This thesis focuses on investigating how streamflow predictions can be improved by using sequential data assimilation with particle filtering in the HBV hydrological model. Particle filtering, which is a Bayesian inference technique, is incorporated into the HBV model to address uncertainties and improve prediction accuracy by updating model states and parameters using observed data.

The study involved implementing particle filtering in the HBV model and assessing its performance against real world streamflow data. The results displayed big improvements in prediction accuracy, indicating the effectiveness of particle filtering in hydrological forecasting.

This research emphasizes the potential of advanced data assimilation methods such as particle filtering to enhance the reliability of hydrological models, thereby contributing to improved water resource management.