eCite Digital Repository

Recurrent Neural Networks for One Day Ahead Prediction of Stream Flow


Mhammedi, Z and Hellicar, A and Rahman, A and Kasfi, K and Smethurst, P, Recurrent Neural Networks for One Day Ahead Prediction of Stream Flow, proc eedings of the Joint Workshop on Time Series Analytics and Collaborative Agents Research & Desig, 5 December 2016, Hobart, Australia, pp. 25-31. ISBN 978-1-4503-4820-1 (2016) [Non Refereed Conference Paper]

Not available

DOI: doi:10.1145/3014340.3014345


Accurate models for one day ahead prediction of stream flow are crucial for water management catchment scale for agriculture. This is particularly important for a country such as Australia where weather conditions can be harsh and varying. The Support Vector Regression (SVR) and the Vector Auto Regression (VAR) are standard methods used for time series prediction [12, 14]. However, since they use fixed-sized time windows, these models cannot capture long-term dependencies that are often present in stream flow time series. Recurrent Neural Networks (RNNs) do not have this weakness, yet they have not been extensively used on this type of time series. In this work, we tested various types of RNN architectures, including the recently introduced clock-work RNN (CW-RNN) [10], on two different stream flow datasets in Tasmania. We compared their accuracy with that of the SVR and VAR methods on the task of one day ahead prediction of the stream flow. In our experiments, the CW-RNNs outperformed the SVR and VAR methods across both datasets. In particular, when evaluated on the largest test set, which contained approximately 4 years of daily records, the normalised Root Mean Squared Error (nRMSE) and the Nash-Sutcliffe Efficiency (NSE) of the best CW-RNN architecture were equal to 0.166 and 0.956, respectively. This is a significant improvement over the best SVR model, which had nRMSE = 0.202 and NSE = 0.936. Our results suggest that RNNs are well suited for the task of one day ahead prediction of stream flow.

Item Details

Item Type:Non Refereed Conference Paper
Keywords:multivariate time series predictions, recurrent neural networks, LSTM, clockwork RNN, support vector regression, vector auto regression
Research Division:Information and Computing Sciences
Research Group:Artificial intelligence
Research Field:Modelling and simulation
Objective Division:Expanding Knowledge
Objective Group:Expanding knowledge
Objective Field:Expanding knowledge in the environmental sciences
UTAS Author:Smethurst, P (Dr Philip Smethurst)
ID Code:117121
Year Published:2016
Web of Science® Times Cited:5
Deposited By:Engineering
Deposited On:2017-05-31
Last Modified:2017-05-31

Repository Staff Only: item control page