Comparison of autoregressive moving average and state space methods for mean monthly time series modelling of river flows in Labrador and South East Quebec

Sparkes, Carissa (2017) Comparison of autoregressive moving average and state space methods for mean monthly time series modelling of river flows in Labrador and South East Quebec. Masters thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (5MB)

Abstract

Time series data such as monthly stream flows can be modelled using time series methods and then used to simulate or forecast flows for short term planning. Short term forecasting can be applied, for example, in the hydroelectric industry to help manage reservoir levels at storage dam facilities as well as energy supply throughout the year at run of river facilities. In this study, two methods of time series modelling were reviewed and compared, the Box Jenkins autoregressive moving average (ARMA) method and the State- Space Time-Series (SSTS) method. ARMA has been used in hydrology to model and simulate flows with good results and is widely accepted for this purpose. SSTS modelling is a method that was developed in the 1990s for modelling economic time series and is relatively unused for modelling streamflow time series. The work described herein focuses on modelling stream flows from three selected basins in Labrador and South-East Quebec, the Alexis, Ugjoktok and Romaine Rivers, using these two time series modelling methods and comparing results. The ARMA and SSTS models for each study basin were compared for fit of model, accuracy of prediction, ease of use and simplicity of model to determine the preferred time series methodology approach for modelling the flows in these rivers. The performance of the methods for both the model fit as well as the forecast accuracy was measured using Nash Sutcliffe Coefficient, Median Absolute Percentage Error and Mean Squared Deviation error calculations. It was concluded that the SSTS method for these three rivers produced better fitting models than the ARMA method, but was generally equivalent in prediction accuracy to the ARMA method. The ARMA models, in contrast, were easier to diagnose and could be used to produce flow simulations, which was challenging using SSTS.

Actions (login required)

View Item