions of Kalman filter were set up based on discretization and linearization Saint-Venant equations by adopting four-point linear implicit form, and the spatial distribution system deviation method (SDM) was used for real-time correction. The calculation of flood forecasting for river section from Cuntan to Fengjie of Yangtze River verifies that SDM is useful in promoting the accuracy of real-time flood forecasting.