In this master thesis the author will estimate the rainfall based on artificial neural network (ANN) models for monsoon season. The accurate rainfall prediction is one of the greatest challenges in hydrology. Forecast of any natural and usual event call for information regarding its phase of occurrence as well as nature. In the present study, artificial neural network (ANN) with different activation functions has been employed, to estimate daily monsoon rainfall of Pusa, Samastipur, in Bihar, India.
The daily mean temperature, relative humidity, vapour pressure and rainfall data of period (1st June to 30th September) for years 1981-1989, 1992-1994, 1996-2002 and 2004-2008 were used for training and data for years 2009-2013 were used to test the models. The sensitivity analysis was carried out to identify the most significant parameter for daily rainfall prediction. The Neuro solution 5.0 software was used for designing of ANN models based on sigmoid axon and hyperbolic tangent axon activation functions. All the ANN networks were trained and tested with feed forward back propagation algorithm. The performance of the models were evaluated qualitatively by visual observation and quantitatively using different statistical and hydrological indices viz. mean square error, correlation coefficient, akaike’s information criterion, coefficient of efficiency and pooled average relative error.
It was found that the performance of the ANN single hidden layer model based on sigmoid axon activation function is better than the ANN model based on hyperbolic tangent axon activation function. The best ANN models revealed that two days lag time was found to be satisfactory for set of inputs to the model. The sensitivity analysis indicated that the most significant input parameter besides rainfall itself is the vapour pressure in daily rainfall prediction for study area.
Table of Contents
1. INTRODUCTION
2. REVIEW OF LITERATURE
2.1 Artificial Neural Networks (ANNs)
2.2 Sensitivity Analysis
3. MATERIALS AND METHODS
3.1 Description of Study Area
3.1.1 Physiographic description
3.1.2 Soil characteristics
3.1.3 Cropping pattern
3.1.4 Climatology
3.2 Data Acquisition
3.2.1 Pre-analysis and formulation of input and output data
3.3 Development of Models
3.3.1 Artificial neural networks (ANNs)
3.3.1.1 General
3.3.1.2 Description and application of ANNs
3.3.1.3 The biological basis of ANNs
3.3.1.4 Basic concept of artificial neural network
3.3.1.5 Network architecture
3.3.1.6 Neurons and layers
3.3.1.7 Output of the nodes
3.3.1.8 Propagation law
3.3.1.9 The Back-propagation algorithm
3.3.1.10 Activation functions
3.3.1.11 Sigmoid axon function
3.3.1.12 Hyperbolic tangent axon function
3.3.2 Selection of network architecture
3.4 Development of ANN Models
3.4.1 Pattern I
3.4.2 Pattern II
3.4.3 Pattern III
3.4.4 Training and testing of ANN models
3.5 Sensitivity Analysis
3.6 Performance Evaluation Indicators
3.6.1 Statistical indices
3.6.1.1 Mean square error (MSE)
3.6.1.2 Correlation coefficient (CC)
3.6.1.3 Akaike’s information Criterion (AIC)
3.6.2 Hydrological indies
3.6.2.1 Coefficient of efficiency (CE)
3.6.2.2 Pooled average relative error (PARE)
4. RESULTS AND DISCUSSION
4.1 Development of Rainfall Prediction Models
4.1.1 Artificial neural network (ANN) models
4.1.1.1 ANN models with nine input parameters and one output parameter (Pattern I)
4.1.1.2 ANN models with six input parameters and one output parameter (Pattern II)
4.1.1.3 ANN models with three input parameters and one output parameter (Pattern III)
4.2 Performance Assessment of Developed Models
4.2.1 Qualitative evaluation
4.2.2 Quantitative evaluation
4.2.2.1 Statistical indices
4.2.2.2 Hydrological indices
4.3 Sensitivity Analysis
4.3.1 Sensitivity analysis of the artificial neural network (ANN) model
5. SUMMARY AND CONCLUSIONS
Research Objectives & Key Themes
The primary objective of this thesis is to develop and evaluate artificial neural network (ANN) models for the accurate prediction of daily monsoon rainfall in the Pusa region of Bihar, India, using two different activation functions to identify the most effective predictive architecture. The study also aims to determine the most significant meteorological factors influencing rainfall through a rigorous sensitivity analysis.
- Development of ANN models for daily monsoon rainfall estimation.
- Comparative analysis of sigmoid and hyperbolic tangent activation functions.
- Evaluation of model performance using statistical and hydrological indices.
- Identification of significant input parameters for predictive accuracy via sensitivity analysis.
Excerpt from the Book
3.3.1.4 Basic concept of artificial neural network
Artificial neural networks (ANNs) are systems of simple computational units that are able to become accustomed to an information environment. This alteration is appreciated by regulation of the inside network connections through putting on a certain set of rules or simply algorithms. In consequence, ANNs are capable to find out and approximate relationships that are contained in the statistics or data that is offered to the network. This approach is based on the human brain and it is flexible in the range of problems it can solve, and highly adaptive to the newer environments.
Basic proposed definition of an ANN (Hecht-Nielsen, 1990):
A neural network is a parallel, distributed information processing structure consisting of processing elements (which can possess a local memory and can carry out localized information processing operations) interconnected via unidirectional signal channels called branches (‘fans out’) into as many collateral connections as desired; each carries the same signal – the processing element output signal. The processing element output signal can be of any mathematical type desired. The information processing that goes in within each processing element can be defined arbitrarily with the restriction that it must be completely local; that is, it must depend only on the current values of the input signals arriving at the processing element via impinging connections and on values stored in the processing element’s local memory.
Summary of Chapters
1. INTRODUCTION: Covers the importance of climate and rainfall prediction for agriculture and water resource management, highlighting the challenges of modeling complex, non-linear atmospheric processes.
2. REVIEW OF LITERATURE: Reviews previous research on the application of ANN models, time series analysis, and sensitivity analysis in the fields of hydrology and rainfall prediction.
3. MATERIALS AND METHODS: Describes the study area's physiography and climate, data collection processes, the development of ANN models with different input patterns, and the evaluation metrics used.
4. RESULTS AND DISCUSSION: Presents the developed ANN models, evaluates their performance based on statistical and hydrological indices, and provides a detailed sensitivity analysis of meteorological inputs.
5. SUMMARY AND CONCLUSIONS: Summarizes the findings of the study, confirming that ANN models are efficient tools for rainfall forecasting and identifying vapour pressure as a highly sensitive input parameter.
Keywords
Artificial Neural Networks, Monsoon Rainfall, Rainfall Prediction, Sigmoid Axon, Hyperbolic Tangent, Sensitivity Analysis, Hydrological Modelling, Mean Square Error, Correlation Coefficient, Akaike’s Information Criterion, Coefficient of Efficiency, Back Propagation Algorithm, Pusa, Bihar, Meteorological Data.
Frequently Asked Questions
What is the core focus of this research?
The research focuses on utilizing Artificial Neural Network (ANN) models to predict daily monsoon rainfall in Pusa, Bihar, by testing different architectures and input variables.
What are the primary thematic areas covered?
The study covers artificial neural network theory, meteorological data analysis, hydrological modeling, rainfall forecasting, and sensitivity analysis of climatic variables.
What is the ultimate goal of the work?
The primary goal is to determine the best-performing ANN architecture and input pattern for accurate rainfall prediction to assist in flood management and agricultural planning.
Which scientific methodology is employed?
The study uses feed-forward back-propagation ANN models with two specific activation functions (sigmoid axon and hyperbolic tangent) to map non-linear relationships in weather data.
What topics are discussed in the main body?
The main body details the data acquisition, the construction and training of various ANN patterns, performance evaluation using statistical indices, and an in-depth sensitivity analysis.
Which keywords characterize this work?
Key terms include ANN, rainfall estimation, monsoon season, sensitivity analysis, back propagation, and hydrological modeling.
Why are different activation functions compared in this study?
Comparing sigmoid and hyperbolic tangent activation functions allows the author to determine which mathematical approach better captures the non-linear dynamics of monsoon rainfall patterns.
What did the sensitivity analysis reveal about the input variables?
The analysis concluded that vapour pressure is the most significant input parameter for rainfall prediction, followed by relative humidity and mean temperature.
How were the ANN models trained and validated?
The models were trained using historical daily meteorological data from 1981 to 2008 and subsequently verified (tested) using data from 2009 to 2013.
- Citar trabajo
- Bhaskar Pratap Singh (Autor), 2014, Rainfall estimation based on artificial neural network (ANN) models for monsoon season, Múnich, GRIN Verlag, https://www.grin.com/document/469953
