Total Column Ozone Variability Over Toronto,
Ontario, Canada
Reza Hosseinian and William A. Gough
Environmental Science, University of Toronto at Scarborough, 1265 Military Trail, Scarborough, Ontario, Canada
Ozone data was examined for Toronto, Ontario, Canada using statistical analysis. It was found that total column ozone thickness
has been decreasing at an average rate of 0.64 percent per year over the last 38 years. Howe ver, there is co nsiderable v ariability
on several time scales. A spectral analysis was performed on a 13 month running mean of the ozone data. Aside from the
expected seasonal c ycle, ozon e varied at ti me scales cons istent with the quasi-bienn ial oscillation (Q BO) of th e stratosphe ric
circulation, and the sun spot cycle.
Keyw ords: total co lumn ozo ne, quasi-b iennial oscilla tion, sun spo t cycle, Toro nto
O zone is capable of absorbing wavelengths (ultraviolet
radiation) of biolog ically dam aging u ltraviolet light.
This radiation has been linked to health and
environmental concerns. Most of this ozone (90
percent) is found in the stratosphere, the layer of the
atmosphe re lying between the altitudes of 10 and 50
kilometers (Kow alok 19 93). He at genera ted from this
absorption causes the tempera ture to incre ase wit h
altitude in the stratosphere. T he resulting temp erature
profile is largely re sponsib le for the dy namic sta bility
of the stratosphere (Shen e t al. 1995). Hence, the
presence of the stratosph eric ozon e layer is vita l both to
human health and to the dynamic stability of the
stratosphere.
Most research on ozone depletion focuses on the
dramatic changes in the Antarctic Ozone Hole. The
purpose of this paper is to examine the temporal
variability in the thickness of the ozone layer over the
Great Lakes area, as typified by data collected at
Toronto, Ontario, Canada (Hosseinian 2000). We w ish
to address the following two questions: has there been
a decrease in total column ozone in this region? And
what is the source of interan nual and interd ecadal
variability in the total column ozone?
In this study, statistical analysis is used to examine
the trend in the total column ozone concentration for
the past four decades (1960 to 1998).
Nonanthrop ogenic variations in the total ozone
The Great Lakes Geographer, Vol. 7, No. 2, 2000
concentration are examined and some causal
mechanisms for these variations are presented. The use
of quantitative statistical analysis of the ozone data can
readily enhanc e the search for unus ual or abnormal
changes in the ozone (Hill 1982). These analyses can be
used to separate physical and chemical mechanisms
from ran dom v ariations.
Background
Ozone Circulation
Stratosp heric circulation p lays an es sential role in
determining the spatial and temporal distribution of
ozone. Ozone circulation can be illustrated through the
Brewe r-Dobs on circula tion mo del (Figu re 1).
It consists of a meridional circulation in each
hemisphere, with air rising into the stratosphere in the
tropics (where there is little seas onal varia tion in
ozone), moving poleward, with descent and
entrainment into the troposphere at high latitudes. This
mass circulation transfers ozone from the tropical
production regions and allows accumulation near the
poles, accounting for the spring polar maximum (James
1994) .
Ozone Chemistry
Human activity has caused a marked increase in the
concentration of certain important trace gases in the
Total Colu mn Oz one Var iability
atmosphere, including OH, NO, ClO, and BrO
(Brasseur and Hitchman 1988).
These ozone
destructive radicals can have a significant impact on
ozone concen tration, either directly through
photochem istry or indirectly by changing the radiative
budget and hence the temperature and chemistry. As
the temperature structure changes, transport of
chemic als by mean circulation and wa ves can also
change .
energy and momentum conservation simultaneously. In
reactions (3) and (5 ), the odd oxyge n particles (i.e ., O
and O 3) are converted into each other, and are much
faster than reactions (1) and (4) in which odd oxygen
particles are created and d estroyed (H oughton , 1986).
However, the more recent theories of ozon e chemistry
not only take into account the classical theory, but they
also accoun t for other c atalytic cycles that result in the
net reaction 2O 3 ÷3O 2 and which do not require the
presence of an O atom. Th ese cataly tic cycles involve
ozone destructiv e radicals. For ex ample, a c atalytic
cycle based on the coupling of the OH and ClO families
involve :
OH+O 3 ÷HO 2 +O 2
Cl+O 3 ÷ClO+O 2
ClO ÷HO 2÷HOCl+O 2
HOCl+h <÷OH+Cl
___________________
Net: 2O3 +h <÷3O 2
(6)
(7)
(8)
(9)
(10)
Also, another c atalytic cycle based on the coupling of
BrO and ClO families involve:
Figure 1: Schematic illustration of the Brewer Dobson
circulation, proposed to account for the observed distribution
of various conserved trace constituents in the lower
stratosphere (James 1994). Arrows indicate direction of
transport.
The chemic al processes that are responsible for the
production and destruction of ozone involve a series of
chain reactions. The simple 'classical' theory of ozone
initially considered reactions involving oxygen only,
namely (H oughton 1986):
O 2 +h <÷O+O
O+O+M ÷O 2 +M
O+O 2 +M ÷O 3 +M
O+O 3 ÷2O 2
O 3 +h <÷O 2 +O
(1)
(2)
(3)
(4)
(5)
Equations (1) and (5 ) describe photod issociation in
the presence of solar radiation. Equation (1) occurs for
wavelengths less than 246 nm and Equation (5 ) occurs
for wavelengths less than 1140 nm but most strongly
below 310 nm. Reactions (2) and (3) are three body
collisions, the third body M being required to satisfy
56
Cl+O 3 ÷ClO+O 2
Br+O 3 ÷BrO+O 2
ClO+B rO ÷Cl+Br+O2
___________________
Net:2O 3 ÷3O 2
(11)
(12)
(13)
(14)
(Shen et al. 1995 ).
Ozone destructive radicals such as chlorine and
bromine are greatly enhanced in the current
atmosphere. They are transported to the stratosphere by
means of other anthropogenically released chemicals,
such as chlorofluoro carbons (CF Cs). Once the y are in
the stratosphere, these ch emicals break d own thro ugh
the process of photodissociation releasing the fixed Cl
and Br radica ls. CFC s are partic ularly effective at
reducing ozone because of their long half-life of
approx imately 100 years and consequently CFCs have
been strongly linked to the observed global ozone
reductio n in the latter part of the in dustrial era .
Ozone Change Processes
Changes in stratospheric ozone in the mid-latitudes can
result from two broad catego ries of mechanisms;
Hosseinian and Gough
chemic al and dynamical. Ozone creation and
destruction both cause variation in ozone concentration.
Ozone destruction, as outlined in the previous section,
is dependent on the introduction of ozone destroying
radicals, such as derivatives of CFCs. Ozone creation
is depend ent on th e supply of solar UV radiation. Any
variations of this supply could result in va riations in
ozone. The solar sun spot cycle of approximately 11
years (Figure 2) provides the most temporally relevant
variation, other than the seasonal cycle. The 11-year
cycle is not constant, however, it varies between nine
and 13 years with variations in recent decades at the
low end of the range (Lean 1991).
There is
approx imately a 0.1 percent decline in total solar
irradiance from the maximum to the minimum of a
solar sunspot cycle (Waple 1999). Dynamically, as
outlined in Figure 1, ozone is transported poleward
from the equatoria l source reg ion via the stratosp heric
circulation .
The most pre valent va riation in th e stratosph eric
circulation is the quas i-biennia l oscillation (QBO)
(James 1994). The QBO is observed in the middle and
lower stratosph ere as a var iation in the equatorial zonal
winds, which oscillate betw een westerlies and easterlies
approx imately every two to three years (Figure 3)
(James 1994). When observed variations in the tropics
are filtered for annual and semiannual periodicities and
long-term trends, the residual variation shows a strong
relationsh ip between total ozone and the zonal tropical
winds of the QBO (London 198 5; Hamilton 1998).
What is less well known, but increasingly becoming
apparent operationally, is that the QBO signal also
exists in the extratropical stratosphere. There is as yet
no generally accepted explanation of how the equatorial
QBO signal is tran smitted to polar latitudes (Tung and
Yang 19 94; Kinners ley and Tu ng 1999 ).
Figure 2: Time series of solar sun spot number for 1960 - 1999.
57
Total Colu mn Oz one Var iability
Figure 3: Tropical stratospheric winds, measured over Singapore, from 1960 to 2000
Methods
Source of Data
The data used in this study was retrieved from the
World Ozone and Ultraviolet Radiation Data Center
(WOU DC). WOUDC is one of seven recognized
World Data C enters w hich are p art of the W orld
Meteorological Organization , Global Atm osphere
Watch program. The WOUDC is operated by the
Experimental Studies D ivision of the atmo spheric
Environment Service, E nvironm ent Canada, located in
Toronto, O ntario, Canada (WOU DC 19 98).
For this study , the Wo rld Ozone Data Center
(WODC) data set was used. There are over 300
stations represented in the archive, some of which
contain over 35 ye ars of continuous data. The data for
this work was retriev ed from station 06 5 located in
Toronto (Elevation: 198 m, Latitude: 43.78o N, and
Longitude: 79.47o W) and maintained by the
Atmo spheric Environment Service of Environment
Canada. The da ta used from this archiv e are the da ily
Total Column Ozone which represents the total
thickness of the ozone lay er in Dobso n Units (DU ),
defined as 0.01 mm thickness at standard temperature
58
and pressure i.e ., 0 /Celsius and o ne atmosph ere
pressure, and it spans a period of nearly 38 years from
January 1, 1960 to July 30, 1998.
To assess the influence of the solar sunspot c ycle,
data from the Solar and Terrestrial Physics Division of
National Geographic Data Center were obtained.
Tropical stratosph eric wind data, m easured over
Singapore, were obtained from the Institute of
Meteo rology at the Free Unive rsity of B erlin.
Statistical Analysis
Mon thly means w ere calculated from the daily total
column ozone data, and these are used throughout this
work. To examine trends, a time series line plot of the
month ly means of total ozone was created (Figure 4),
revealing the presence of strong seasonal cycles and
resulted in a low coefficient of determination
(r 2 =0.033). A low r2 value indicates that the regression
line can accoun t for only a sma ll percentage of the
variation s in total ozo ne.
Hosseinian and Gough
Figure 4: Monthly total column ozone data for Toronto from 1960 to 1998
To illustrate larger time scale trends, the seasonal
variations were minimized using two separate methods.
The first method was th e use of a 13 month running
mean. The ozone value for every month was calculated
as a mean of the data six months before and after. The
running mean method is a helpful graphical aid for
visualizing any long term patterns or apparent trends
(Hill 198 2). Afterw ard, a time series spectral analysis
was performed on the running m ean data in order to
observe any possible long term variations other than the
obviou s seasona l cycles.
The second method that was used to minimize the
data variability is a metho d suggested by Hill (198 2),
where the average of all values corresponding to the
same month is subtracted from that month; e.g., total
month ly ozone for January 1982 minus the mean of
month ly ozone for all months of January , from 19 60 to
1998. To compare the results of these two methods, a
basic statistical analysis w as perform ed on th e data
obtained from eac h metho d.
Results
Figure 4 is a time series line plot of monthly total
column ozone for Toronto, from 1960 to 1998. One
gross feature is an apparent decrease in total ozo ne with
time. As the regression line indica tes, the slop e is
0.053. Every month, on average, there is a 0.053
percent decrease in total ozo ne (0.64 perce nt per year,
or 6.4 percent per d ecade). Ano ther feature is the
reduction of season al amplitu de with time, which
appears to be tied to the absolute value of the ozone
concentration, a point ex plored in more d etail in
Hosseinian (2 000).
Figures 5a and 6 are the line plots of monthly total
colum n ozon e for Tor onto for the same time period as
Figure 4, applying the two trend analysis methods
described above. In Figure 5a, the running mean
method of de-trending is used and in Figure 6, the
second method is used. U sing a linea r regressio n fit, the
same decreasing trend observe d in Figu re 4 is found in
Figures 5a and 6. In these two figures, however, the
seasona l variations have be en redu ced.
59
Total Colu mn Oz one Var iability
Figure 5: Monthly total column ozone over Toronto for 1960 - 1998. Thirteen month running mean trend analysis has been
applied. a) Linear regression fit. b) Polynomial fit.
60
Hosseinian and Gough
Figure 6: Monthly total column ozone data for Toronto from 1960 to 1998. The second method of trend analysis has been
applied (see text for details)
A comparison of Figures 5a and 6 reveals that the 13
month running mean trend analysis method (Figure 5)
is a more effective way of removing the seasonal cycles
(high frequen cy variatio n) from th e data. This is
further substantia ted by T able 1 w here som e basic
statistical results of the three data sets (i.e. original
month ly total ozone, monthly total ozone with 13
month running average applied, and monthly total
ozone with seasonal cycles removed) are summarized:
The coefficient of determ ination (r 2 ) increases from
three percent to 57 per cent for th e first meth od and to
19 percent for the second method. Other methods for
removing the seasonal cycle are reviewed in Hosseinian
(2000).
In Figure 5(b) a polyno mial fit is also included. The
polynomial fit indicates that mo st of the de crease in
total ozone that was evident in Figure 4 took place
between 1979 a nd 199 2, with a recent increasing trend
in ozone c oncentr ation beginning in 1993. The
influence of volcan ism can a lso be seen as there is a
significant decrease in total ozone concentration at or
shortly after 1991 and a similar, albeit muted, p attern is
also apparent after 1982. This is coincident with the
eruption of Mt. P inatubo in 1991 and El C hichòn in
1982.
Figure 7 is a time series spectral analysis for the mean
month ly total column ozone data with the 13 month
running mean applied. This graph summarizes some of
the mecha nisms th at can resu lt in variations in total
ozone concentration at different time scales. The three
distinct peaks that are highlighted in this figure
correspond to long p eriod w aves that a re respon sible
for causing variations in the total ozone at Toronto.
Peak one has a perio d of 12 m onths, an d it simply
represen ts the seasonal v ariation ca used du e to
meridional circulation which was no t entirely
suppressed by the 13 month running mean. Peak two,
which is not as sharp as the other peaks (representing a
wider range of periods), h as a period of appro ximately
2.7 to 3.0 years, which qualitatively matches the quasi
biennial oscillation (Q BO) described in Figure 3. To
further substantiate that the period of QBO from 1960
to 1998 match es the perio d of peak 2 in Figu re 7, a
cospectral density analysis was performed on the QBO
61
Total Colu mn Oz one Var iability
Table 1 Regression Analysis of Mean Monthly Ozone Data
Original Data
Regression line
Seasonal cycles removed
Thirteen month running
mean applied
y=359.8-0.053x
y=360.158-0.0x
y=12.97-0.053x
-0.053
-0.053
-0.053
Coefficient of Determination (r2)
0.033
0.57
0.16
Variance
1347
93
308
Slope
data from 19 60 to 1998 and annual ozone concentration
data from 1960 to 1998, as seen in Figure 8. As Figure
8 shows , there is a strong p eak with a period o f 2.5
years (frequency = 0.034) suggesting the source of the
second peak is the zonal tropical wind in the
stratosphere. The extrem es in the prevailing w inds vary
from cycle to cycle, but the peak easterly usu ally
exceeds 30 metres/secon d (m/s), w hile the w esterly
extreme is usually between 10 to 20 m/s (Hamilton
1998). Hence, it is deduced that easterlies are gene rally
stronger and capable of transporting more ozone from
the equatorial region to higher latitudes, thereby
contributing to variations in the total ozone
concentration over Toron to. The third pea k in Figure 7
has a period o f approx imately 9 .5 years an d this clearly
matches the solar sunspot cycle period (Figure 2)
(Wilso n 1994 ).
To further examine this, a cospectral density a nalysis
was performed on the solar activity data from 1960 to
1998 and annual ozone concentration data from 1960 to
1998 (Figure 9). As Figure 9 shows, there is a strong
peak with a period of 9.52 years (frequency = 0.105)
confirming that the source of the third peak is the solar
sunspot cycle activ ity consisten t with oth er results
(Haigh 1994).
The final peak occurs at the low frequency end of the
spectrum and represents the secular trend of ozone for
the entire span of the data. This informa tion is already
contained in the regre ssion ana lysis. Thu s, the
variations in total ozone in the Greater Toronto Area,
aside from seasonal cycles, can also be related to quasi
biennial oscillations, to sun spot activity and to a long
term secular tren d, likely the result o f anthrop ogenic
emission s of CFC s.
It is important to note that these cycles coexist and at
any given time, depending on their relative phase, they
can amplify or mitigate ozone variability. For example,
62
the recent rise in ozone since 1993 (Figure 5) can be
attributed to a solar activity maximu m in 1996 (Figure
2), the dominance of easterlies in tropical w inds in the
stratosphere since 1996 (Figure 3), and likely the
decline in CFCs due to emission restrictions resulting
from the 198 7 Mo ntreal Pro tocol. Knowledge of these
cycles provides the basis for a re alistic ozon e forecast.
Discussion
The 38 year s of ozon e data that w ere analy zed in this
study revealed that the total ozone over the Greater
Toron to Area over the period of record showed an
overall decrease of 0.64 percent per year or 6 .4 percent
per decade. This regiona l estimate is co nsistent w ith
the results obtained by others. Miller et al. (1995),
using data covering the period between 1968 to 1991,
found ozone trends in both the lower stratosphere (1520 km) and the upper stratosphere (35-50 km), of about
-6 percent p er decad e. Similarly, Randel et al. (1999),
for the period between 1979 to 1996, showed trends in
total ozone concentration at all altitudes between 10 to
45 km, of about approximately -7 percent per decade.
Chen and Nunez (1998) studied the temporal and
spatial variability of ozone in Southwest Sweden for the
period 1988 to 1997, and found that the long term total
ozone trend has been decreasing at a rate of 0.79
percent per ye ar (or 7.9 percent p er decade).
As discussed previously, seasonal variations of ozone
are due to the presence of meridional circulation as
illustrated in the Brewer Dobson circulation model
(Figure 1). Two separate statistical procedures were
used to minim ize season al variations from the ozone
data. The run ning m ean me thod resu lted in an effective
mitigation of seasonal variation and th erefore, the
adjusted total ozone data obtained using this method,
was used in the subsequent analy sis.
Hosseinian and Gough
Figure 7: Spectral analysis of monthly total column ozone data for Toronto from 1960 to 1998 with 13-month running mean applied
Figure 8: Cospectral analysis of annual ozone data and QBO variations
63
Total Colu mn Oz one Var iability
Figure 9: Cospectral analysis of annual ozone data and solar cycle variability
The smoothed data showed four spectral peaks which
were linked to a residual seasonal varia tion, qua sibiennial oscillation (QBO), solar sun spot variation and
a longer term secular trend likely the results of
anthrop ogenic influences (CFCs). The solar variab ility
provided a precise m atch to a spectral peak in the ozone
data. The QB O pro vided a close match with another
spectral peak, ho wever, th e spectral p eriod w as slightly
longer than the tropical QBO data. It is possible that
only the stronger QBO episodes (with longer periods)
influence ozone d istant from the equator. On-going
work on the spatial variability of the QBO influence
will decid e this issue. It should be n oted, howe ver, we
are making inferences about causal relationships based
solely on statistical analysis. Althoug h the results are
consistent with pro posed p hysical m echanism s, this
does not constitute direct evidence of causality.
The analysis in this paper illus trates the diff iculty in
predicting future oz one le vels. Although a decreasing
trend was fou nd for the 38 year period, th ere is
conside rable variability on a variety of time scales.
Thus, a prediction for a specific year needs to consider
64
not only this long term trend but also the state o f the
QBO and the sun spot cycle.
Acknowledgements
The authors wish to thank the A tmospheri c
Environment Service of Environment Canada for
providing access to the ozone data. The research was
supported in part by the National Science and
Engineering Researc h Cou ncil of Ca nada.
We
acknowledge the assistance of two anonymous
reviewe rs.
References
Brasseur, G., and H. M., Hitchman, 1988.
Stratosp heric Response to Trace Gas Perturbations:
Changes in Ozone and Tem perature Distributions.
Science, 240, 634 637.
Chen, D. L., and Nune z, M., 19 98. Tem poral and
spatial variability of total ozone in Southwest Sweden
revealed by two ground-based instruments.
International Journal of Climatology 18(11), 12371246.
Hosseinian and Gough
Haigh, J. D. 1994. The role of stratospheric ozone in
modulating the solar radiative fo rcing of climate.
Nature 370, 544-546.
Hamilton, K., 1998. Dynamics of the Tropical Midd le
Atmosphere: A Tuto rial Review. Atmosphere-Ocean
36(4), 319-354.
Hill, J. W., 1982. Ozone Trend Analysis. In: Bower,
A. F., and Ward, B. R. Stratospheric Ozone and
Man. Volume I. CRC Press, Inc. Boca Raton,
Florida. 70-86.
Hosseinian, R., 2000. Status of total Column Ozone
Concentration over North America. M aster Thesis,
Department of Geograp hy, Unive rsity of Toronto,
134p.
Houghton, T. J., 1986. The Physics of Atmosph eres.
Cambridge Unive rsity Press . Cambridge, U.K . 271p.
James, N. I., 1994. Introduction to Circulating
Atmospheres.
Cambridge University Press.
Camb ridge, U .K. 422 p.
Kinnersley, J. S., and K . K. Tun g, 1999 . Mechanisms
for the extratropical QBO in circulation and ozone.
Journal of the Atmospheric Sciences, 56(12), 19421962.
Kowalok, M. E., 1993. Common Threads - Research
Lessons from Acid Rain, Ozone Depletion, and
Global Warming. Environment 35(6), 13-38.
Lean, J., 1991. Variations in the sun ’s radiative output.
Reviews of Geophysics 29, 505-535.
London, J., 1985. Distributio n of Atm ospheric Ozone.
In: Whitten, C. R., and S. S. Prasad. Ozone in the
Free Atmosphere. Van Nostrand Reinhold Company.
New York. 11-80.
Miller, A. J., Tiao, G. C., Reinsel, G. C, Wu ebbles, D .,
Bishop, L., Kerr, J., Nagatani, R. M., Deluisi, J. J.,
and Mater, C. L., 1995. Comparisons of observed
ozone trends in the stratosphere through examination
of umkehr and balloon ozonesonde data. Journal of
Geophysi cal Research-Atmospheres 100(D6), 1120911217 .
Rande l, W. J., Stolarski, R. S., Cunnold, D. M., Logan,
J. A., Newchurch, M. J., and Zawodny, J. M., 1999.
Atmosphere-Trends in the vertical distribution of
ozone. S cience 28 5(543 4), 1689 -1692 .
Shen, L. T., Wooldridge, J. P., and J. M., Molina, 1995.
Stratosp heric pollution and ozone depletion. In:
Singh, B. H. Compo sition, Ch emistry, a nd Clim ate
of the Atmosphere. Van Nostrand Reinhold. New
York. 394 -442.
Tung, K.K. and H. Yang 1994. Glo bal QB O in
circulation and ozone. 1. Reexamination of
observational evidenc e. Journa l of the Atm ospheric
Sciences, 51:2699-2707.
Waple, A. M., 1999. The sun -climate rela tionship in
recent centuries: a review. Progress in Physical
Geography 23(3), 309-328.
Wilson, R. P., 1994. Solar and Stellar Activity Cycles.
Cambridge Unive rsity Press . Camb ridge, U.K. 274p.
WOUDC, 1998.
World Ozone and Ultraviolet
R a d i a t i o n
D a t a
C e n t e r .
http://ww w.tor.ec.g c.ca/wo udc/.
65