A novel method of visualizing and understanding the internal functional behaviour of an artificial neural network (ANN) river flow model is presented. The method hypothesizes that an ANN is able to map a function similar to the flow duration curve while modelling the river flow. A mathematical analysis of the hypothesis is presented, and a case study of an ANN river flow model confirms its significance. The proposed approach is also useful within other models that improve the performance of an ANN. The reasons why these models improve a raw ANN can be clearly understood using this approach. While the field of ANN knowledge-extraction is one that continues to attract considerable interest, it is anticipated that the current approach will initiate further research and make ANNs more useful to the hydrologic community. © 2004 John Wiley and Sons, Ltd.

K. P. Sudheer

Department Of Civil Engineering

Flow of water

flow visualization

Hydrology

Knowledge based systems

Neural Networks

Knowledge Extraction

ROVER FLOW MODELS

Rivers

Artificial neural network

Hydrological modeling

mathematical analysis

Rainfall-runoff modeling

river flow

IIT Madras is a public technical and research university located in Chennai, Tamil Nadu. Founded in 1959, it is recognised as an Institute of National Importance.

IIT Madras has been ranked as the top engineering institute in India for four years in a row by the National Institutional Ranking Framework of the MHRD

It currently offers undergraduate, postgraduate and research degrees across 16 disciplines in Engineering, Sciences, Humanities and Management. About 596 faculty belonging to science and engineering departments and centres of the Institute are engaged in teaching, research and industrial consultancy.

IIT Madras

Hydrological Processes

Over the past 15 years, artificial neural networks (ANNs) have been used increasingly for prediction and forecasting in water resources and environmental engineering. However, despite this high level of research activity, methods for developing ANN models are not yet well established. In this paper, the steps in the development of ANN models are outlined and taxonomies of approaches are introduced for each of these steps. In order to obtain a snapshot of current practice, ANN development methods are assessed based on these taxonomies for 210 journal papers that were published from 1999 to 2007 and focus on the prediction of water resource variables in river systems. The results obtained indicate that the vast majority of studies focus on flow prediction, with very few applications to water quality. Methods used for determining model inputs, appropriate data subsets and the best model structure are generally obtained in an ad-hoc fashion and require further attention. Although multilayer perceptrons are still the most popular model architecture, other model architectures are also used extensively. In relation to model calibration, gradient based methods are used almost exclusively. In conclusion, despite a significant amount of research activity on the use of ANNs for prediction and forecasting of water resources variables in river systems, little of this is focused on methodological issues. Consequently, there is still a need for the development of robust ANN model development approaches. © 2010 Elsevier Ltd. All rights reserved.

Environmental Modelling and Software

Methods used for the development of neural networks for the prediction of water resource variables in river systems: Current status and future directions

The paper describes some of the recent developments and applications for knowledge extraction from trained artificial neural network based hydrological models. The methods are in general derived from the casual relationship between inputs and outputs, and analysis of the hidden neuron behaviour. The discussions are focused on works related to explaining the internal behaviour and embedded physical process identification in a trained ANN rainfall-runoff model. It is envisaged that knowledge extraction is important for artificial neural networks to gain wider degree of acceptance especially in the context of hydrologic modeling. Therefore, this domain has to become a major field of research since it validates the use of neural networks for applications where reasons or explanations on why or how a result has been achieved are important. © 2009 Taylor & Francis Group, LLC.

ISH Journal of Hydraulic Engineering

Recent advances in knowledge extraction from neural network based hydrologic models

Estimation of evapotranspiration (ET) requires a knowledge of the values of many climatic variables, some of which require special equipment and careful observations. Although ET is an important component of water balance, the data required for its accurate estimation are commonly available only at widely spaced measurement stations. The major objective of this study was to estimate ET using an artificial neural network (ANN) technique and to examine if a trained neural network with limited input variables can estimate ET efficiently. The results indicate that even with limited climatic variables an ANN can estimate ET accurately. The paper also outlines a procedure to evaluate the effects of input variables on the output variable using the weight connections of ANN models. Such an analysis performed on the ANN-ET models developed was able to explain the reasons for the ANN's potential in estimating the ET effectively from limited climatic data. Copyright © 2008 John Wiley & Sons, Ltd.

Models for estimating evapotranspiration using artificial neural networks, and their physical interpretation

This paper presents a novel hybrid linear-neural (LN) model formulation to effectively model rainfall-runoff processes. The central idea of the proposed model framework is that the hidden layer of an artificial neural network (ANN) model be designed with a combination of linear and nonlinear neurons. A training algorithm for the proposed model is designed based on minimum description length criteria. The advantage of the algorithm is that the final architecture of the LN model is arrived at during the training process, thus avoiding selection from a class of models. The proposed model has been developed and evaluated for its performance for forecasting the river flow of Kolar basin, in India. The values of three performance evaluation criteria, namely, the coefficient of efficiency, the root-mean-square error, and the coefficient of correlation, were found to be very good and consistent for flows forecasted 1 hour in advance by the LN model. The value of the relative error in peak flow prediction was within reasonable limits for the model. The forecasts by the LN model at higher lead times (up to 6 hours) are found to be good. A relative evaluation of LN model performance with that of an ANN model and of a multiple linear regression model indicates that the LN model effectively combines the strength of the other two, implying that the LN model seems to be well suited to exploit the information to model the nonlinear dynamics of the rainfall-runoff process. Copyright 2006 by the American Geophysical Union.

Fulltext

Water Resources Research

A hybrid linear-neural model for river flow forecasting

Artificial neural networks (ANNs), due to their excellent capabilities for modeling complex processes, have been successfully applied to a variety of problems in hydrology. However, one of the major criticisms of ANNs is that they are just black-box models, since a satisfactory explanation of their behavior has not been offered. They, in particular, do not explain easily how the inputs are related to the output, and also whether the selected inputs have any significant relationship with an output. In this paper, a perturbation analysis for determining the order of influence of the elements in the input vector on the output vector is discussed. The approach is illustrated though a case study of a river flow model developed for the Narmada Basin, India. The analyses of the results suggest that each variable in the input vector (flow values at different antecedent time steps) influences the shape of the hydrograph in different ways. However, the magnitude of the influence cannot be clearly quantified by this approach. Further it adds that the selection of input vector based on linear measures between the variables of interest, which is commonly employed, may still include certain spurious elements that only increase the model complexity. Journal of Hydrologic Engineering © ASCE.

Journal of Hydrologic Engineering

Knowledge extraction from trained neural network river flow models

The emergence of artificial neural network (ANN) technology has provided many promising results in the field of hydrology and water resources simulation. However, one of the major criticisms of ANN hydrologic models is that they do not consider/explain the underlying physical processes in a watershed, resulting in them being labelled as black-box models. This paper discusses a research study conducted in order to examine whether or not the physical processes in a watershed are inherent in a trained ANN rainfall-runoff model. The investigation is based on analysing definite statistical measures of strength of relationship between the disintegrated hidden neuron responses of an ANN model and its input variables, as well as various deterministic components of a conceptual rainfall-runoff model. The approach is illustrated by presenting a case study for the Kentucky River watershed. The results suggest that the distributed structure of the ANN is able to capture certain physical behaviour of the rainfall-runoff process. The results demonstrate that the hidden neurons in the ANN rainfall-runoff model approximate various components of the hydrologic system, such as infiltration, base flow, and delayed and quick surface flow, etc., and represent the rising limb and different portions of the falling limb of a flow hydrograph. © 2004 John Wiley and Sons, Ltd.

Identification of physical processes inherent in artificial neural network rainfall runoff models

Journal of Hydrology

Integrated approach to model decomposed flow hydrograph using artificial neural network and conceptual techniques

Journal of Computing in Civil Engineering

Data Division for Developing Neural Networks Applied to Geotechnical Engineering

Development of effective and efficient rainfall-runoff models using integration of deterministic, real-coded genetic algorithms and artificial neural network techniques

Hydrological Sciences Journal

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

Explaining the internal behaviour of artificial neural network river flow models

Journal	Hydrological Processes
ISSN	08856087
Open Access	No