International Journal of Engineering & Technology, 7 (2.

7) (2018) 219-224

International Journal of Engineering & Technology

Website: www.sciencepubco.com/index.php/IJET

Research Paper

Design and implementation of weather fore casting system

based on cloud computing and data mining techniques
Rajarajeswari. P 1 *, J. Pradeep kumar 2, D. Vasumathi 3
1Associate professor, Department of Computer science and Engineering Aditya college of Engineering, Madanapalle ,Chitoor (Dt)
,Andhra pradesh, India
2 Assistant Professor, Department of Computer science and Engineering Aditya college of Engineering,Madanapalle,Chitoor(Dt),An-
dhra Pradesh, India
3 Professor,Department of Computer science and Engineering Jntu,Hyderabad,Telegana

*Corresponding author E-mail: rajacse77@gmail.com

Abstract

Software engineering is an engineering discipline which is used to analyze, modeling and development of the system. Normally Weather
conditions are changing continuously with respective to atmospheric conditions. In the phase of modeling of the Weather fore casting
system is performed by taking the data which is collected on atmospheric conditions and record temperature, rainfall, humidity and other
parameters. The main objective of this paper is focused on Design and implementation of Weather forecasting system. Weather forecasting
model is prepared by using Cloud computing and Data mining techniques. This System is analyzed by using temperature, rainfall conditions.
Statistical framework is provided on Modeling for surface process of weather forecasting with Data assimilation. This paper presents the
Conceptual level architecture of Weather prediction model by using decision tree. Proposed algorithm is used for the construction of
decision tree. Weather forecasting report is implemented by using Android.

Keywords: Weather Forecasting; Cloud Computing; Data Assimilation; Atmospheric Conditions.

1. Introduction
2. Related work
Weather forecasting is considered as one of the meteorological and
challenging issue in the world. Number of Scientists has been fo- In this section we discuss some important aspects which are related
cusing on characteristics of weather system by using various meth- to modeling of weather forecasting system.
ods. Normally Weather prediction model is performed by using
mathematical equations with the description of atmospheric tem- 2.1. Climate changes
perature, pressure and moisture with respective to time [1]. Weather
conditions are obtained by considering ground, ship, aircraft obser- Climate data gives the official data record that is provided after per-
vations. Modern computers transfer the observations onto surface forming quality control activities. Synoptic data is considered as
and map the lines from meteorologists for correcting the errors. The real-time data provided for doing forecast modeling. So many risks
forecaster analyzes each model and does the predictions on best as- are produced to human society due to Climate changes.
pects of Weather forecasting model. It can be performed for 12 to
24 hours. Here Data mining techniques are used for analyzing the 2.2. Weather forecasting
Weather prediction model. In this section 3 modeling for surface
process of Weather forecasting is presented. Statistical framework It is a vital application in meteorology and facing problems in the
for data assimilation is described in section 4. Cloud computing world. Chen and Dudia have focused on hydrology model with cou-
techniques are applied for Weather forecasting system in section 5. pling process of an advanced land surface. Kalyankar and Alaspur-
Data mining techniques for weather forecasting system is presented kar [2] proposed a data mining techniques to analyze Metrological
in section 6. Proposed algorithm and their results is presented in data and to model the Weather forecasting system by taking the pa-
section 7.Different constraints are given for the construction of de- rameters such as temperature pressure and rainfall.
cision tree is given section 8.Conceptual level of architecture for
Weather forecasting system is given in section 9. Design pattern for 2.3. Data assimilation
Weather forecasting system is provided in section 10. Implementa-
tion of weather forecasting report by using android and Conclusions The main objective of data assimilation provides production of ini-
are given in section 11 and 12. tial conditions for operational forecasts. It also provides the con-
struction of long term reanalyzes’ of atmospheric state. Data assim-
ilation is considered as one type of an analysis method where as
information from the accumulation of observations over a period of
time and converted into model state. There are three components in
Copyright © 2018 Rajarajeswari. P et al. This is an open access article distributed under the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
 220 International Journal of Engineering & Technology

Data-assimilation process. Observations, Background information

about atmosphere state based on model forecast, dynamic con-
straints forecast modeling. Ghil and Rizzoli have introduced in Data
assimilation in meteorology and oceanography. Heintzen and
Charlson are given detailed description about Weather forecasting
System by using Cloud computing techniques [3]. Yadav and
Khatri is proposed weather forecasting model using Data mining
techniques [4]. Olaiya and Adeyemo have focused on different ap-
plications of weather prediction and Climate change studies by us-
ing Data mining Techniques [5]. Hemalatha is focused on the report
of weather forecasting system for the implementation of Data Min-
ing and Global Positioning Systems with the help of ID3 algorithm.

3. Modeling for surface process of Weather Fig. 2: The Process of Intermittent, Data Assimilation.
forecasting
4. Statistical framework for data assimilation
Weather and climate –prediction models are represented by the
movement of heat and water
State vectors ‘p’ may be defined on a model grid. Depending on the
analysis, unknown and known back ground vector are represented
as pa, pb. It can define the values of a single variable in a two-dimen-
sional space, Tb (p, q).The vector length n is the product of the num-
ber of variables and the number of grid points.
State and observation vectors are defined as follows
pt = True model state vector
pa = analysis model state vector
pf = forecast model state vector
pb= background model state vector
q = vector of observations.
Error covariance matrices are defined as follows
B is the covariance matrix of forecast errors. The dimensions of er-
ror matrix are nxn. It controls the function in terms of shape and
Fig. 1: Modeling for Surface Process of Weather Forecasting. magnitude. At an observation point information in the innovation
vector is defined. It can be translated into B into a variable analysis
Within the plant canopy and the ground beneath. Seasonal Variation increment at surrounding grid points to reduce the analysis error.
of Surface heating provides the difference between continents and
oceans which gives monsoon circulations on larger scales. Vertical
 
2
B   a  b (1)
movement of heat at the atmosphere interface occurs through con-
duction process. It consists of a very shallow layer of atmosphere
which is called the laminar sub layer. It is having a depth of few In a multi-dimensional system
molecules to a few millimeters. Generally, forecasting of two or

  a  b 
more week prepared with two variables such as water temperature, T
ice cover to the atmospheric-model simulation. The following fig- B   a  b (2)
ure shows the physical process with the movement of heat and mass
in Ocean [1]. Large-scale weather Philomena in a region into a specific number
of different, dominant weather regimes, classes based on a variable
3.1. Model initialization process such as sea-level pressure. The impact of Climate impacts must be
forecast, understood and dealt with at local and regional levels. This
Initialization of model is performed by taking two requirements. need has motivated many of current and future climate downscaling
One is dependent variables and second one is gridded mass field activities. In current situation, it can be generalized that the atmos-
variables and momentum field variables. Two approaches are re- pheric models are good tools for evaluating the historical changes
quired for doing the initialization. One is called static initialization on climate. [7].
and second one is called dynamic initialization. In case of static in-
itialization, the observations are interpolated to a model grid such
as data analysis. Dynamic initialization involves the reforecast in- 5. Cloud computing techniques for weather
tegration of the model to produce an initial state which is dynami- forecasting model
cally consistent with the equations for the forecast.
Cloud computing is considered as one type of internet -based com-
3.2. Data assimilation puting .It provides shared computer processing resources and data
to computers and other devices on demand. It is a model for sharing
Normally data assimilation is to be obtained by taking the combi- computer resources based on demand. Cloud computing and stor-
nation of information from observations and Numerical methods. It age solutions provide users and enterprises with various capabilities
should act as major role for the improvement of weather forecasts. to store and process their data in either privately owned that may be
Data assimilation provide mathematical framework from the re- located far from the user–ranging in distance from across a city to
sources independently. At the initial time of forecast Data assimila- across the world.
tion and data analysis both are referred to the processes which em-
ploy observations for the construction of gridded datasets which
consists of dependent variables at the initial time of a forecast. The
main objective of data assimilation can be the creation of initial
conditions for operational forecasts [6]
 International Journal of Engineering & Technology 221

Now a day’s weather conditions are obtained by taking ground ob-

servations and observations are taken from ships and aircraft, radi-
osonde, Doppler radar, and satellites. This type of information is
sent to various meteorological centers. In these centers, the data col-
lection analyses on data operations are well performed. It is con-
verted into in the form of variety of charts, maps, and graphs [9].
Modern high-speed computers transfer the many thousands of ob-
servations onto surface and upper-air maps. Computers draw the
lines on the maps with help from meteorologists, who correct for
any errors. A final map is called an analysis. Computers not only
draw the maps but predict how the maps will look sometime in the
future. The forecasting of weather is performed by using computer
is known as numerical weather prediction. Weather prediction is
performed by using numerical analysis. Atmospheric models are
developed by using by using mathematical equations to provide de-
scription of atmospheric temperature, pressure and moisture will
change over time. These equations are converted in the form of
computer programming and data is present with atmospheric con-
Fig. 3: Weather Forecasting Model by Using Cloud Computing Techniques ditions which are fed into the computer. These equations are solved
[5]. with the help of computer and show the equations to determine how
the different atmospheric variables will change over the next few
Cloud computing techniques are suitable for meteorological appli- minutes. This procedure is repeated again and again using the out-
cations. Cloud resources serves as the backbone to countless put from one cycle as the input for the next cycle.
webpage and web driven applications are used by general public Climate change is considered as significant change in Statistical
including data processing and distribution systems that disseminate distributions of weather patterns with respective weather conditions
key weather forecasting, severe weather warning, and climate in- [8]. Here data mining techniques are used for finding relevant in-
formation. We use cloud services to execute now casting algorithm formation from huge quantity of data.
for our Cloud cast application [5]. Cloud computing platforms
lower than cost of computation by leveraging economies of scale. Table 1: Attributes of Meteorological Dataset
It reduces cost and energy and time.

4.1. Implementation

We present the results of live analysis performed with our own ra-
dar located on our campus. Calculate Now casting’s
cost and computation time for one hour of weather data for the sim-
ulation instance types offered by Amazon EC2.We bring up the in-
stances with the Now casting image and start the ingest of weather
data from the radars. Once the cloud based now cast instance re-
ceives the first set of radar scans, it starts to generate 1 to 15
minutes.
We carry out this operation for one hour of weather data and deter-
mine the cost for one hour. Now cast operation using cost tracking
services provided by Amazon EC2.We can do liver measurement
on each of four cloud instance for calculating overall time taken for Forecasting the weather based on Rainfall, Temperature, Air qual-
now casting process. This process is suitable for cloud cast applica- ity for selected cities:
tion on cloud services. Generation of data is performed by using Analysis of numerical data values for Rainfall, Temperature and Air
radar. quality in select metros on 29th March 2017 [11]. These values can
Data is transmitted to a particular instance by the execution of al- be shown above figure.
gorithm within 15 minutes.
Now cast images are generated and are sent to a central web server
which is used by client. Here Amazon EC2 is a cloud service that
provides resizable compute capacity to execute applications on de-
mand. EC2 provides on-demand resources with pricing depending
on the type of resources used and duration of usage.

6. Data mining techniques for weather predic-

tion model

6.1. Data mining method

In the data mining methods the following steps are follows; Data
Collection, data selection, data transformation and Data mining
Graph 1: Analysis of Numerical Data Values for Rainfall, Temperature
methods are used [10]. and Air Quality
a) Data cleaning
In this data model is prepared with consistent format by taking care
of data missing. Cleaned data is more suitable for data mining.
b) Data selection
Data relevant to the analysis and meteorological dataset had 10 at-
tributes.
 222 International Journal of Engineering & Technology

Table 2: Max & Min Temperature Values for Various Cities 6.1. Results and analysis

1600
1400 Dataset size
1200 ID3 alogrithm
1000
800
600
400
200
0
1 2 3 4 5

Fig. 4: Variation in between Proposed Algorithm and ID3 Algorithm.

Table 3: Numerical Dataset Values for Air Quality
The results and analysis about parameters have been shown the fol-
lowing figure.

6.2. Decision tree

Decision tree is constructed with this proposed algorithm. Set of

records are having structure, attribute values. Usually Successful
attributes are taken as true or false or failure, success.

Table 4: Attribute Name-Possible Values

Particular Matter In the Air You are Breathing

Air Quality
Good
8. Constraint s for the construction of decision
Moderate tree
Poor
Weather prediction model is based on with respective weather con-
7. Proposed algorithm ditions. The following values are taken which are present in the
given table.
Constraints for the construction of decision tree using the above
Decision tree is built up with proposed algorithm with a following values.
set of values. Constraint 1: If the climate is Sunny then Stormy is true when
Input: Set of attributes Temperature && Humidity is in the range of 40 0 c to 45 0 c then It
Output: Weather prediction model is in safe state.
A: = {A1……An} is the set of target attributes value. Constraint 2: If the climate is Cloudy then Stormy is False or True
{ when
If D= Temperature && Humidity is in the range of 25 0 c to35 0 c then it
Then it returns anode with unsuccessful attribute value and pro- is in Unsafe state.
duces a decision tree. Constraint 3: If the climate is Rainy then Stormy is True or false
If (D consists of number of records are having the same value for when
the target attribute values) Temperature && Humidity is in the range of 20 0 c to 30 0 c then it
Then it returns a single node with that single value is in Safe state.
If C=
Then it returns a single node with Successful attribute values are Table 5: Numerical Values for Decision Tree Construction
placed in number of records Z.
Let us take the attribute values are ranging from 1……n for ri in C.
Let us take the subsets of T containing the values from 1……….K
for Tj consisting of record with dj of attribute R
It Returns a tree with roots is labeled D and arcs are labeled d1, d2
…dm
Values are (C-{Z}, A, T1), (C-{R}, A, T2)…,(C-{R},C, Sm)
End
}
 International Journal of Engineering & Technology 223

11. Design pattern for weather forecasting

system
Design pattern provides the flow of Weather forecasting Model.
Weather forecasting Model is an Abstract class. Numerical Weather
prediction and Cloud based Architecture for Weather Forecasting
system are subclasses of Weather forecasting system. Concrete
Fig. 5: State Chart Diagram for Climate Changes.
classes are Mathematical Model and Cloud Model. In case of Math-
ematical Model, Mathematical operators are attributes and Deriva-
9. Wind parameters tive function is an operation. In case of Cloud Model data is an at-
tribute and Data is stored by using an Amazon Ec2.
Motion of air is called wind. It is expressed in terms of speed and
direction. Wind is measured in terms of meters. Humidity, Rainfall,
Wind speed, Wind direction, temperature condensation and water
levels all these parameters are measured in terms of various sensors.
Wind speed is low then the turbulence will be high. Higher the tem-
perature and turbulence is also increased. Generation of turbulence
is high by the temperature difference. Normally deviation of wind
is 5 0 to 10 0 degrees.

6
Minutes
5
Fig. 8: Design Pattern for Weather Forecasting System.
4
3 12. Implementation of weather report by
2 using android material
1 Mobile Application models are developed by J2ME and Android .It
0 provides security and applications from uncorrected code. Java is
well suited for an execution model of the system which is based on
1 2 3 4 5 6 7 8 a virtual machine. Application permissions are not possible with
Fig. 6: Direction of Wind Speed with Respective Wind Speed, Wind Angle J2ME. Consequently, read and write files are not running normally
and Minutes. by using J2ME applications and arbitrary places are connected by
open network connections. These are to be possible with Android.
Mobile applications which are running by using J2ME and Android
10. Conceptual level architecture of weather are made from assemblies of Java classes packaged into archives.
prediction model by using decision tree
11.1. Weather forecasting application by using android

Conceptual level provides Formal way of the system. Here weather This process is implemented by considering three main areas. One
prediction model is designed with the conceptual architecture is Toolbar area, second one is weather icon and temperature and
model by using decision tree. This process is easily understood by third one is weather data .Weather forecasting model is developed
users. Here rounded rectangle is used for representing the concepts by using current weather information. In this process the following
[12]. parameters are used. Select various cities such as Bangalore, Chen-
nai, and
Hyderabad. Choose one of the cities and display the temperature in
Celsius. It also gives conditions of Weather.

11.2. Result analysis

Here weather application is considered as one type of real time ap-

plication. Select one of the cities that can be given in first figure.
Here I have taken various cities such as Bangalore, Chennai, and
Hyderabad in the second figure.

Fig. 7: Conceptual Level Architecture of Weather Prediction Model by Us-

ing Decision Tree.

First level: It shows Outlook of Decision tree.

Second level: It shows the Climate, Temperature, Humidity,
Stormy, State parameters Fig.7: Conceptual level architecture of
Weather prediction model for modeling Weather forecasting sys-
tem.
Third level: Climate parameters: Sunny, Cloudy, Rainy, Tempera-
ture parameters: 40 0 c to 45 0 c, Humidityparameters: 40Va-
pour/mm3to 45 Vapour /mm3, Stormy: True or False
 224 International Journal of Engineering & Technology

References
[1] K.Rupakumar,A.K.Sahai,K.Krishnakumar,S.K.Pat-
wardhan.etal”High resolution climate change scenarios for India for
the 21stcentury.Indian Institute of Tropical Meteorol-
ogy,Pune411008,India.Currentscience ,Vol.90,NO.3,10 February
2006.
[2] MeghaliA.Kalyankar,Prof.S.J.Alaspurkar,“DataMining Technique
to analyse the Metrological Data,International Journal of Advanced
Research in Computer science and Software Engineering,Volume 3
Issue 2,February 2013.
[3] Chen.F and J.Dudhia Coupling an advanced land surface –hydrology
model with Penn state-NCAR MM5 modeling System. Part I: Model
implementation and sensitivity.Mon.Wea.Rev, 129, 569-585.
[4] Rohitkumaryadav,Ravi Khatri, “A Weather forecasting model using
the DataMiningTechnique”,International Journal of Computer Ap-
plications Volume 139-N0.14,April 2016.
Fig. 9: A) Model of Weather Application. [5] Fororunso Olaiya Application of Data mining Techniques in
Weather Prediction and Climate Change Studies.”International Jour-
nal of Information Engineering and Electronic Business 2012, 1, 51-
59. DOI:10.5815/ijieeb.2112.01.07.
[6] Ghil,M and P.PMalanotte-Rizzoli.Data assimilation in meteorology
and oceanography.Adv.Geophys.,33,141-266.
[7] Glahn, B. Comments-Reforecast: An important data set for improv-
ing weatherpredictions.Bull.Amer.Meteor.Soc. 89, 1373-1378.
[8] HansonC.E., J.P.Palutikof,M.T.J.Livermore, etal.Modelling the im-
pact of climate extremes: an overview of the MICE project. Climate
change, 81, 163-177.
[9] Heintzenberg, J. andR.J.Charlson (2007).Clouds in the perturbed
Climate System: Their relationship to energy balance, Atomspheric
Dynamics and precipitation Cambridge,USA: MIT press.
[10] MeghaliA.Kalyankar,Prof.S.J.Alaspurkar,“DataMining Technique
to analyse the Metrological Data,International Journal of Advanced
Research in Computer science and Software Engineering,Vol-
Fig. 9: B) Selection of Various Cities. ume3Issue2,February 2013.
[11] ChybickiA,Lubniewski.Z,Bruniecki.K,Markiewicz.L”Numerical
weather prediction-Data fusion to GIS systems and potential appli-
cations,Croatian information technology association.-GIS Fo-
rum,university of Silesia,Zagerb 2011,56-61.
[12] S.L.Belousov,L.V.BerkovichandI.G.Sitnikov“Mathematical model-
ing in Meteorology and Weather forecasting “Mathematical models
of life support system.
[13] P.Rajarajeswari,Dr.A.Ramamohanreddy,Dr.D.Vasumathi “Problem
Solving Process Based on Conceptualmodel with Mathematical
Techniques, “International Journal of Computer & Mathematical
Sciences ISSN 2347 – 8527 Volume 3, Issue 6 August 2014
[14] https://developer.android.com/index.html.
[15] Evans, J.P.21st century climate change in the Middle East. Climate
Change, 92, 417, 432, doi: 10.1007/s10584-008-9438-5.

Fig. 10: Weather Conditions of Selected City.

The above screenshot shows Weather conditions such as tempera-

ture conditions for a particular city and also shown the cloud pa-
rameters.

13. Conclusion
This paper presents the study based on land surface model and pro-
vides heat fluxes with surface boundary conditions for Weather
forecasting system. This paper also provides analysis for modeling
of weather forecasting system with various techniques and Weather
forecasting model by using Cloud computing techniques. The pre-
sent paper mainly focused on the construction of decision tree with
proposed algorithm.