Incorporating Prior Knowledge in the Calibration of Hydrological Models for Water Resources Forecasting

Abstract

The management of water resources in Australia faces increasing challenges due the rise of conflicting demands and a highly variable climate. In this context, the Bureau of Meteorology developed a dynamic seasonal forecasting service providing probabilistic forecasts of river flow at selected locations across Australia by coupling rainfall forecasts from a Global Circulation Model with a rainfall–runoff model. The chapter presents a method to improve the Bayesian inference of the rainfall–runoff model parameters by using an informative prior derived from the calibration of the model on a large sample of catchments. This prior is compared with a uniform prior that is currently used in the system. The results indicate that the choice of the prior can have a significant impact on forecast performance for both daily and monthly time steps. The use of an informative prior generally improved the performance, especially for one test catchment at daily time step where prediction intervals were narrowed without compromising forecast reliability. For other catchments and time steps, the improvement was more limited.