Transcript CECILIA extremes - Ensembles RT3

Validity of Pattern Scaling Investigated with a
Multi-Model RCM ensemble over Europe
O. B. Christensen1, S. Yang1, F. Boberg1, C. Fox Maule1, P. Thejll1, M. Olesen1, M.
Drews2, H. J. D. Sørup1,3, J. H. Christensen1
1Danish
Meteorological Institute, Copenhagen, Denmark ([email protected])
2DTU Management Engineering, Danish Technical University, Roskilde, Denmark
3DTU Environment, Danish Technical University, Lyngby, Denmark
What did we do?
In order to study the consequences of a much warmer World, the GCM EC-EARTH was used to simulate a
scenario with 1%/y CO2 increase. From this simulation, two 30-year time slices were chosen with a difference
in global temperature of approximately 6 degrees C. These two periods were downscaled with the HIRHAM5
RCM over Europe in around 25km resolution. This integration area is identical to the one used by many
different RCMs in the EU FP6 ENSEMBLES project (http://ensemblesrt3.dmi.dk). Furthermore, a transient RCM
simulation with HIRHAM5 following the RCP8.5 scenario as simulated with EC-EARTH, was available due to a
simulation in the EU FP7 ClimateCost project, and finally one time-slice simulation for RCP4.5 2071-2100 was
available in the same model setup.
A 6 degree warmer world
The GCM EC-EARTH performed a simulation following a 1%/y CO2 increase. The
period 1976-2005 was chosen as control, and a 30-year time slice was chosen at 6
degrees increased global temperature. Figures show changes in temperature (C)
and precipitation (%).
From the available set of transient simulations, 15 ENSEMBLES simulations
following SRES A1B plus the one RCP8.5 simulation, 3 future time slices were
chosen: 2011-2040, 2041-2070, and 2071-2100. This means that we have
available 16*3+2=50 different 30-year future time slices with a climate change
of varying amplitude, along with control simulations, which were chosen as the
period 1976-2005 in the respective simulations.
For analyses of average quantities, extremes, and parameters of the Gumbel
distribution, we have taken averages over land points within areas defined in
the map to the right, the same areas as used since the PRUDENCE project.
The two uncertainty bars shown at the 6-degree points in the plots below are
estimates based on the uncertainty in the linear regression of the other points
(black) and on the typical deviation of all points from the line (red).
Pattern scaling results
Change in temperature (K,
upper row) and precipitation
(%, bottom row) for winter (left
column) and summer (right
column). Dashed lines indicate
uncertainty in the slope of lines
through 0.
A 6 degree warmer Europe
The downscaling simulation over Europe shows considerable change for the
globally 6 degrees warmer climate. The patterns are well-known, but the
amplitudes are not.
Change in 30-year return values
of daily maximum temperature
for Europe (left) and
Scandinavia (right).
Relative change in wet-day 95th
percentile of daily precipitation
(top row) and 30-year return
value (bottom row) for Europe
(left) and Scandinavia (right).
What would Denmark be like in
such a future? To the right is
shown the deviation of current
temperature from the future
Denmark, winter (left) and
summer (right). Basically,
Danish winters will be warmer
than anywhere in Europe
currently.
Conclusions
Change in Gumbel parameters
for Europe (left) and
Scandinavia (right). μ is the
most frequent annual maximum,
and β is a scaling parameter,
which is dominant for very high
extremes.
We have investigated pattern scaling of several quantities over Europe as a whole
and over European sub-regions individually using a multi-model ensemble of a total
of 50 simulated 30-year periods in the future according to various scenarios. The
same integration domain and resolution has been used in all cases.
There are no obvious deviations from linearity, except when looking at high
extremes.
Other European sub-regions display the same behaviour as Scandinavia: More
spread than the entire domain, but the same degree of linearity in global
temperature change. The promised inclusion of CORDEX results has not been
possible yet, partly due to lack of data access.
This study has been published as O. B. Christensen, S. Yang, F.
Boberg, C. Fox Maule, P. Thejll, M. Olesen, M. Drews, H. J. D.
Sørup, J. H. Christensen (2015) Scalability of regional climate
change in Europe for high-end scenarios, Clim. Res., 64, 2538; QR code to the right
Acknowledgments:
The ENSEMBLES simulations in this study have been made under EU FP6, contract 505539. The HIRHAM5 RCP8.5 simulation was part of EU FP7
ClimateCost, contract 212774. The 6-degree simulation was performed under the Danish Centre for Regional Change in the Earth System (CRES).