Christian W. Dawson

SDSM - a decision support tool for the assessment of regional climate change impacts

General Circulation Models (GCMs) suggest that rising concentrations of greenhouse gases will have significant implications for climate at global and regional scales. Less certain is the extent to which meteorological processes at individual sites will be affected. So-called ‘downscaling’ techniques are used to bridge the spatial and temporal resolution gaps between what climate modellers are currently able to provide and what impact assessors require. This paper describes a decision support tool for assessing local climate change impacts using a robust statistical downscaling technique. Statistical DownScaling Model (sdsm) facilitates the rapid development of multiple, low-cost, single-site scenarios of daily surface weather variables under current and future regional climate forcing. Additionally, the software performs ancillary tasks of predictor variable pre-screening, model calibration, basic diagnostic testing, statistical analyses and graphing of climate data. The application of sdsm is demonstrated with respect to the generation of daily temperature and precipitation scenarios for Toronto, Canada by 2040–2069.  2002 Elsevier Science Ltd. All rights reserved.

Hydrological modelling using artificial neural networks

This review considers the application of artificial neural networks (ANNs) to rainfall-runoff modelling and flood forecasting. This is an emerging field of research, characterized by a wide variety of techniques, a diversity of geographical contexts, a general absence of intermodel comparisons, and inconsistent reporting of model skill. This article begins by outlining the basic principles of ANN modelling, common network architectures and training algorithms. The discussion then addresses related themes of the division and preprocessing of data for model calibration/validation; data standardization techniques; and methods of evaluating ANN model performance. A literature survey underlines the need for clear guidance in current modelling practice, as well as the comparison of ANN methods with more conventional statistical models. Accordingly, a template is proposed in order to assist the construction of future ANN rainfall-runoff models. Finally, it is suggested that research might focus on the extraction of hydrological ‘rules’ from ANN weights, and on the development of standard performance measures that penalize unnecessary model complexity.

An artificial neural network approach to rainfall- runoff modelling

Abstract This paper provides a discussion of the development and application of Artificial Neural Networks (ANNs) to flow forecasting in two flood-prone UK catchments using real hydrometric data. Given relatively brief calibration data sets it was possible to construct robust models of 15-min flows with six hour lead times for the Rivers Amber and Mole. Comparisons were made between the performance of the ANN and those of conventional flood forecasting systems. The results obtained for validation forecasts were of comparable quality to those obtained from operational systems for the River Amber. The ability of the ANN to cope with missing data and to “learn” from the event currently being forecast in real time makes it an appealing alternative to conventional lumped or semi-distributed flood forecasting models. However, further research is required to determine the optimum ANN training period for a given catchment, season and hydrological contexts.

HydroTest: A web-based toolbox of evaluation metrics for the standardised assessment of hydrological forecasts

This paper presents details of an open access web site that can be used by hydrologists and other scientists to evaluate time series models. There is at present a general lack of consistency in the way in which hydrological models are assessed that handicaps the comparison of reported studies and hinders the development of superior models. The HydroTest web site provides a wide range of objective metrics and consistent tests of model performance to assess forecasting skill. This resource is designed to promote future transparency and consistency between reported models and includes an open forum that is intended to encourage further discussion and debate on the topic of hydrological performance evaluation metrics. It is envisaged that the provision of such facilities will lead to the creation of superior forecasting metrics and the development of international benchmark time series datasets.

Detection of conceptual model rainfall—runoff processes inside an artificial neural network

Abstract The internal behaviour of an artificial neural network rainfall—runoff model is examined and it is demonstrated that specific architectural features can be interpreted with respect to the quasi-physical dynamics of a parsimonious water balance model. Neural network solutions were developed for daily discharge series simulated by a conceptual rainfall—runoff model given observed daily precipitation totals and evaporation rates for the Test River basin in southern England. Neural outputs associated with each hidden node, produced from the output node after all other hidden nodes had been deleted, were then compared with state variables and internal fluxes of the conceptual model (including soil moisture, percolation, groundwater recharge and baseflow). Correlation analysis suggests that hidden nodes in the neural network correspond to dominant processes within the conceptual model. In particular, different hidden nodes are associated with distinct “quickflow” and “baseflow” components, as well as a threshold state in the soil moisture accounting. The results also demonstrate that, for this river basin, a neural network with seven inputs and three hidden nodes can emulate the gross behaviour of the conceptual model.

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

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.

The effect of different basis functions on a radial basis function network for time series prediction: A comparative study

Abstract Many applications using radial basis function (RBF) networks for time series prediction utilise only one or two basis functions; the most popular being the Gaussian function. This function may not always be appropriate and the purpose of this paper is to demonstrate the variation of test set error between six recognised basis functions. The tests were carried out on the Mackey–Glass chaotic time series, Box–Jenkins furnace data and flood prediction data sets for the Rivers Amber and Mole, UK. Each RBF network was trained using a two-stage approach, utilising the k-means clustering algorithm for the first stage and singular value decomposition for the second stage. For this type of network configuration the results indicate that the choice of basis function (and, where appropriate, basis width parameter) is data set dependent and evaluating all recognised basis functions suitable for RBF networks is advantageous.

Practical proposals for managing uncertainty and risk in project planning

Abstract Standard planning techniques, such as PERT, and the popular software tools that support them are inadequate for projects involving uncertainty in the project direction and task durations. Probability distributions for task durations and generalized activity networks with probabilistic branching and looping have long been established as viable techniques to manage these project uncertainties. Unfortunately, their complexity has meant that their use in industry is minimal. This paper proposes extensions to existing software tools to specify and manage such uncertainties that would be easy to learn and use. A survey has shown that if these extensions were available, commercial and government organizations would regularly use them.

2006 Special issue: Symbiotic adaptive neuro-evolution applied to rainfall-runoff modelling in northern England

This paper uses a symbiotic adaptive neuro-evolutionary algorithm to breed neural network models for the River Ouse catchment. It advances on traditional evolutionary approaches by evolving and optimising individual neurons. Furthermore, it is ideal for experimentation with alternative objective functions. Recent research suggests that sum squared error may not result in the most appropriate models from a hydrological perspective. Models are bred for lead times of 6 and 24 hours and compared with conventional neural network models trained using backpropagation. The algorithm is also modified to use different objective functions in the optimisation process: mean squared error, relative error and the Nash-Sutcliffe coefficient of efficiency. The results show that at longer lead times the evolved neural networks outperform the conventional ones in terms of overall performance. It is also shown that the sum squared error objective function does not result in the best performing model from a hydrological perspective.

Neural network and GA approaches for dwelling fire occurrence prediction

Abstract This paper describes three approaches for the prediction of dwelling fire occurrences in Derbyshire, a region in the United Kingdom. The system has been designed to calculate the number of fire occurrences for each of the 189 wards in the Derbyshire. In terms of the results from statistical analysis, eight factors are initially selected as the inputs of the neural network. Principal Component Analysis (PCA) is employed for pre-processing the input data set to reduce the number of the inputs. The first three principal components of the available data set are chosen as the inputs, the number of the fires as the output. The first approach is a logistic regression model, which has been widely used in the forest fire prediction. The prediction results of the logistic regression model are not acceptable. The second approach uses a feed-forward neural network to model the relationship between the number of fires and the factors that influence fire occurrence. The model of the neural network gives a prediction with an acceptable accuracy for the fires in dwelling areas. Genetic algorithms (GAs) are the third approach discussed in this study. The first three principle components of the available data set are classified into the different groups according to their number of fires. An iterative GA is proposed and applied to extract features for each data group. Once the features for all the groups have been identified the test data set can be easily clustered into one of the groups based on the group features. The number of fires for the group, which the test data belongs to, is the prediction of the fire occurrence for the test data. The three approaches have been compared. Our results indicate that the neural network based and the GA based approaches perform satisfactorily, with MSEs of 2.375 and 2.875, respectively, but the GA approach is much better understood and more transparent.