Reconstruction natural flow in a regulated system, the Murrumbidgee River, Australia, using time series analysis

Abstract

In the area of floodplain restoration, river managers and ecologists emphasise the need to partially or fully restore a range of natural variations to flow regimes. The most common approach to investigate the natural flow has been the use of hydrological models. This study illustrates that multivariate time series analysis is a valid alternative. Thirty years of pre-regulation time series of monthly discharge, rainfall, and maximum temperature in Murrumbidgee Catchment are used to build the regressive models which represent the natural state of catchment hydrology. The models successfully arrest the dynamic behaviour of river discharges for both upstream and downstream stations; and demonstrate the lagged correlations between climatic variables and river discharges. Furthermore, the models also indicate there are strong autocorrelation in river discharge time series. The correlations between lagged climatic variables and river discharge and autocorrelations can be used to better understand the hydrologic response to climatic forcings. To demonstrate the application of this new approach, the fitted models are using to forecast the monthly river discharges from 1921 to 2007 at Balranald, Hay, and Wagga Wagga. The results indicate that river flows were increased significantly upstream (Wagga Wagga) on an annual base although the winter/spring flows were reduced. In addition, the peak flows are shifted from winter to summer, especially during drought periods. Conversely, for the downstream stations (Hay and Balranald), the river discharges are significantly reduced for all months; and the flow seasonality is close to the natural pattern. The results are generally consistent with the integrated quality and quantity model (IQQM) approach. The method to reconstruct river flow time series illustrated in the study can be adopted to evaluate the effects of river management plans and policies.