ArticlePDF Available

Stratospheric Ozone in the perspectives of Exploratory Data Analysis for Atmospheric Region

Authors:

Abstract and Figures

In order to study physical behavior of Stratospheric ozone in Pakistan’s atmospheric regions (PARs), the Exploratory Data Analysis (EDA) is performed. Using this analysis Probabilistic and mean deviation models are developed to study the ozone layer depletion for Pakistan Atmospheric Regions. These models provided a comprehensive description of the underlying process. The information attained from these forecasts by analyzing these models, can be further employed to vary possible parameters and variables in the physical system to achieve an optimal performance. Such an approach is well explained within the likelihood of computational analysis. The models presented in this paper along with their physical interpretations are very useful for public and private sector organizations.
Content may be subject to copyright.
Journal of Basic and Applied Sciences Vol. 6, No. 1, 45-49, 2010 ISSN: 1814-8085
STRATOSPHERIC OZONE IN THE PERSPECTIVES OF EXPLORATORY DATA
ANALYSIS FOR PAKISTAN ATMOSPHERIC REGIONS
M. Ayub Khan Yousuf Zai,1 M.R.Kamal Ansari2 and Jawaid Quamar3, M. Arif Husain4 and Jawaid Iqbal5
1Department of Applied Physics University of Karachi, Karachi, 75270, Pakistan
and Institute of Space and Planetary Astrophysics, University of Karachi
2 Mathematical Sciences Research Center, Federal University of Arts, Science and Technology
3-5 Institute of Space and Planetary Astrophysics, University of Karachi
ABSTRACT
In order to study physical behavior of Stratospheric ozone in Pakistan’s atmospheric regions (PARs), the Exploratory
Data Analysis (EDA) is performed. Using this analysis Probabilistic and mean deviation models are developed to study
the ozone layer depletion for Pakistan Atmospheric Regions. These models provided a comprehensive description of the
underlying process. The information attained from these forecasts by analyzing these models, can be further employed to
vary possible parameters and variables in the physical system to achieve an optimal performance. Such an approach is
well explained within the likelihood of computational analysis. The models presented in this paper along with their
physical interpretations are very useful for public and private sector organizations.
Keywords: Atmospheric phenomenon, Ozone layer Depletion, stratospheric region, exploratory data analysis,
coefficient of variations.
INTRODUCTION
Atmospheric variables such as precipitation, temperature,
humidity, concentration of ozone, CFCs, CO, CO2, UV
radiations etc. exhibit both temporal and spatial
fluctuations. On the other hand, an overall substantial
mode of identity is present between observations of these
quantities, taken either at the same location like PARs
separated by short temporal sections; or at the same time
at different locations separated by small distances
(Wayne, 1991, Garcia 1994; and Yousufzai, 2001) In
order to depict such variables in a realistic manner, it is
essential to consider probabilistic approaches that allow
for both variability and dependence. Such approaches are
useful for characterizing atmospheric processes such as
OLD in terms of few comprehensive parameters. Also it
is necessary to make decision about atmospheric data sets.
We present some probabilistic approaches that found
application in atmospheric sciences (in the ozone
concentration reduction). To demonstrate the statistical
inferences about these dependent atmospheric
observations, the probability theory for stochastic
processes will be considered. The quantitative study of
our data of ozone depths for PAR we perform EDA
(which is now being incorporated into formal statistical
theory). It will provide a comprehensive characterization
of the underlying system in the form of mathematical
models. The information attained from such analyses can
be further employed to vary possible parameters and
variables in the system to achieve an optimal
performance. Such a statistical approach is well within the
feasibility and approach of available computational
software/packages that rely on graphical techniques such
as histograms, probability plots, P-P plots, Q-Q plots, and
Kolmogorove-Smirnov probability plots.
Evaluation of descriptive parameters
(a) Regarding ozone volume data (1960-1998) for
Pakistan’s air space and the corresponding histogram we
refer to (Fig.1). The illustration signifies the probability of
each value of the depth of ozone in terms of its calculated
volume.
Number of observatin
(5.7,0)
(5.9,2)
(6.1,8)
(6.3,17)
(6.5,27)
(6.7,23)
(6.9,16)
(7.1,14)
(7.3,3) (7.5,2) (7.7,1) (7.9,0)
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
<= 5.8
(5.8,6.]
(6,6.2]
(6.2,6.4]
(6.4,6.6]
(6.6,6.8]
(6.8,7.]
(7,7.2]
(7.2,7.4]
(7.4,7.6]
(7.6,7.8]
> 7.8
Fig. 1. Histogram of various number of observations of
volume of ozone contents
The area under the horizontal line segment at a particular
depth represents the probability of each value of the
depth. (Fig. 2) depicts the gaussian distribution function
for ozone depths for the above said period.
Corresponding author: email: ayubzai@yahoo.com
J. basic appl. sci. 46
Observed Value
Expected Normal Value
-3.5
-2.5
-1.5
-0.5
0.5
1.5
2.5
3.5
180 220 260 300 340 380 420
Fig. 2. Gaussian distribution function for ozone layer
depth at Pakistan’s atmospheric region
(b) For normal data, the sample mean and variance are the
unbiased estimators of location of the underlying
distribution. Most physical data sets are not normally
distributed even after transformation, because the
assumption of an underlying normal distribution is a
mathematical idealization that is never met exactly in
practice because large data sets inevitably contain
outliers.
(c) Robust estimates for location can be obtained by
making the arithmetic mean less sensitive to outliers. In
fact it is to trim a proportion of the data from each end of
the ordered data set and then to compute the mean of the
remaining values (Pandit and WU 1983; Yousuf Zai and
Quamar 1998; Yousuf Zai and Quamar 2001; Garcia,
1994). These trimmed means form a family of estimators
that can be indexed by α, the fraction of the observations
trimmed at each end of the sample. Estimates that are not
unduly influenced by a small number of outliers are called
resistant.
The trimmed mean
α
X is of the form
)21( 1
α
α
=n
X
]})[1({ )()1(
1
2)( knk
kn
ki ixxnkx +
+=
+++
α
,
][
2
1)()1( knk xxX + += , (1)
Here k is the largest integer n
α
and 0<
α
< 0.5.
The ideas underlying trimmed means can also be applied
to the estimation of the variance. In the estimation of
higher order moments (such as the variance) one needs to
protect the estimates from outliers.
(i) 1970-1994 (N = 296)
Trimmed mean = 287.47 DU
Trimmed variance = 332.43 DU
(ii) 1960-1999 (N = 480)
Trimmed mean = 283.33 DU
Trimmed variance = 381.46 DU
A robust measure of scale, is the inter-quartile range
which is given by IQ = upper quartile-lower quartile
It is useful because of its simplicity of the computation
and low efficiencies. Using the parametric values given in
Table1, for ozone layer depths, the median = 283.5 DU,
and lower and upper quartiles are 272.000 and 294.5000
respectively.
Table 1. Calculation of descriptive parameters of ozone
layer from 1960-1998 N = 468.
Serial # Num. Values
Mean 284.03
Conf.Int(-95%) 282.25
Conf.Int(+95%) 285.81
Median 283.50
Max. 200.00
Min 385.00
Variance 384.94
Std.Dev. 19.62
Std.Error 0.91
Lower Quart 272.00
Upper Quart 294.00
so that, Upper MAD = Upper quartile median = 294.5
283.5 = 11.0
Similarly, Lower Mad = Median lower quartile =
283.50 272.00 = 11.5.
The Inter-quartile range IQ = Upper quartile lower
quartile = 294.5 272.0 = 22.5,
(d) Median absolute deviation (MAD) is a scale estimate
that is the counterpart of the median as the most resistant
estimator for location and to make it a biase-free estimate
of the standard deviation it is to be divided by a
normalizing constant cMAD. In the normal case the
asymptotic value of cMAD for n tending to infinity is 0.67
so that we can define the scale estimate
MAD
MAD c
MAD
s=.
For the symmetric case MAD can be computed as MAD =
IQ / 2 = 11.5
SMAD = 11.5 / 0.67 = 17.05
Analysing data in this way is one of the major
applications of robust/resistant methods (Bartoszynski
Yousuf Zai et al. 47
and Buggi, 1996). Computational details of our analysis
are depicted in Table 2.
Table 2. Computational analysis of descriptive measures
(parameters) of ozone layer data (1970-1997) N = 296
Serial # Num. Values
Mean 288.36
Conf.Int(-95%) 290.46
Conf.Int(+95%) 286.00
Median 283.51
Max. 385.00
Min 245.00
Variance 384.94
Std.Dev. 018.40
Std.Error 001.00
Lower Quartile 277.00
Upper Quartile 298.00
(e) The graphical techniques such as described so far
provide a reasonably good idea about the shape of the
distribution of the data under investigation, but do not
determine how well a data set conforms to a given
theoretical distribution i.e. how much the data are
significantly different from a given theoretical
distribution. To diagnose that whether a particular data set
seems to have come from a specific probability
distribution goodness-of-fit-test such as chi square,
Shapiro-Wilk and Kolmogorov-Smirnov (KS) tests are
performed. Most of the cases deal with the goodness-of-
fit-test that are used to assess how well a data set appears
to come from a normal or lognormal distribution, because
classical parametric hypothesis tests assume the data to
come from a normal distribution. Normality, however is
not as important as assumption of independent
observations and equal variance (Millard, 1996; Michaels,
1994).
We will avoid using small sample sizes (n 35) because
most goodness-of-fit tests are not able to detect deviations
from the hypothesized distribution unless the deviations
are quite extreme. Our data set is large enough (n = 480,
468, 296, and for that each test will reject the null
hypothesis, since the “real” data are never distributed
according to any theoretical distribution. For this reason
we have used KS goodness-of-fit test ([]) to show that the
population distribution of ozone depth data is normal(
Fig.3). This test uses the maximum absolute difference D
between the Fo (the cumulated observed frequency) and Fe
(the cumulated expected frequency for each class that has
been created) so that
D = maximum Fo - Fe (3)
Table 3. No.observations = 100, # catagories =
19,Kolomogorove-Smirnov D = 0.0581770,Chi-square=
5.215153, df = 5, p = .3902121
# Observations Cumulative %
expected (Fe) Cumulative %
Observed (F0)
210 0,015 0.000
220 0.089 1.000
230 0.431 1.000
240 1.664 1.000
250 5.152 2.000
260 12.884 9.000
270 26.321 24.000
280 44.628 48.000
290** 64.182 70.000
300 80.558 84.000
310 91.311 93.000
320 96.845 97.000
330 99.078 98.000
340 99.785 99.000
350 99.960 99.000
360 99.994 99.000
370 99.999 99.000
380 99.999 100.00
Infinity 100.00 100.00
Table 3. Depicts differences using (3) for our ozone layer
depletion data with maximum difference about 0.058
whereas from the KS table with level of significance
α
=
0.05 for the sample size 100 (n > 35), D = 100
36.1 = 0.136.
This value is greater than the value of D obtained by (3)
so that we accept the null hypothesis that the observed
distribution does not differ significantly from the
theoretical (normal) distribution. We continue in a similar
way to accept Ho and assume that ozone layer depletion
can be simulated by sampling from a normal distribution
with a mean 282.71DU and standard deviation of 20.06
DU.
(f) Coefficient of Variation (CV) is one of the measures
that describe the scatter of the distribution relative to the
size of the estimated mean and the diversity of the data
from normality. It also indicates stability and consistency
of the data and is given by
CV= Mea
n
Deviation Standard (4)
CVs for computed descriptive parameters given in tables
1 and 2 are 0.069, 0.064 The sufficiently low value of the
calculated CVs (just about 7.0 %) indicates a good degree
of normality.
A probability graph shows at a glance how the
distributions differ from each other (Bartoszynski, 1996;
Hussain and Quamar, 2000). There are two types of
J. basic appl. sci. 48
probability plots, P-P plots and Q-Q plots. These plots can
be best explained as cumulative distribution functions. P-
P plots and Q-Q plots for our data are depicted in figures
2, 3 and 4 The Q-Q plot displayed in (Fig. 5). assesses
that whether a set of data appears to come from a
particular probability distribution or not. We can also use
two data sets that come from the same parent distribution
(with the same shape but not necessarily the same
location and scale).
Theoretical Quantile
Observed Value
.01 .05 .1 .25 .5 .75 .9 .95 .99
150
200
250
300
350
400
450
-4-3-2-101234
Fig. 3. Q-Q representation for ozone layer at Pakistan’s
atmosphere.
Theoretical cumulative distribution
Empirical cumulative distribution
0.00
0.25
0.50
0.75
1.00
0.00 0.25 0.50 0.75 1.00
Fig. 4. The p-p plot of ozone layer depth for Pakistan
Atmospheric region.
Expected
Category (upper limits)
No of obs
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 390
Fig. 5. Kolmogorov-Smirnov test to assess how well the
data set apperas to come from a normal distribution (100
observations), d = 0.058, Chi square = 5.21, df = 5, p =
0.39, Maximum difference = % observed % expected =
70.00 64.2, d = 5.8% at 290 DU or 0.058 at 290 DU.
HYPOTHESIS TESTING
Now we test the hypothesis that
µ
is greater than 284.03
DU or is less than 284.03 DU. This means that we are
concerned with the variability of the mean in both
directions over the entire span of ozone concentration. It
should be noted that the characteristics of this span do not
change.
For this purpose we have chosen a small sample of 36
observations (1960-1962), a large sample of 144
observations (1987-1998) and another large sample of
152 observations following a normal distribution.
Our calculations maintaining 95% confidence level are as
follows:
(a) For Sample size n = 36 with µ = 278.71
1. Let Ho : µ = 278.71 H1 : µ 278.71
2. The test statistic is t
3. Significance level α = 0.05 t = ± 2.042 critical region
= ± 2.5% degrees of freedom (d f) = 35
4. Decision rule for the acceptance or rejection of H0
reject H0 if our observed t < 2.042 reject H0 if our
observed t > + 2.042
5. Test statistic t on the assumption that H0 is true t=
0.94 sample mean = 278.71 sample standard deviation
S1= 33.63
6. Our observed t value of 0.94 is fairly low but it does
not lie in the critical region so we cannot reject Ho: µ =
278.71.
(b) For large sample size n = 144 with µ = 279.94
1. State the hypothesis to be tested and its alternative Ho:
µ = 279.94 H1 : µ < 279.94
2. State the test statistic t to be used and its probability
distribution.
The test statistic t has a standard normal distribution
3. Choose the value of the significance level α and
determine the critical regions Significant level α = 0.05,
(p) = 0.95.
As we now have a two-tail test our critical region of 5 %
is split between the tails. Looking up 0.025 as the α / 2 in
the table we find that this corresponds to a Z value of
1.96. That is the limiting of the critical region are Z ±
1.96. This is recalled as confidence intervals.
4. State the decision rule for the acceptance or rejection of
H0 reject H0 if Z < 1.96 or Z > 1.96 α = 0.05 (p = .95)
accept H0 if 1.96 Z 1.96
5.Calculate the value of the test statistic t on the
assumption that H0 is true test statistic Z = 2.75
6. Make the decision to accept or reject H0. Our observed
Z value of 2.75 is fairly low and less than 1.96 and
Yousuf Zai et al. 49
so falls in the critical zone. On the basis of this sample we
must reject the null hypothesis H0. It will support the
alternate hypothesis H1. From this sample it seems that
mean depth of ozone layer in the Pakistan’s air space is
less than 279.94 DU.
(c) For sample size n = 152 with µ = 284.64
1. State the hypothesis to be tested and its alternative Ho :
µ = 284.64, H1 : µ < 284.64
2. State the test statistic to be used and its probability
distribution
The test statistic has a standard normal distribution
3.Choose the value of the significance level α and
determine the critical regions level α = 0.05, (P) = 0.95.
As we now have a two-tail test our critical region of 5 %
is split between the tails. Looking up 0.025 as the α / 2 in
the table we find that this corresponds to a Z-value of
1.96. That is the limiting of the critical region are Z ±
1.96. This is recalled as confidence intervals.
4. State the decision rule for the acceptance or rejection of
H0 reject H0 if Z < 1.96 or Z > 1.96 α = 0.05 (p = .95)
accept H0 if 1.96 Z 1.96
5. Calculate the value of the test statistic t on the
assumption that H0 is true test statistic = Z = 0.46
6.Make the decision to accept or reject H0. Our observed
Z value of 0.46 is fairly low and less than 1.96 and so
falls in the critical zone. On the basis of this sample we
must reject the null hypothesis H0. It will support the
alternate hypothesis H1. From this sample it seems that
mean depth of ozone layer in the Pakistan’s air space is
less than 284.64 DU.
CONCLUSION
In this communication we have described physical
behavior of Stratospheric ozone in Pakistan’s atmospheric
regions (PARs), using Exploratory Data Analysis (EDA).
In this analysis Probabilistic and mean deviation models
are developed to study the ozone layer depletion for
Pakistan Atmospheric Regions. These models provided a
comprehensive description of the process. This approach
is well explained within the likelihood of computational
analysis that along with their physical interpretations are
very useful for public and private sector organizations.
ACKNOWLEDGMENTS
This work was carried out under a study leave from the
Department of Applied Physics, Faculty of Science
University of Karachi. Grateful acknowledgement is also
due to the Director and the staff of Geophysical Center,
Quetta, Pakistan for having provided the data recorded by
the recently installed Dobson Spectrophotometer under
the auspices of WMO.
REFERENCES
Bartoszynski, R.and Buggi-Niewiadomska M. 1996.
Probability and Statistical Inference,Wiley-Interscience,
New York.
Bisio, A. and Boots S., 1997. Encychlopedia of Energy
and Environment, J Wiley, New York.
Garcia, R,1994. Physics World, April Issue.
Hussain, A, and Quamar J., 2000 In Procd. 5th
International Conference of Science Technology &
Development in the New Millennium, held at University
of Karachi, Pakistan, April: 24- 27.
Kärcher B., Hirschberg M. M., and Fabian,P., 1996.J.
Geophys.Res.101(10):15169-15190.
Koop T. and Carslow, K. S, 1996. Science, Volume 272:
1638-1641.
Millard. S. P., 1998. “Environmental Stats for S-Plus”,
Springer-Verlag, New York.
Michaels P. J., Singer S. F., and Kerr J. B. 1994.Science.
264: 1341- 1343.
Pandit S. M.and Wu,S. M, 1983. Time Sries and System
Analysis with Applications, J. Wiley, New York.
Prather M.,P Midgley, Rowland S. R., and Stolarski R.,
1996.Nature.381 (6583): 551-554.
Robert C. W., and Sheo, S. P. 1987.Ozone in the Free
Atmosphere, van Nostrand Reinhold, New York.
Schoch, M. R.and Mckinney L M. 2004. Environmental
Science,Systems and Solutions. Jones and Bartlett,
Tennesee.
Sugden, T. M and West,T. F. 1985. Chlorofluorocarbons
in the environment. Ellis Horewood, Chichester
Thibodeaux J.L. 1996. Environmental Chemodynamics
Movement of Chemicals in Air, Water and Soil. J. Wiley,
New York.
Tie,X. X. and Brasseur G. 1995. Geophys. Res. Lett.
22(22):3035-3038
Tong, T. 1990. Nonlinear Time Series, A Dynamical
System Approach. Clarendon Oxford.
Wayne, R. P. (ed.). 1991. Chemistry of Atmosphere.
Oxford University Press Oxford, World Meteorological
Organization, Scientific Assessment of Ozone Depletion ,
UNO Report (25), WMO, Geneva. 1991.
Yousuf Zai M. A. K., and Quamar J. 1998. Sing. J.
Phys., 1(1): 69-79.
Yousuf Zai M. A. K., and Quamar J. 2001. The study of
phenomenon of ozone layer depletion as a physical
process. Indian J. Phys. 75B(4): 307-314.
... The ozone layer is passed to Pakistan air space by means of the geographical position (Zahid and Rasul 2010), vertical lifting, horizontal shifting, and ozone mixing ration in the stratosphere that has revealed a decreasing trend in the ozone depth over Pakistan atmospheric region (Zai and Quamar 2001;Zai et al. 2008;Khan et al. 2010;Sumera et al. 2017). Several researchers were working on the techniques and methodologies of nonlinear dynamics to analyze their irregular nature on observed time series data. ...
Article
The stratospheric ozone behavior is of great importance for the shelter of life on the earth’s biosphere from harmful UV radiation. A wavelet technique and explorative analysis are applied on ozone monthly time series data set from 1970 to 2013 for the stratospheric region of Pakistan. Such analysis was performed in order to fully characterize the distinct time-frequency ozone variability. Moreover, in the analysis of ozone irregularity, the main frequency constituents in the data set are estimated by the global wavelet scale. The results obtained in this investigation confirmed particular patterns that could be used to describe the regional condition of ozone. The wavelet approach illustrates a strong power spectrum along with the 2–16-month cycle, indicating an annual oscillation of 12 months for whole data set, which is investigated by the peak of the integration of time average of power over time that also reveals powerful annual pulses. The low variance period results in a band emphasized the presence of variation for seasonal interval of stratospheric ozone for each month starting from January to December. The main objective of this paper is to study the dynamics of ozone concentration at the atmospheric region of Pakistan. It also aims to examine the role of the particular ozone concentration patterns indicating inter-annual irregularity of Ozone status.
... and longitude 61ºE and 75.5ºE) from 1979-1993 by comparing TOMS satellite data and Dobson spectrometer ground observation data. Both the data sets have shown sharp decline in ozone layer thickness over Pakistan [13][14]. The long-term effect of ozone layer depletion appears to be an increase in the ultraviolet radiation reaching the earth. ...
Article
Full-text available
The ozone layer depletion and its harmful impact on living beings have been a greater concern of all the scientists all over the world. The aim of this paper is to reveal the current status of stratospheric ozone over Pakistan. The annual, monthly and seasonal analyses have been performed in order to check the status. The variation in total column of ozone has been observed during these analyses and decrease in total column of ozone has been seen in all the investigations from 1987-2008. The correlation coefficient for JRA forecasted data and observed ozone data is 0.6. Both the data sets show decline in ozone concentration. The total change calculated in annual depth of ozone is −5.67 D.U and −4.2 D.U in monthly depth of ozone. The seasonal analysis shows that the total change in ozone in summer is −6.3 D.U, in spring −10.5 D.U, in winter −3.15 D.U and in autumn −2.0 D.U. Maximum change in ozone thickness has been found in spring and minimum in autumn. The solar radiations, decrease in temperatures of stratosphere and carbon dioxide (CO 2) play significant role in ozone layer depletion. According to the findings of this study solar radiations and carbon dioxide (CO 2) are inversely proportional to the total column of ozone. The correlation coefficient for solar radiations and ozone on annual basis is 0.44 (R 2 = 0.44) and on monthly basis is around 0.35 (R 2 = 0.35). Therefore the more intense the solar radiations the more ozone layer thinning will occur. The correlation coefficient for ozone and carbon dioxide is around 0.3 (R 2 = 0.3) during the study period. The decrease in stratospheric temperatures will cause the cooling of stratosphere which is ultimately responsible for ozone layer depletion. The total decrease analyzed in stratospheric temperatures during the study period is about −1.3℃. It is observed that alarming rise in carbon dioxide (CO 2) concentration is not only contributing to global warming in troposphere but cooling in the stratosphere.
Article
Full-text available
These lectures introduce key concepts in probability and statistical inference at a level suitable for graduate students in particle physics. Our goal is to paint as vivid a picture as possible of the concepts covered.
Book
This extensively revised and expanded edition is based entirely on the multimedia approach to chemical fate in nature. New sections have been added on equilibrium models for environmental compartments, dry deposition of particles and vapours onto water and soil surfaces, chemical profiles in rivers and estuaries, fate and transport in the atmospheric boundary layer and within subterranean media and particles and porous media.
Article
The present book provides a summary of the state of scientific knowledge of stratospheric and free tropospheric ozone as it exists at the beginning of 1983. Ozone photochemistry in the stratosphere is discussed, taking into account fundamental molecular properties, the absorption spectrum of ozone, photodissociation, ozone formation and destruction in the upper atmosphere, the photochemistry of odd-hydrogen, the photochemistry of odd-nitrogen, the photochemistry of odd-chlorine, and photochemistry-temperature coupling. The observed distribution of atmospheric ozone and its variations are considered along with ozone transport, ozone in the troposphere, stratospheric ozone perturbations, and climatic and biological effects. Attention is given to the techniques of observing atmospheric ozone, horizontal-vertical ozone transport and conservative quantities, measurements of tropospheric ozone, the tropospheric ozone budget, ozone models, natural ozone variations, and anthropogenic ozone perturbations.
Chemistry of Atmosphere World Meteorological Organization, Scientific Assessment of Ozone Depletion
  • R P Wayne
Wayne, R. P. (ed.). 1991. Chemistry of Atmosphere. Oxford University Press Oxford, World Meteorological Organization, Scientific Assessment of Ozone Depletion, UNO Report (25), WMO, Geneva. 1991.
The study of phenomenon of ozone layer depletion as a physical process
  • Yousuf Zai
  • M A K Quamar
Yousuf Zai M. A. K., and Quamar J. 2001. The study of phenomenon of ozone layer depletion as a physical process. Indian J. Phys. 75B(4): 307-314.
  • X X Tie
  • G Brasseur
Tie,X. X. and Brasseur G. 1995. Geophys. Res. Lett. 22(22):3035-3038
  • Yousuf Zai
  • M A K Quamar
Yousuf Zai M. A. K., and Quamar J. 1998. Sing. J. Phys., 1(1): 69-79.