Author(s): D. E. Robertson; Q. J. Wang; T. C. Pagano; H. A. P. Hapuarachchi
Linked Author(s):
Keywords: Seasonal streamflow forecasting; Water-balance modelling
Abstract: Statistical seasonal streamflow forecasting methods commonly use antecedent streamflows as predictors to represent initial catchment conditions. However, antecedent streamflows do not necessarily capture the full dynamics of catchment processes. This paper investigates the use of water balance model output as an alternative representation of the initial catchment conditions. The Bayesian joint probability modelling approach is used to produce seasonal forecasts using two sets of predictors representing initial catchment conditions: predictors selected from a pool of candidates comprising antecedent streamflow and rainfall totals, and simulations made using a monthly water balance model. In general, using water balance model output as predictors results in more skilful streamflow forecasts, with the greatest differences being in seasons where the dominant source of streamflows transitions between surface runoff and base flow. The use of water balance model output as predictors also has the advantage of reducing artificial skill due to predictor selection.
Year: 2011