The authors have declared that no competing interests exist.
This research paper focuses on rainfall variations in Tamil Nadu, India using Wavelet, Linear regression and Artificial Neural Networks model from 2004 to 2017. As the rainfall is the key factor in understanding climate change, the seasonal datasets from 2004-2017 of Tamil Nadu state has been taken for study. The salient feature of this study is the application of Neural Networks and wavelet analysis. It reveals that the rainfall variations are ambiguous that it does not maintain a constant pattern. Wavelet coefficients of multiresolution spectrogram reveals that the intensity of rainfall in each year. Linear regression model divulge the pattern of rainfall followed in every season and the results show that except winter season all other season suffers deficient rainfall. The deficiency of rainfall may be due to different parameters like ElNino or LaNina pattern or global warming. Results showed that all seasons except winter does not maintain consistency in the rainfall variability. Winter season provides the positive slope values of 4.7 and 0.6 for January and February respectively. Moreover Artificial Neural Networks training provides prominent results of Regression value 0.98 which is comparably high with other seasons taken for study.
Rainfall a natural occurrence plays major role in determining groundwater level, and in particular helps agricultural sector. Rainfall is considered to be the fundamental sector and it has major contribution in the development of the country. India receives most of the rainfall on monsoon so it is very essential for understanding the rainfall pattern and its trend. Rainfall pattern was analysed by various researchers of India. Rainfall pattern of Krishna-Godavary river basin was analysed by Jaagannadha sarma, 2005
For past few years there has been a continuous increase in the temperature of Tamil Nadu which led to increase in global warming and reduction in the rainfall days. Tamil Nadu receives rainfall two times a year –
According to Tamil Nadu State Action Plan for Climatic Change (TNSAPCC), the trend in the past decade has shown the southwest monsoon has been decreasing from 48% to 24%, and the northeast monsoon has been increasing from 34% to 63%. Thus average rainfall in Tamil Nadu may not vary but the distribution changes. Due to the climatic change there has been increase in the intensity of carbon dioxide, soil erosion, rise in sea levels which led to cyclones, drastic increase in pests and diseases
Weather- which is always changing- is comprised of the following elements:
i)
ii)
iii)
iv)
Weather forecasting is the application of science and technology to predict the conditions of the atmosphere for a given location and time. Human beings have attempted to predict the weather informally for millennia and formally since the 19th century. Weather forecasts are made by collecting quantitative data about the current state of the atmosphere at a given place and using scientific understanding of atmospheric processes to project how the atmosphere will change.
The state of Tamil Nadu is situated in the southern part of the country between north latitude between 8° 5' and 13° 35' east longitude between 76° 15' and 80° 20' 1 . The total area of the state is 1,30,058 sq km making it the eleventh largest state in the country (
Tamil Nadu can be divided broadly into two natural divisions (a) the coastal plains and (b) the hilly western areas. The average temperature in the hilly areas varies between a minimum of 21.27o C to a maximum of 35.86o C. The average temperature in the plains varies between a minimum of 10.46o C to 24.86o C.
Out of 15 agro-climatic zones of India as per Planning Commission of India, Tamil Nadu comes under two zones, namely Southern Plateau and Hills region and East Coast and Hills region . Under National Agricultural Research Project of ICAR, with in the broad classification of Planning Commission’s 15 agro-climatic zones, Tamil Nadu has been divided in to seven agro climatic sub-zones and those are
North western zone
North eastern zone,
Western zone,
Cauvery Delta zone,
Southern zone,
High rainfall zone and
High Altitude and
Hilly zone
Wavelet transform plays major role in analysing time series data efficiently. Event related potential determination can be easily done through this transformation more clearly. The spatial and temporal information will provide the real time data strategy in eagle’s eye. Advantages over the conventional time-frequency analysis lies in decompose the signal of any time-frequency multiresolution functions
ψ (t) ϵ L2 (R) satisfies certain admissibility conditions as
ψ(t) is called wavelet; (ω) is fourier transform of
ψ(t). The Continuous Wavelet Transform (CWT) is defined as the sum over all time of the signal multiplied by scaled, shifted versions of the wavelet function,
CWT uses discrete sampling of data in order to obtain the finer details depends on the scaling parameter of the wavelet. Finer resolution will obtain from increased computational time and more memory required calculating the coefficients
Neural networks are simplified imitations of the central nervous system, and obviously therefore, have been motivated by the kind of computing performed by the human brain. The structural constituents of a human brain termed
A human brain develops with time and this, in common parlance is known as
|
|
|
---|---|---|
1. | Number of Layers | 3 (Input, Hidden and Output) |
2. | Algorithm | Levenberg-Marquardt Optmization |
3. | Learning | Supervised |
4. | Data Preprocessing | Normalization |
5. | Data Partition | 70% (Training), 15% (Testing), 15% validation |
6. | Momentum constant | 0.7 |
7. | Learning Rate | 0.9 |
8. | Activation Function (Hidden unit) | tansig |
9. | Activation Function (Output unit) | linear |
10. | Number of neurons in hidden layer | 25 |
11. | Number of epochs | 1000 |
During the month of March, rainfall decreases at the rate of -1.8202 and Aril and May month it increases slope increases to certain value of 1.8202 and 1.6497 respectively (
Linear regression for the month of June, August reveals the increasing trend with the slope value of 1.16 and 0.96308 whereas during the month of July and September it shows the steep decline of rainfall with a value of -2.6073 and -2.89 respectively (
For the month of October, November and December i.e., North East (NE) monsoon season, the state receives deficient rainfall (
Apart from the effect of monsoon, the rainfall variation may occur due to ElNino/La Nina effect. (
S.No. | Season | RMSE | NRMSE | R Value | Computational Time (sec) |
1. | Winter | 2.376 | 0.1978 | 0.985 | 7.150 |
2. | Pre Monsoon | 3.5788 | 0.2184 | 0.694 | 2.246 |
3. | Monsoon | 2.8127 | 0.1166 | 0.9405 | 8.035 |
4. | Post Monsoon | 8.4395 | 0.2307 | 0.7987 | 2.190 |
The investigation of rainfall data for all the seasons in Tamil Nadu in India had revealed that the trends of annual and seasonal rainfall. Temporal variations are analyzed using linear regression method and Artificial Neural Network for the year 2004 - 2017. This examination demonstrated significant changes in seasonal rainfall variations in Tamil Nadu during the year 2004-2017. The rainfall trend shows that there is a slow decrease of rainfall in the pre monsoon season. Wavelet trend shows the intensity of rainfall is completely distributed and it is not consistent. The distribution of rainfall over the years is completely random and it cannot be predicted or forecasted easily. Neural Networks attempts to estimate and the error percentage are very less in training the winter season. Thus winter maintains some consistency in rainfall and other error percent of neural networks are very high when compared with other seasons. Artificial neural network methods are used to predict the rainfall values for the year 2004-2017 and the predicted values are compared with the actual value of rainfall. The errors between the actual and predicted value are plotted. Such predictions are done with the help of MATLAB R2012a. The variation of temperature and rainfall and the prediction of rainfall would make a helpful guide for possible crop planning and also used for irrigation and agriculture.
We would like to express our sincere gratitude to the Management of Loyola College, Chennai and SDNB Vaishnav College, Chennai for providing permission to publish this article.