Journal of Physics: Conference Series

PAPER • OPEN ACCESS

The Weather Forecast Using Data Mining Research Based on Cloud

Computing.
To cite this article: ZhanJie Wang and A. B. M. Mazharul Mujib 2017 J. Phys.: Conf. Ser. 910 012020

View the article online for updates and enhancements.

This content was downloaded from IP address 107.173.189.180 on 02/11/2017 at 01:39

CTCE2017 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
910 (2017) 012020 doi:10.1088/1742-6596/910/1/012020

The Weather Forecast Using Data Mining Research Based on

Cloud Computing.

ZhanJie Wang and A. B. M. Mazharul Mujib

Electronics Information & Electrical Engineering, Dalian University of Technology,
Liaoning, China
Email: [email protected], [email protected]

Abstract. Weather forecasting has been an important application in meteorology and one of
the most scientifically and technologically challenging problem around the world. In my study,
we have analyzed the use of data mining techniques in forecasting weather. This paper
proposes a modern method to develop a service oriented architecture for the weather
information systems which forecast weather using these data mining techniques. This can be
carried out by using Artificial Neural Network and Decision tree Algorithms and
meteorological data collected in Specific time. Algorithm has presented the best results to
generate classification rules for the mean weather variables. The results showed that these data
mining techniques can be enough for weather forecasting.

1. Introduction
Weather Prediction is the application of science and technology to predict atmospheric conditions ahead
of time for a particular region. Prediction is one of the basic goals of Data Mining. Data Mining is to
dig out knowledge and rules, which are hidden and unknown. User may be interested in or has potential
value for decision-making from the large amounts of data. Such potential knowledge and rules can
reveal the laws between the data. There are many kinds of technical methods of data mining, which
mainly include: association rule mining algorithm, decision tree classification algorithm, clustering
algorithm and time series mining algorithm, etc. [1]. How to store, manage and use these massive
meteorological data, discover and understand the law and knowledge of the data, to contribute to
weather forecasting completely and effectively has attracted more and more Data Mining researcher’s
attention[2]. This article constructs the Weather Forecasting platform, using data mining for
meteorological forecast and the forecast results are analyzed.

2. Weather Forecasting
Weather forecasting plays a significant role in meteorology [3]. Weather forecasting has remained a
formidable challenge because of its data intensive and frenzied nature. Generally, two methods are used
to forecast weather: a) the empirical approach and b) the dynamical approach. The first approach is
based on the occurrence of analogues and often referred to as analogue forecasting. This approach is
useful in predicting local scale weather if recorded cases are plentiful. The second case is based upon
equations and forward simulations of the atmosphere and often referred to as computer modeling. Most
weather prediction systems use a combination of both of these techniques.

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
CTCE2017 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
910 (2017) 012020 doi:10.1088/1742-6596/910/1/012020

Figure 1 FAAS Models

This framework as a service (FAAS) has selected seven common forecasting methods. These are
Regression (R), Logistic Regression, Time Series, Artificial Neural Network, Random Forest, Support
Vector Machine and Multivariate Adaptive Regression Splines (MARS). For instance, Regression may
encounter them collinearity among variables. Logistic Regression could only deal with the dataset
where the dependent variable is nominal.

3. Weather Prediction Architecture

Artificial Neural Networks (ANN) and Decision Trees (DT) were used to analyze meteorological data,
gathered in-order to develop classification rules for the Application of Data Mining Techniques in
Weather Prediction. Artificial Neural Networks (ANN) has received special attentions among different
forecasting methods in recent years [4, 5]. Main reason for the popularity of ANN is its capability of
supervised learning from complex relations using non-linear functions [6]. This algorithm combines
with both of the time series and regressional approaches. Weather parameters [7] over the study period
use available historical data for the prediction of future weather conditions. The targets for the
prediction are those weather changes that affect our daily life e.g. changes in minimum and maximum
temperature, rainfall, evaporation and wind speed. These techniques are often more powerful, flexible,
and efficient for exploratory analysis than the statistical techniques. The most commonly used
techniques in data mining are: artificial neural networks, logistic regression, discriminant analysis and
decision trees. By this model, temperature (T), rainfall (R) and wind (W) speed can easily be predicted.
Now Prediction method that only single parameter for example some researchers [8], [9] used wind
speed and other researchers used wind Power for prediction [10], [11]. We have provided more accurate
prediction model data and it has more parameters and efficiency. (Figure.2)

Figure 2 Weather Data Prediction

There are three basic elements of a neuron model. Figure.3 shows the basic elements of neuron
model with the help of a perceptron model, which are, (i) a set of synapses, connecting links, each of
which is considered by a weight/strength of its own (ii) an adder, for summing the input signals,
weighted by respective neuron’s synapses (iii) an activation function, for limiting the amplitude of
neuron’s output. A typical input-output relation can be expressed as shown in Equation 1.

2
CTCE2017 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
910 (2017) 012020 doi:10.1088/1742-6596/910/1/012020

Figure 3 Model of a perceptron

n

n e tj   W ijX i  b j
j 1

O i  f i ( n e ti )
(1)
Where = inputs to node in input, = weight between input node and hidden node, b – bias at node, net =
adder, f = activation function.
The type of transfer/activation function affects the size of the steps taken in weight and space [12].
ANN’s architecture needs determination of number of the connecting weights and the way in which the
information flows through this network is carried out via the number of layers, nodes number in each
layer, and their connectivity. The output nodes numbers are fixed, according to the estimated quantities.
The input nodes numbers are dependent on the existing problem under consideration, and the modeler’s
choice to utilize knowledge of domain. The neurons in the hidden layer are enhanced gradually, and the
network performance in the form an error is examined.

4. Weather Forcast In Cloud Computing

Before Cloud computing has improved the efficiency of data storage, delivery, and dissemination
across multiple platforms and applications, allowing easier collaboration and data sharing, including
data processing and distribution systems that disseminate key weather forecasting, severe weather
warning, and climate information. Data mining techniques and forecasting applications are very much
needed in the cloud computing paradigm. In this study, data mining in Cloud Computing allows
weather forecasting and data storage, with assurance of efficient, reliable and secure services for their
users. The implementation of data mining techniques through Cloud computing will allow the users to
retrieve meaningful information from virtually integrated data warehouse that will reduce the costs of
infrastructure and storage.

5. Implement And Process

5.1. Forecasting Model

Our prediction model is based on a Multi-Layer Perceptron ANN. The central, innovative point of the
method is the construction of the data set and the choice of input data unlike the other works [13][14].
Along with the historical weather data of the site, chosen for the prediction, nearby sites are also taken
into account for the evolution of the weather fronts. Therefore, for the identification of the site where to
make the wind forecast (called WFS), I have chosen four cardinal directions and the four intermediate
directions, two points for each direction, one more far away (1000 km) and second one halfway (500
km) from the place identified. For sites at maximum distance, weather data referred to the instant t-2,
for sites at minimum distance, weather data refer to the instant t-1. Because, weather data was referred
to instant t0 for the WFS. The implemented MLP, has one hidden layer using transfer function, F(x),

the .

3
CTCE2017 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
910 (2017) 012020 doi:10.1088/1742-6596/910/1/012020

The MLP network is trained through the back-propagation learning algorithm. This training
algorithm is to minimize the cost function (E) of the network, expressed as mean square error amongst
all the outputs of the network and the respective desired values defined in the training set. The mean
square error is calculated by using the following expression:

5.2. Input Data

The data used in this study for each point are temperature, rainfall and wind. The data used are the daily
average values. The input Dataset variable are (T), (R) and (W) used in this Model and this model is
similar to Fig.2. We have collected the data of Dalian weather forecasting and input the data in the
model for computing and got the results. Xi (i=1, 2, 3) used are site-dependent. The prediction of
weather forecasting, with a resolution of one day, requires the knowledge of the weather conditions in
previous time instants to the one chosen for the prediction. Earlier, Mesoscale Meteorological Modeling
[15], Numerical Weather Prediction Models [16] and other methods [17] were used for prediction.
Unlike these traditional methods, we have improved this model, using different time instants for
weather forecasting.

5.3. Simulation Results

In this section we have introduce a case study; the test site is a real weather forecast located in Dalian
Liaoning Province, China. In this study of data analysis for Dalian weather forecasting comes from the
Dalian Meteorological Bureau in the last five years. In our research, we have set up a new model with
more parameters and finished the simulation later. Our provided model has shown more accurate results
than the old model. In this prediction model, we have used ANN and FAAS. The figures of merit used
for analyzing the performance. The forecasting data are compared with the real data recorded in the site.
In this work, we have used last five years data analysis, the variables including in the highest
temperature, the lowest temperature and wind speed. Variables of analysis are described in following
chart:

Figure 4 Weather Forecasting Temperature for Last Five Years (2011-2015) Highest and Lowest
Value

Figure 5 Results of wind speed prediction Value in Last Five Years. (2011-2015)

4
CTCE2017 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
910 (2017) 012020 doi:10.1088/1742-6596/910/1/012020

The results have shown that concerning Wind Speed and Temperature highest and lowest Value
displayed the simulation results (figure 4 and 5) in Intervals of last five years. While using more input
data values, the execution time can be as longer as expected.

6. Vi. Discussion and Conclusions

In this Study Forecasting Models algorithm are used for classifying weather parameters such as
maximum temperature, minimum temperature, and wind speed in terms of the month and years. The
implementation of data mining approach to solve the wind forecasting problems for wind farm
production, in particular, for predicting wind speed. The data mining prediction algorithm-ARIMA time
series prediction algorithm is also integrated into the system. The platform has the ability of mass
storage of meteorological data, efficient query and analysis, weather forecasting and other functions. In
this study we also adapted the method of Artificial Neural Networks, it can detect the relationships
between the input variables and generate outputs based on the observed patterns inherent in the data
without any need for programming or developing complex equations to model these relationships. An
artificial neural network (multi-layer perceptron) was applied and several simulations have been
conducted for comparison purposes. ANN’s can detect the relationships between weather parameters
and use these for future prediction. Weather conditions are important to climatic change studies because
the variation in weather conditions in term of temperature and wind speed can be studied using these
data mining techniques. ANNs are implemented, in order to compare their effectiveness in changing the
network topology and the training mode. The results obtained from real data are based on time series of
meteorological data provided by the Dalian Meteorological Bureau. The test cases pointed out that the
proposed approach gives a very interesting performance of the implemented network and shows good
performance in term of MSE. For future perspective, there is still significant potential for improvement
in weather forecasting by using ANN model, through introducing climate change and global warming
variables, in order to forecast more realistic weather parameters.

7. Acknowlegment
This research was supported by the National Natural Science Foundation of China (Grant No.
61672126) and (Grant No. 51579041), We would like to thank the anonymous reviewers for their
attentive reading and for their constructive comments that have helped to further strengthen this paper.

8. References
[1] M. Zhang, “Application of Data Mining Technology in Digital Library”, Journal of Computers,
vol. 6, no. 4, (2011) April, pp. 761-768.
[2] Z. Danping and D. Jin, “The Data Mining of the Human Resources Data Warehouse in University
Based on Association Rule”, Journal of Computers, vol. 6, no. 1, (2011) January, pp. 139-146.
[3] Introduction to Data Mining and Knowledge Discovery, Third Edition, Two Crowds Corporation,
http://www.twocrows.com/introdm.pdf, accessed on 12 April 2009.
[4] L. M. Saini and M. K. Soni, “Artificial neural network-based peak load forecasting using
conjugate gradient methods,” IEEE Transactions on Power Systems, vol. 12, no. 3, pp. 907–
912, . 2002.
[5] S. Fan, C. X. Mao, and L. N. Chen, “Peak load forecasting using the self-organizing map,” in
Advances in Neural Network-ISNN 2005. New York: Springer-Verlag, 2005, pt. III, pp. 640–
649.
[6] Kourentzes, N., “Intermittent demand forecasts with neural networks”, International
Journal of Production Economics, Volume 143, Number 1, pages 198-206, 2013.
[7] Elia G. P., 2009, “A Decision Tree for Weather Prediction”, Universitatea Petrol-Gaze din Ploiesti,
Bd. Bucuresti 39, Ploiesti, Catedra de Informatică, Vol. LXI, No. 1.
[8] A. R. Finamore; V. Calderaro; V. Galdi; A. Piccolo; G. Conio; S. Grasso, “A day-ahead wind
speed forecasting using data-mining model– a feed-forward NN algorithm", IEEE International
Conference on Renewable Energy Research and Applications, 2015, pp. 1230-1235.
[9] E. Erdem, J. Shi, “ARMA based approaches for forecasting the tuple of wind speed and direction”,
Applied Energy 88, ELSEVIER, 2011, pp. 1405–1414.

5
CTCE2017 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
910 (2017) 012020 doi:10.1088/1742-6596/910/1/012020

[10] N. Chen, Z. Qian, I. T. Nabney, and X. Meng, “Wind Power Forecasts Using Gaussian Processes
and Numerical Weather Prediction,” IEEE Transaction on Power Systems, vol. 29, no. 2, 2014.
[11] EWEA Wind in power: European Statistics. 2013.
[12] Martin T. H., Howard B. D, Mark B., 2002, Neural Network Design, Shanghai: Thomson Asia
PTE LTD and China Machine Press.
[13] N. Chen, Z. Qian, I. T. Nabney, and X. Meng, “Wind Power Forecasts Using Gaussian Processes
and Numerical Weather Prediction,” IEEE Transaction on Power Systems, vol. 29, no. 2, 2014.
[14] R. R. B. de Aquino, H. T. V. Gouveia, M. M. S. Lira, A. A. Ferreira, O.N. Neto, M. A. Carvalho
Jr., “Wind Forecasting and Wind Power Generation: Looking for the Best Model Based on
Artificial Intelligence”, IEEE World Congress on Computational Intelligence, 2012.
[15] Roger A. Pielke Sr., “Mesoscale Meteorological Modelling”, International Geophysics Series,
Volume 98, oct, 2013.
[16] David J. Stensrud ,“Parameterization schemes – Keys To Understanding Numerical Weather
Prediction Models”, Cambridge University Press, 2007.
[17] Renewables – Global Status Report- REN21, 2015.