Transcript Mastertitelformat bearbeiten

Consistency of observed winter
precipitation trends in northern Europe
with regional climate change projections
Jonas Bhend and Hans von Storch
GKSS Research Institute, Geesthacht, Germany
Mastertitelformat bearbeiten
Motivation
>
Gap between formal detection and attribution studies and
“significant trends” studies
>
Are the recent trends consistent with regional climate change
projections?
— Plausibility arguments
— A priori assumption about the mechanism
— Less informative than DnA but no estimate of natural variability
needed
Mastertitelformat bearbeiten
Data
>
Observations:
— CRU TS 2.1 monthly precipitation
— 0.5° latitude-longitude grid
>
Climate change scenarios:
—
—
—
—
RCAO simulations of the SMHI (PRUDENCE)
0.44° rotated grid
Two different driving GCMs, HadAM3H and ECHAM4/OPYC3
Two emission scenarios SRES A2 and B2
— Four climate change scenarios defined as the difference
between 1961-1990 and 2071-2100 mean
Mastertitelformat bearbeiten
Method
>
Pattern correlation
S  O
i
PCC 
i
 S  O
2
i
i

>
i
2
i
S: Climate change signal
O: Trends in observations
i
Ratio of Intensities
O
S
with:
1
O 
Oi2

n i
and:
1
S 
Si2

n i
Mastertitelformat bearbeiten
Climate change scenarios...
Mastertitelformat bearbeiten
... and observations
Mastertitelformat bearbeiten
Pattern correlation
> Patterns are similar
> Better correspondence with ECHAM scenarios
> Better correspondence with stronger GHG forcing (A2)
Mastertitelformat bearbeiten
Sensitivity of PCCs
> Bootstrap with CRU precip fields
— randomly select precip fields
— compute trends
— correlate trend fields
> Autocorrelation:
moving blocks bootstrap, 5 years
Histogram of PCCs for the Baltic catchment
(shaded) and northern Europe (hatched)
Mastertitelformat bearbeiten
Pattern correlation
> PCCs for the Baltic catchment significant
> PCCs for all of northern Europe are not significant for HadAM B2
> Above findings robust to removal of the NAO
Mastertitelformat bearbeiten
Intensity
> Change in the observations much stronger than in scenarios
— RCM simulations are wrong
— additional forcings
— natural variability
Mastertitelformat bearbeiten
Mean change
> Change in the observations much stronger than in scenarios
— RCM simulations are wrong
— additional forcings
— natural variability
Mastertitelformat bearbeiten
Different trend lengths
>
>
>
PCCs decrease with increasing trend length
Significance levels are not affected by choice of trend length
Intensity and mean change decrease with increasing trend
length
Mastertitelformat bearbeiten
Conclusions - pattern correlation
>
Baltic catchment:
— Regional climate change scenarios are consistent
— Observed and expected patterns are similar and significant
>
Northern Europe:
— Regional climate change scenarios are partly consistent
— Pattern similarity with HadAM signal could be random
Mastertitelformat bearbeiten
Conclusions - intensity
>
Both intensity and mean change suggest that:
>
Assuming the model response to anthropogenic forcing is
correct, a large part (30 to 70 percent) of the observed trends
is due to other factors (e.g. natural variability).
Mastertitelformat bearbeiten
Thank you for your attention.