Two decades of anarchy? Emerging themes and outstanding challenges for neural network river forecasting

Abstract

This paper traces two decades of neural network rainfall-runoff and streamflow modelling, collectively termed ‘river forecasting’. The field is now firmly established and the research community involved has much to offer hydrological science. First, however, it will be necessary to converge on more objective and consistent protocols for: selecting and treating inputs prior to model development; extracting physically meaningful insights from each proposed solution; and improving transparency in the benchmarking and reporting of experimental case studies. It is also clear that neural network river forecasting solutions will have limited appeal for operational purposes until confidence intervals can be attached to forecasts. Modular design, ensemble experiments, and hybridization with conventional hydrological models are yielding new tools for decision-making. The full potential for modelling complex hydrological systems, and for characterizing uncertainty, has yet to be realized. Further gains could also emerge from the provision of an agreed set of benchmark data sets and associated development of superior diagnostics for more rigorous intermodel evaluation. To achieve these goals will require a paradigm shift, such that the mass of individual isolated activities, focused on incremental technical refinement, is replaced by a more coordinated, problem-solving international research body.

Developing predictive insight into changing water systems: use-inspired hydrologic science for the Anthropocene

“…To begin with, data-driven modeling approaches offer an avenue towards the identification of process coupling as revealed in time series records (Abrahart et al, 2012;See et al, 2007;Young, 2013). The information theoretic approach of Ruddell and Kumar is one methodology that can extract signatures of coupling from time series records (Ruddell and Kumar, 2009a, b).…”

Successive-station monthly streamflow prediction using different artificial neural network algorithms

“…In recent years, ANN are increasingly used for prediction and pattern recognition problems in various fields of water and environmental science and technology such as total ozone forecasting (Bandyopadhyay and Chattopadhyay 2007), sea level prediction (Altunkaynak 2007, Imani et al 2013, rainfallrunoff modeling (De Vos and Rientjes 2005, Kuok et al 2010, Nourani et al 2011), water quality prediction (Emamgholizadeh et al 2013). A comprehensive literature review about the application of ANN in river forecasting was presented by Abrahart et al (2012). The review shows inadequate comparative investigations of ANN-based streamflow prediction models.…”

Prediction of dissolved oxygen concentration in hypoxic river systems using support vector machine: a case study of Wen-Rui Tang River, China

“…In addition, BPNN and GRNN both have shown great potential in predicting DO concentration (Antanasijević et al 2013a(Antanasijević et al , 2014aWen et al 2013). MLR is a well-known statistical technique for establishing linear links between input and output variables and is the minimum standard for model intercomparison (Abrahart et al 2012). We expect that the proposed SVM model will be an efficient tool for water quality management and pollution control of Wen-Rui Tang River and other hypoxic river systems.…”