BMS Institute of Technology and Management

Department of MCA

Major Project [18MCA63]

Student Name: Thejashwini N L USN: 1BY19MCA72

Dateof Submission: 03/04/2021

Project Title: Rainfall Detection Using Wavelets

Company Name: VMD Technologies

1. INTRODUCTION

In Today’s era global warming is affecting all over the world which majorly effect on mankind and
cause the expedite the change in climate. Due to this air and oceans are warming, sea level is rising
and flooding and drought etc. One of the serious consequences due to this climate change is on
Rainfall. Rainfall prediction now days is an arduous task which is taking into the consideration of most
of the major world-wide authorities. Rainfall is a climatic factor that affects several human activities
on which they are depended on for ex. agricultural production, construction, power generation and
tourism, among others [1]. This make the rainfall serious concern and requirement of better rainfall
prediction. Rainfall is a complex atmospheric process, and due to the climate changes, it become more
difficult to predict it. Since due to the arbitrary characteristics of rainfall series, they are often labelled
by a stochastic process [2]. Now flood and drought are very common as in Uttarakhand state of India
has confronted worst natural disaster in June 2013. There was received approx. 400% more rainfall
compare to
regular monsoon rainfall. Due to such type of heavy rainfall, roads and bridges was completely
destroyed and 100,000 pilgrims and tourists was trapped which are on their “Char Dham Yaatra”[3]

1
and this disaster could not be predicted by the government, big industries or risk management entitles,
as well as

the scientific community before the incident. These are also may lead to the land slide which is also a
most serious geo-hazard causing the loss of life and property all over the world.

2. PROBLEM STATEMENT

Climate is a important aspect of human life. So, the Prediction should accurate as much as possible. In
this paper we try to deal with the prediction of the rainfall which is also a major aspect of human life
and which provide the major resource of human life which is Fresh Water. Fresh water is always a
crucial resource of human survival – not only for the drinking purposes but also for farming, washing
and many other purposes.

Making a good prediction of climate is always a major task now a day because of the climate
change.Now climate change is the biggest issue all over the world. Peoples are working on to detect
the patterns in climate change as it affects the economy in production to infrastructure. So as in rainfall
also making prediction of rainfall is a challenging task with a good accuracy rate. Making prediction
on rainfall cannot be done by the traditional way, so scientist is using machine learning and deep
learning to find out the pattern for rainfall prediction.

A bad rainfall prediction can affect the agriculture mostly framers as their whole crop is depend on the
rainfall and agriculture is always an important part of every economy. So, making
an accurate prediction of the rainfall somewhat good. There are number of techniques are used of
machine learning but accuracy is always a matter of concern in prediction made in rainfall. There are
number of causes made by rainfall affecting the world ex. Drought, Flood and intense summer heat
etc. And it will also affect water resources around the world. Our major concern is the major downfall
to the rainfall on yearly bases as we can see in Fig 1 the graph show there is the major downfall to the
yearly rainfall in millimetre.

2
2.1 OBJECTIVES

1.To predict when rainfall will occurs because sometimes rainfall will occur in non-rainy season also.
This will yields some problems.
2.To build a robust system for analysis several years of data and to monthly predict rainfall.
3.To design a ML modules which includes wavelets and neural networks to predict rainfall.
4.To obtain the optimal weights(parameters) of neural network structure,LM back propagation
algorithm was used to train the network.
5.To choose the best for model and reject model with underfit or overfit

3. LITERATURE SURVEY

1. Navianti et al. [2] in the journal Application of Fuzzy Inference System on Rainfall
Prediction in North Surabaya conducted a study aimed at predicting rainfall in North
Surabaya by applying basic logic and fuzzy logic rules by applying Fuzzy Inference
System method.
 Information:
This study uses six input variables that affect the occurrence of rain that is in the form of air
temperature, relative humidity, wind speed, air pressure, total cloud layer, and the duration of
solar irradiation. While the output variable in the form of rain prediction. The results of this
study obtained the accuracy of rainfall prediction of 77.68% of eleven experiments and it can
be concluded that the results of rainfall prediction have characteristics similar to the actual data
of rainfall in North Surabaya.
 Drawback:
The advantages of this research is the use of Mamdani method that is more intuitive, covers a
wide field, and according to the process of human information input so that the results obtained
quite accurate, while the lack of use of the data used must be precise.

3
2.Alexandra L’Heureux, et.al (2017) presented new ways of processing Big Data through
Machine Learning Algorithms.

• Information
Due to Big Data characteristics, traditional tools are now not capable of handling its storage, transport
or its efficiency. Machine Learning is regarded as a fundamental component of Data Analytics as it has
power to learn from data and provides data driven insights, prediction and decision. The tremendous
increase in size, space and time complexity of Support Vector Machine (SVM) would affect both the
complexities thus making computational efficiency infeasible.
• Drawback
 Curse of Modularity in which increase in size of data leads to collapse of the given boundary of
algorithm is solved by Map Reduce 

2.Yuniar et al. [3] Conducting research that aims to improve the method of weather forecasting
at Abdulrahman Saleh Airport with neural network backpropagation algorithm.

 Information
The variable input used in this study consists of three inputs namely air temperature, air humidity, air
pressure and two output variables i.e. wind speed and rainfall. The results show the forecast has been
close to the target value. The model has successfully performed the forecast process well with the
MSE value of wind speed forecast of 0.0086 and the MSE forecasted precipitation value of 0.004846.
The advantage of this research is to predict the weather for every day.

 Drawback
While the drawback is to use only one year data earlier as input variables. The 2008 forecast uses 2007
input data, 2009 forecasts using input data for 2008, and so on

4. Research conducted by Haryanto et al [4] aims to predict daily weather in Bengkulu City by
implementing artificial neural network backpropagation method using temperature, humidity
and rainfall data from 2008 to 2013 as input variables.

 Information

4
While the ouput variables used in the study are daily weather categories such as very light rain, light
rain, moderate rain, heavy rain, and very heavy rain .. The tests conducted in this study are: a).
Maximum iteration of 100, 500, 1000, 1500, 10000, and 15000. b). Hidden layers: 1, 2, 3, 4, 5, 6, 7, 8,
9, and 10 c). Maximum error of 0.01 and 0.001. In the three tests that have been done the best results
obtained by using the maximum iteration of 15000, 7 hidden layer, and maximum error 0.001. The
advantages of this study is to compare the daily weather on the observation of Meteorology
Climatology and Geophysics (BMKG) Bengkulu with daily prediction of weather with artificial neural
networks.

• Drawbacks
. While the shortcomings of the research is only using three variables as input, and the used of data is
still only as much as 5 years.

5.Research conducted by Nhita and Adiwijaya [5] A Rainfall Forecasting using Fuzzy System
Based on Genetic Algorithm, Based on the result of experiment, it is concluded that the
combination of GA and Fuzzy System prediction model have accuracy in several different
population sizes and crossover probability more than 90%.

Drawbacks:
This obtained accuracy is highly depending on data which been used as input and classification which
been established due to the vast missing value in the data

3.1 Limitations of Existing system

1. Several numerical weather forecasts have been proposed for weather prediction but most of these
models are limited to short period forecasts.
2. Some algorithms can able to take only one year data and some algorithms take 50 years of data.

4. PROPOSED SYSTEM

In the proposed system we combined wavelets with artificial neural networks. The Wavelet Neural
Network model was formed in which the weights are learned with feed forward neural network with

5
Back Propagation algorithm. The number of hidden neurons BPNN was determined by trial and error
procedure.
 Wavelet Analysis
 Discrete wavelet Transform(DWT)
 Mother Wavelet
 Artificial neural network
 Linear Auto-Regressive(AR) Modelling

1.Wavelet Analysis
The wavelet analysis is an advanced tool in signal processing that has attracted much attention
since its theoretical development (Grossmann and Morlet 1984). Its use has increased rapidly in
communications, image processing and optical engineering applications as an alternative to the Fourier
transform in preserving local, non-periodic and multiscaled phenomena. The difference between
wavelets and Fourier transforms is that wavelets can provide the exact locality of any changes in the
dynamical patterns of the sequence, whereas the Fourier transforms concentrate mainly on their
frequency. Moreover, Fourier transform assume infinite length signals, whereas wavelet transforms
can be applied to any kind and any size of time series, even when these sequences are not
homogeneously sampled in time (Antonios and Constantine 2003). In general, wavelet transforms can
be used to explore, denoise and smoothen time series which aid in forecasting and other empirical
analysis.

2.Artificial Neural Network

An ANN, can be defined as a system or mathematical model consisting of many nonlinear artificial
neurons running in parallel, which can be generated, as one or multiple layered. Although the concept
of artificial neurons was first introduced by McCulloch and Pitts, the major applications of ANN’s
have arisen only since the development of the back-propagation method of training by Rumelhart
(Rumelhart et al. 1986). Following this development, ANN research has resulted in the successful
solution of some complicated problems not easily solved by traditional modeling methods when the
quality/quantity of data is very limited. ANN models are ‘black box’ models with particular properties,
which are greatly

suited to dynamic nonlinear system modeling. The main advantage of this approach over traditional
methods is that it does not require the complex nature of the underlying process under consideration to
be explicitly described in mathematical form. ANN applications in hydrology vary, from real time to
event based modeling.

6
3.Linear Auto Regressive Model
A common approach for modeling univariate time series is the autoregressive (AR) model: Xt ¼ δ þ
ϕ1Xt−1 þ ϕ2X2−1 þ ⋯⋯ þ ϕ2X2−1 þ At ð4Þ where Xt is the time series, At is white noise, and δ ¼
1− Xp i−1 ϕi !μ ð5Þ where ‘μ’ is the mean of the time series. An autoregressive model is simply a
linear regression of the current value of the series against one or more prior values. AR models can be
analyzed with linear least squares technique. They also have a straightforward interpretation. The
determination of the model order can be estimated by examining the plots of Auto Correlation
Function (ACF) and Partial Auto Correlation Function (PACF). The number of non-zero terms (i.e.
outside confidence bands) in PACF suggests the order of the AR model. An AR (k) model will be
implied by a sample PACF with k non-zero terms, and the terms in the sample ACF will decay slowly
towards zero. From ACF and PACF analysis for rainfall, the order of the AR model is selected as 1.

4.1 Advantages of Proposed System

1.By predicting rainfall,it will help formers to make plan for agriculture and we can also control
floods.
2.We can also plan for storing Rain water for other purposes.
1.Several time series models have been proposed for modelling monthly rainfall series and annual
rainfall series such as AR and the disaggregation multivariate model examine the poosibility of long
term prediction forecast by the application of ANNs.
2.Wavelets have proven to be a powerful tool for the analysis and synthesis of data from long memory
process.
3. A larger dataset is used to improve the accuracy of the results. More robust machine learning
algorithms are used to make better predictions.

5. Modules Identified

1.Wavelet Analysis : The basic aim of wavelet analysis is to determine the frequency content of a
signal and then it assess and determine the temporal variations of this frequency content.
2.Artificial Neural Network : It is a mathematical model consisting of many non-linear artificial
neurons running in parallel, which can be generated, as one or multiple layer.
7
3.Dataset Collection : Details of rainfall of 10 years will collected to form a Dataset.
4.Model Training: here we used Levenberg-marquardt method(LM) is used for training of given
network.

6.TOOLS OR LANGUAGES USED/ HARDWARE DETAILS

 Hardware Requirements

Processor - Intel dual core

Speed - 1.1Ghz

RAM - 1 GB

Hard Disk - 20GB

Key Board - Standard Windows keyboard

Mouse - Two or Three Button Mouse

Monitor - SVGA

 Software Requirements

Operating System - Windows XP/7

Application Server - Tomcat 5.0/6.x/8.x

Front End - Python Tkinter

Database Connectivity - MySQL

Recommendations: Accepted(Y/N)
Suggestions by the Guide:

8
Signature of the Guide Signature of Project Coordinator

9
10