ONLINE MONITORING OF CLIMATIC PARAMETERS:A STATISTICAL STUDY ABOUT ENVIRONMENTAL CHANGES IN QATAR Tahir Mahmood*, Saddam Akber Abbasi** & Muhammad Riaz* [*] Department of Mathematics and Statistics, King Fahd University of Petroleum and Minerals,Dhahran, Saudi Arabia. [**] Department of Mathematics, Statistics and Physics, Qatar University, Doha, Qatar. Corresponding Email:
[email protected] Poster No: QULSS2016-025
ACF of Temperature
0.5 −0.5
−0.5 −1.0
50
60
0
10
20
30
40
50
60
Forecasting of Temperature
35 30
30
25
25
20
20
15
15
1 2 3
1960
1970
1980
1990
2000
2010
2010
time
2014
2016
2018
0.05 0.00 −0.10 −0.05
0.4
2010
2012
0.10
1.0 0.8
2000
0.2
1990
0.0
1980
2008
PACF of the Residuals
0.6
−1 1970
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
50
60
Figure 1: Time series analysis for Temperature
0.0
0.1
0.2
0.2 0.4 0.6 0.8 1.0
60 40
0 2 4 6 8
0
−0.2
−0.1
20
data
PACF of Rainfall 0.3
ACF of Rainfall
Decomposition of Rainfall
0
20
30
40
50
60
0
10
30
40
10 15 20 25 30 35
60 50 40
0
5
20 10 0
20
Forecasting of Rainfall
30
10 8 6 4
1960
1970
50
2
trend
10
Rainfall
1950
1980
1990
2000
2010
2008
2010
2012
2014
2016
2018
PACF of the Residuals
2000
0.05
0.4
2010
−0.05
1990
0.2
1980
0.0
1970
0.00
0.6
10
0.8
30
1.0
ACF of the Residuals
−10 1960
Materials and Methods
time
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Figure 2: Time series analysis for Rainfall ACF of CO2
ACF of the Residuals
1.0
−0.5
30
−0.2
0.0
40
0.0
0.2
50
0.4
60
0.5
0.6
70
0.8
80
1.0
90
CO2 emmision from fossils
0
10
20
30
40
1970
1980
PACF of CO2
1990
2000
2010
0
5
10
15
20
25
30
35
30
35
PACF of the Residuals
−0.2
40
0.0
−0.1
50
0.2
0.0
60
0.4
0.1
70
0.6
0.2
80
0.8
0.3
90
Forecasting of CO2
−0.3
30
−0.2
where, φ (x) = 1 − φ1x − φ2x2 − ... − φpxp;Φ (x) = 1 − Φ1xs − Φ2x2s − ... − Φpxps
40
35
−5 −10 28.5 27.5 26.5
1960
1950
By using time series techniques, this study aims to analyze the time series model for environmental parameters (Temperature, Rainfall and CO2) of Qatar and to forecast the effect of these parameters in future. The present study depends on the monthly data sets of Temperature (c) and Rainfall (mm) over the period 1951 to 2012 which is extracted from the web page of World Bank 1. The annual data CO2 emissions (metric tons/capita) between 1970 and 2008 used in stated study has been fetched from the web page of CDIAC 2. The statistical schemes used to meet the objectives of this study are given below. Seasonal ARIMA Models: The Box-Jenkins approach for modeling and forecasting has the advantage in analyze the seasonal time series data (for more detail see 3;4;5;6 ). In this case where the seasonal components are included, the model is called as seasonal ARIMA model or SARIMA model. The model can be abbreviated as SARIM A (p, d, q) (P, D, Q)S where the lowercase for nonseasonal part meanwhile the uppercase for seasonal part. The generalized form of SARIMA model can be written as: φ (x) Φ (x) Yt = θ (x) Θ (x) et
30
5 0
seasonal trend
1950
remainder
1. To monitor the behavior of environmental parameters (i.e. Temperature, Rainfall and CO2) and forecast the environmental effects on Qatar. 2. Diagnostic analysis on the residuals of selected models through Shewhart and EWMA control charts.
20
ACF of the Residuals
−4
Main Objectives
10
Temperature
1950
seasonal
The world is facing several environmental issues such as global warming, loss of biodiversity, deforestation, ocean acidification, ozone layer depletion, acid rain and pollution. These challenging environmental problems result into slight or drastic effects on the ecosystem. The source of continuous variations in the climate are greenhouse gases which are causing the rise in temperature all over the world. Specifically, the Gulf region facing the significant effects by the change of climate such as less and more unpredictable rainfall, average increased in temperature, rise in sea level, dryness, lack of drinking water and regular drought. Qatar is the richest state in the Gulf region whose economic growth depends on petroleum and natural gas industries that directly relates with environmental development. In this study, we have focused on the monitoring of climatic parameters such as Temperature, Rainfall and CO2 emission in Qatar through time series analysis.
0.0
0.5 0.0
35 25 15
data
0
−3
Introduction
PACF of Temperature
1.0
Decomposition of Temperature
remainder
Abstract Twentieth century has witnessed unprecedented changes in the climate whose profound effects are also observed on ecosystem and human life. The source of these changes are presumed to be increasing concentration of greenhouse gases which result into rise in temperature worldwide. Unwanted effects have also been observed in the Gulf region in terms of reduced but intensive and unpredictable rainfall, average increase in temperature, sea level rise, lack of drinking water and regular drought. Qatar, being a richest country whose economic growth depends on petroleum and natural gas industry, is paying focus on its environmental development programs, which is also a goal of recent national vision. In this study, we have focused on monitoring of temperature, CO2 emission and rainfall pattern in Qatar through different control charting schemes, i.e., memory less (Shewhart) and memory type (EWMA and CUSUM) control charting structures; while time series analysis was performed for the period of 1950-2012. It has been observed that temperature and CO2 emission have increasing trend while rainfall depicts decreasing trend in last decades. Furthermore, forecasting of average weather is made by memory type structures which may serve as principle tool in environmental development initiatives.
θ (x) = 1 − θ1x − θ2x2 − ... − θq xq ; Θ (x) = 1 − Θ1xs − Θ2x2s − ... − Θq xqs
10
20
30
40
1970
1980
1990
2000
2010
0
10
15
20
25
For the Shewhart chart, flase alarm is fixed say α = 0.002976 while for EWMA chart, L is the control charting constant and reported as 2.64, 3.15 and 2.5 for the Temperature, Rainfall and CO2 respectively at fixed ARL0 = 168. Shewhart chart for CO2 Residuals
Shewhart chart for Rainfall Residuals
UCL 4
40
where L is the control charting constant and decision criteria is defined as: If the plotting statistic (Zi) exceeds from either side (U CLi or LCLi) then the process is said to be out of control.
Results The appropriate models are selected through the use of different measures such as AIC, AICc, SBC, Durbin Watson (DW) test, Autoregressive Conditionally Heteroscedastic (ARCH), significant (sig.) parameters and Ljung-Box Q-test for each environmental variable which are reported in Table 1.
0
20 10
−2
0
0
−4
−1
200
300
400
500
600
700
LCL
−6 −8
−20
LCL
−10
LCL
−2 −3
100
0
200
300
400
500
600
700
0
10
20
30
40
EWMA chart for CO2 Residuals
EWMA chart for Rainfall Residuals
4 −6
UCL
UCL LCL
−2
0
0
0.0
2
5
0.5
UCL
6
10
1.0
EWMA chart for Temperature Residuals
100
−4
0
100
200
300
400
500
600
700
0
100
200
300
400
500
600
700
LCL
−5
LCL
−0.5
where i is the sample number and λ is a constant such that 0 < λ≤1. the quantity Z0 is the starting value and it is taken equal to the target mean or the average of initial data in case when the information on the target mean µ0 is not available. The control limits for the EWMA statistic are given as r r λ λ LCLi = µO − LσO ( 1 − 1 − λ)2i ; U CLi = µO + LσO ( 1 − 1 − λ)2i , 2−λ 2−λ
0
−1.0
If the plotting statistic (θ) exceeds from either side (UCL or LCL) then the process is said to be out of control. EWMA Control Charts: EWMA, a new control charting scheme termed as exponentially weighted moving average (EWMA) control chart 8. EWMA is a memory type structure that utilizes both current and previous information. Such charts are very useful for the detection of small or persistent shifts in the process parameters (location or scale). An EWMA control chart for monitoring the location of a process is based on the statistic Zi = λθi + (1 − λ) Zi−1,
1
30
2
2
UCL
UCL
6
Shewhart chart for Temperature Residuals
3
LCL = θα/2 ; U CL = θ1−α/2,
5
Figure 3: Estimation and the forecasted results of selected model for CO2
50
Shewhart Control Charts: Shewhart control chart 7 is used to detect the large or transient shifts in the process parameters such as location or scale. Shewhart charts are memory less charts due to its structure that only depends on current observations related to the quality of interest (θ). The structure of the Shewhart chart consist of two decision lines upper control limit (UCL) and lower control limit (LCL) which are defined on specified false alarm rate say α.
0
0
10
20
30
40
Figure 4: Investigation of the residual through control charting schemes
Conclusions The time series and statistical process control analysis of environmental variables reveal the following: • The Temperature and CO2 emission exhibit an increasing pattern over the time referring to special cause such as green house gases and the burning of fossil fuels (coal, oil and natural gas). • The Rainfall shows an irregular pattern over the next decade, that may be resulted due to the deforestation and global warming.
Table 1: Selected models for each environmental parameter
Variable
Model
AICc
DW Test ARCH Sig. Ljung-Box Q-test
Temperature SARIM A(1, 0, 2)(0, 1, 1)12 1814.59 2.002649 Rainfall SARIM A(1, 0, 0)(0, 1, 1)12 4613.2 1.999438 CO2 ARIM A(1, 0, 0) 281.27 2.039606
No No No
Yes Yes Yes
White Noise White Noise White Noise
References [1] The World Bank. Climate change knowledge portal for development practitioners and policy makers. url: http://sdwebx.worldbank.org/climateportal/index.cfm?page=downscaled data downloadmenu=historical: Online; accessed 26-November-2016. [2] Oak Ridge National Laboratory Tennessee United States Carbon Dioxide Information Analysis Center, Environmental Sciences Division. http://databank.worldbank.org/data/reports.aspx?source=2series=EN.ATM.CO2E.PCcountry=QAT: Online; accessed 26-November-2016.
Co2 emissions (metric tons per capita).
url:
[3] Maximilian Auffhammer and Richard T Carson. Forecasting the path of china’s co 2 emissions using province-level information. Journal of Environmental Economics and Management, 55(3):229–247, 2008. [4] Ying Zhang, Peng Bi, and Janet E Hiller. Weather and the transmission of bacillary dysentery in jinan, northern china: a time-series analysis. Public health reports, pages 61–66, 2008. [5] Osabuohien-Irabor Osarumwense. Applicability of box jenkins sarima model in rainfall forecasting: A case study of port harcourt south south nigeria. Canadian journal in Computing Mathematics, Natural Sciences, Engineering and Medicine, 4(1):1–4, 2013. [6] Hsiao-Tien Pao and Chung-Ming Tsai. Modeling and forecasting the co 2 emissions, energy consumption, and economic growth in brazil. Energy, 36(5):2450–2458, 2011. [7] Abdaljbbar Dawod, Muhammad Riaz, and Saddam Akber Abbasi. On model selection for autocorrelated processes in statistical process control. Quality and Reliability Engineering International, 2016. [8] Mu’azu Ramat Abujiya, Saddam Akber Abbasi, and Muhammad Riaz. A new ewma control chart for monitoring poisson observations. Quality and Reliability Engineering International, 2016.
