Rainfall - Runoff Modelling Using Artificial Neural Networks ( ANNs )

نویسنده

  • Aman Mohammad Kalteh
چکیده

Over the last decades or so, artificial neural networks (ANNs) have become one of the most promising tools for modelling hydrological processes such as rainfall-runoff processes. In most studies, ANNs have been demonstrated to show superior result compared to the traditional modelling approaches. They are able to map underlying relationships between input and output data without detailed knowledge of the processes under investigation, by finding an optimum set of network parameters through the learning or training process. This thesis considers two types of ANNs, namely, self-organizing map (SOM) and feed-forward multilayer perceptron (MLP). The thesis starts with the issue of understanding of a trained ANN model by using neural interpretation diagram (NID), Garson’s algorithm and a randomization approach. Then the applicability of the SOM algorithm within water resources applications is reviewed and compared to the well-known feed-forward MLP. Moreover, the thesis deals with the problem of missing values in the context of a monthly precipitation database. This part deals with the problem of missing values by using SOM and feed-forward MLP models along with inclusion of regionalization properties obtained from the SOM. The problem of filling in of missing data in a daily precipitation-runoff database is also considered. This study deals with the filling in of missing values using SOM and feed-forward MLP along with multivariate nearest neighbour (MNN), regularized expectation-maximization algorithm (REGEM) and multiple imputation (MI). Finally, once a complete database was obtained, SOM and feedforward MLP models were developed in order to forecast one-month ahead runoff. Some issues such as the applicability of the SOM algorithm for modularization and the effect of the number of modules in modelling performance were investigated. It was found that it is indeed possible to make an ANN reveal some information about the mechanisms governing rainfall-runoff processes. The literature review showed that SOMs are becoming increasingly popular but that there are hardly any reviews of SOM applications. In the case of imputation of missing values in the monthly precipitation, the results indicated the importance of the inclusion of regionalization properties of SOM prior to the application of SOM and feed-forward MLP models. In the case of gap-filling of the daily precipitation-runoff database, the results showed that most of the methods yield similar results. However, the SOM and MNN tended to give the most robust results. REGEM and MI hold the assumption of multivariate normality, which does not seem to fit the data at hand. The feed-forward MLP is sensitive to the location of missing values in the database and did not perform very well. Based on the one-month ahead forecasting, it was found that although the idea of modularization based on SOM is highly persuasive, the results indicated a need for more principled procedures to modularize the processes. Moreover, the modelling results indicated that a supervised SOM model can be considered as a viable alternative approach to the well-known feed-forward MLP model.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Rainfall-runoff modelling using artificial neural networks (ANNs): modelling and understanding

In recent years, artificial neural networks (ANNs) have become one of the most promising tools in order to model complex hydrological processes such as the rainfall-runoff process. In many studies, ANNs have demonstrated superior results compared to alternative methods. ANNs are able to map underlying relationship between input and output data without prior understanding of the process under in...

متن کامل

Monthly runoff forecasting by means of artificial neural networks (ANNs)

Over the last decade or so, artificial neural networks (ANNs) have become one of the most promising tools formodelling hydrological processes such as rainfall runoff processes. However, the employment of a single model doesnot seem to be an appropriate approach for modelling such a complex, nonlinear, and discontinuous process thatvaries in space and time. For this reason, this study aims at de...

متن کامل

Flood Forecasting Using Artificial Neural Networks in Black-Box and Conceptual Rainfall-Runoff Modelling

The paper presents a comparison of lumped runoff modelling approaches, aimed at the realtime forecasting of flood events, based on or integrating Artificial Neural Networks (ANNs). ANNs are used in two ways: (a) as black-box type runoff simulation models or (b) for the real-time improvement of the discharge forecasts issued by a conceptual-type rainfall-runoff model. As far as the coupling of A...

متن کامل

Optimising training data for ANNs with Genetic Algorithms

Artificial Neural Networks (ANNs) have proved to be good modelling tools in hydrology for rainfall-runoff modelling and hydraulic flow modelling. Representative datasets are necessary for the training phase in which the ANN learns the model’s input-output relations. Good and representative training data is not always available. In this publication Genetic Algorithms (GA) are used to optimise tr...

متن کامل

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 A...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2018