DACH
Deutsch - Österreichisch - Schweizerische
Meteorologen - Tagung
18. - 21. September 2001, Vienna, Austria

Statistical modelling of total ozone measurements of the northern hemisphere

Jörg A. Mäder and Johannes Stähelin
Institute for Atmospheric and Climate Science, ETH Zürich

September 12. 2001

Introduction and Goals

Multiple regression models are widely used in ozone research because they allow to separate long-term ozone trends from natural variability by using explanatory variables (commonly used: Quasi Biennial Oscillation (QBO) and solar cycle, see e.g. Staehelin et al. 2001). Several recent studies provided evidence, that a significant part of the long-term total ozone winter trends (particularly over Europe) can be attributed to changes in the atmospheric condition rather than by the increase in the stratospheric concentrations of ozone depleting substances (EESC: Effective Equivalent Stratospheric Chlorine) (Steinbrecht et al., 1998, 2001, Appenzeller et al., 2000). In this study we describe the variability of total ozone of 65 ground sites of the northern hemisphere (minimal length: 20 years) by using a statistical model in which we introduce a monthly factor which allows to treat the 12 calendar months by one single statistical model instead of using 12 separate model. We start from an extended data set of explanatory variables which is sequentially reduced to select those which explain most of the variability.

Method

The method is based on a multiple linear regression model using the following equation: log O₃ = log O_{3 mean} + c₁ log P_1,m + c₂ log P_2,m + c_x P_x,m + ... + log EESC + e where:

O₃	measured monthly total ozone value of the respective month (see Table 1)
O_{3 mean}	mean value of the respective calendar month of the entire period
P₁, P₂, ... P_x	explanatory variables (see used data) with the respective coefficients c₁, c₂, ... c₃
EESC	effective equivalent stratospheric chlorine (stratospheric concentrations of the ozone depleting substances weighted over their reactivities)
m	monthly factor
e	noise term

The monthly factor m is introduced to describe (i) the seasonal variation of ozone and (ii) the influence of the seasonal variation of the relationship of the explanatory variable on total ozone. We used logarithmic variables if possible to give extreme values less weight. In the procedure we started from 24 explanatory variables of Table 1, thereafter the variables which explain less of the variability were stepwise eliminated using F-values, until all variables except one were eliminated. In the procedure the monthly factor of the worst variable is excluded first before the explanatory variable itself is eliminated.

Used Data

Local variables			Zonal variables			Regional climate pattern, teleconnection indices

total ozone	1	O3	aerosol loading	6	Aero	north arctic oscillation	9	nao	pacific north america	9	pna
pressure at tropopause	2	Ptp	Hemispheric pattern			east atlantic	9	ea	east pacific	9	ep
Global variables			atlantic oscillation	7	ao	east atlantic jetstream	9	eaj	west pacific	9	wp
light flux at 10.7cm	3	Lf107	Sea surface pressure			east atlantic - west russia	9	eawr	north pacific	9	np
EESC	4	Chlor	jakarta sea level pressure	8	jak	polar eurasi	9	pe	pacific transition	9	pt
quasi biennale oscillation	5		enso sea surface pres.	8	sst	scandinavia	9	sca	subtropical zonal	9	sz
positive phase		QBOpos				asian summer	9	as	tropical - north. hemisphere	9	tnh
negative phase		QBOneg							southern oscillation	8	so

Results and Discussion

We obtain for every station a set of ranking of all variables (see below for a example). The set of variables obtained for the mean of all stations (see Figure 1) shows the following characteristics. Beside the seasonal variation the local quantity of the tropopause pressure (tropopause altitude) explains the largest fraction of the variability, followed by EESC, which describes (an almost linear) ozone trend. The next variables include stratospheric aerosols, describing the effects of large volcanic eruptions which have significantly influenced total ozone (e.g. after the eruptions of Mt. Pinatubo and El Chichon), followed by the Arctic Oscillation (ao), which is a hemispheric variable and which is believed to describe the behaviour of the Arctic stratospheric vortex. The teleconnection pattern representing large scale weather conditions explain a rather small part of the variability in this method, possibly because the combination of the hemispheric variable AO and the local variable of the tropopause pressure already partially describes their effects.

rank at arosa, switzerland
1-10	11-20	21-30	31-40	41-49
Ptp	nao	m:pe	m:QBOneg	pt
m	np	pna	m:Aero	m:pt
EESC	m:QBOpos	m:pna	eawr	eaj
ao	m:nao	m:soi	m:eawr	m:eaj
Aero	tnh	m:ep	Lf107	as
sst	m:tnh	jak	m:Lf107	sz
ea	m:np	sca	m:ea	m:jak
QBOpos	m:ao	m:sca	wp	m:as
ep	soi	m:EESC	m:wp	m:sz
m:Ptp	pe	QBOneg	m:sst

The tropopause pressure shows a large signal all over the northern hemisphere, except northern America, where it is much less dominant (see Figure 2).
QBO shows in both phases a dependence on latitude.
The Pacific - North America index yields the largest regional pattern of all teleconnection indices, showing largest signals over North America and smallest signals over Japan (see Figure 3).
item the solar cycle signal is only significant at a few stations.

Conclusion and Outlook

In the next step we will extend the investigation by introducing lower stratospheric temperatures as further variables and we will study the pattern of the long-term trends because the statistical models allow to separate the long-term behaviour caused by changes in the atmospheric conditions from those which have to be attributed to the increase in the concentrations of the ozone depleting substances.

References

Appenzeller, C., Weiss, A.K., and Staehelin, J., North Atlantic Oscillation modulates total ozone winter trends, Geophys. Res. Lett., 27, 1131-1134, 2000.
Staehelin, J., Harris N.R.P., Appenzeller C., and Eberhard, J., Ozone trends: A review, Rev. Geophys., 39, 231-290, 2001.
teinbrecht, W., Claude, H., Köhler, U, and Hoinka, K.P., Correlations between tropopause height and total ozone: Implications for long-term changes, J. geophys. Res., 103, 19,183-19,192, 1998.
Steinbrecht, W., Claude, H., Köhler, U, and Winkler, P., Interannual changes of total ozone and northern hemispheric circulation patterns, Geophys. Res. Lett., 28, 1191-1194, 2001

Data Sources

Download

Poster (.ps): a0 - a4

Figures (.ps): 1 - 2 - 3
Figures (.gif): 1 - 2 - 3

Mail

home

Stand: Tue 15. Jun 2004

DACH Deutsch - Österreichisch - Schweizerische Meteorologen - Tagung 18. - 21. September 2001, Vienna, Austria