A combined deterministic and self-adaptive stochastic algorithm for streamflow forecasting with application to catchments of the Upper Murray Basin, Australia

S. Yu. Schreider, a A. J. Jakeman, a* B. G. Dyer, b R. I. Francis b aCentre for Resource and Environmental Studies, Australian National University, Canberra 0200, ACT, Australia bMurray-Darling Basin Commission, Canberra 2601, ACT, Australia Abstract This paper describes the results of runoff modelling for nine catchments of the Upper Murray Basin (Basin 401) of the Murray- Darling Drainage Division (MDDD), Australia. The work aimed firstly to provide adequate models for long-term streamflow prediction in nine catchments of this Basin feeding the Hume and Dartmouth reservoirs. The development and testing of flow forecasting algorithms for operational management by the Murray-Darling Basin Commission was another purpose of the work reported here. The conceptual lumped parameter rainfall-runoff model IHACRES (Jakeman et al., 1990, 1993; Jakeman and Hornberger, 1993) was selected as the modelling tool for streamflow prediction in the catchments. The conceptual rainfall-runoff model IHACRES (with a snow melt/formation module in snow-affected catchments) and a self- adaptive linear filtering approach for the IHACRES residuals were combined and applied for forecasting daily streamflow in the Upper Murray Basin catchments. Different orders of AutoRegressive Integrated Moving Average (ARIMA) models for the residuals were considered in order to select the most appropriate forecasting algorithm. Linear filtering of the conceptual model residuals provides considerable improvement in forecasting for both low and high values of streamflow for developing the operational streamflow forecast system. © 1997 Elsevier Science Ltd. Keywords: Streamflow modelling; operational streamflow forecasting; ARIMA algorithm 1. Introduction 1.1. Background Overton and Meadows (1976) define three basic categories of streamflow forecasting methods based *Corresponding author. 93 upon: (1) regression analysis, (2) time series analysis and (3) flow frequency analysis. The regression-type analysis uses an optimisation procedure where a causal model is structured as a linear or slightly non-linear approximation. Least squares (or modified least squares) regression serves as a tool for this approxi- mation of modelled values against empirical data. Although not strictly a regression model, the IHACRES

[1]  R. E. Kalman,et al.  A New Approach to Linear Filtering and Prediction Problems , 2002 .

[2]  A. I. McLeod,et al.  On the Distribution of Residual Autocorrelations in Box–Jenkins Models , 1978 .

[3]  I. Rodríguez‐Iturbe,et al.  Random Functions and Hydrology , 1984 .

[4]  Armando Brath,et al.  Adaptive calibration of a conceptual model for flash flood forecasting , 1993 .

[5]  W. E. Watt,et al.  Multivariate Transfer Function-Noise Model of River Flow for Hydropower Operation , 1986 .

[6]  Anthony J. Jakeman,et al.  Assessing uncertainties in hydrological response to climate at large scale , 1993 .

[7]  S. Yakowitz,et al.  Nearest Neighbor Methods for Time Series, with Application to Rainfall/Runoff Prediction , 1987 .

[8]  Anthony J. Jakeman,et al.  Runoff modelling for snow-affected catchments in the Australian alpine region, eastern Victoria , 1997 .

[9]  G. Kember,et al.  Forecasting river flow using nonlinear dynamics , 1993 .

[10]  Anthony J. Jakeman,et al.  A parametrically efficient model for prediction of streamflow in an Australian benchmark catchment with complex storage dynamics , 1995 .

[11]  Konstantine P. Georgakakos,et al.  Real‐Time Flash Flood Prediction , 1987 .

[12]  S. Yakowitz,et al.  Rainfall-runoff forecasting methods, old and new , 1987 .

[13]  G. Galeati.  A comparison of parametric and non-parametric methods for runoff forecasting , 1990 .

[14]  S. Yakowitz.  NEAREST‐NEIGHBOUR METHODS FOR TIME SERIES ANALYSIS , 1987 .

[15]  M. Rosenblatt,et al.  Multivariate k-nearest neighbor density estimates , 1979 .

[16]  George E. P. Box,et al.  Time Series Analysis: Forecasting and Control , 1977 .

[17]  Anthony J. Jakeman,et al.  MODELLING RAINFALL‐RUNOFF FROM LARGE CATCHMENT TO BASIN SCALE: THE GOULBURN VALLEY, VICTORIA , 1996 .

[18]  Anthony J. Jakeman,et al.  Estimation of possible climate change impacts on water availability, extreme flow events and soil moisture in the Goulburn and Ovens Basins, Victoria , 1996 .

[19]  A. Jakeman,et al.  How much complexity is warranted in a rainfall‐runoff model? , 1993 .

[20]  E. Todini.  The ARNO rainfall-runoff model , 1996 .

[21]  Z. Şen.  Adaptive prediction of hydrologic series by Walsh-Kalman model , 1991 .

[22]  P. Whetton,et al.  Climate change and snow-cover duration in the Australian Alps , 1996 .

[23]  Gwilym M. Jenkins,et al.  Time series analysis, forecasting and control , 1972 .

[24]  A. Jakeman,et al.  Computation of the instantaneous unit hydrograph and identifiable component flows with application to two small upland catchments , 1990 .

[25]  J. Nash,et al.  River flow forecasting through conceptual models part I — A discussion of principles☆ , 1970 .