final_program_abstracts[1]

11 IMSC Session Program
Statistical analysis of global surface temperature and sea
level using nonstationary methods
Thursday - Parallel Session 10
T. Schmith 1 , S. Johansen 2 and P. Thejll 1
1
Danish Meteorological Institute, Denmark
2 University of Copenhagen, Denmark
Global averages of both surface temperature and sea level have increased through the
past century. The relationship between these two non-stationary series is however
quite complicated. For instance: the warming period in the 1940es is not very evident
in mean sea level. This would also be expected a priori, due to the complicated
physical cause-effect relationships involved: Surface temperature has increased
primarily due to external radiative forcing, but is affected by the large heat capacity of
the ocean. Mean sea level in turn has risen due to the effect of rising temperatures t
partly hrough thermal expansion and partly through glacier melt-off.
We analyse this relationship using a bivariate Vector Error Correction Model
(VECM) approach. This is a well-known method within the field of econometrics and
can with advantage be applied to climate time-series analysis. The method is wellsuited
for revealing cause-effect relationships (in a statistical sense) between time
series.
We find that the two series are cointegrated, i.e. inter-related in the long run which
confirms our a priori expectation. But the VAR analysis tells us more: namely that
changes in the sea-level influences changes in the surface temperature while the
opposite is not the case. This is in accordance with the notion that the major part of
the heat capacity of Earth’s climate system resides in the ocean. By adding historical
estimates of long- and short-wave forcing as external explanatory variables we are
able to explain the differences in temperature and sea level behaviour in terms of the
radiative budget.
These results have application in the field of forecasting sea-level rise, given
greenhouse gas emission scenarios.
Abstracts 258