How, why, what, when?
David RaynerUniversity of Gothenburg
Why?
'Global' is a place where nobody lives
John D. Cox
Ätran (Falkenberg)
Local impacts
Why?
There are no floods in the average climate.
What is a Global Climate Model?
https://www.e-education.psu.edu/earth103/node/524
6
Climate change impact modeling
Hydraulisk modell
Modified from SMHI
IPCC (AR4 WG 1 Chapter 1 page 113 Fig. 1.4).
Effe
ct o
f Res
olut
ion
Rummukainen, M., 2010. State-of-the-art with regional. doi:10.1002/wcc.008
Regional Climate Models
Source: Scale challenges in high resolution climate modelling
Regional Climate Models
11
Climate change impact modeling
Hydraulisk modell
Source – modifeid from SMHI
GCM outputs are “biased”
Observations GCM
ECHAM5 grid.
Climate model outputs are “biased”
K.M.A. Gabriel, W.R. Endlicher / Environmental Pollution 159 (2011) 2044e2050
Mortality rates for Berlin, 1994
Simulated flow in ÄtranTemperaturePrecipitation
Inflow: obs model
When to downscale?
When impact-assessment requiresrealistic climate time-series for the future.
"The most that can be expected from any model is that it can supply a useful approximation to reality: All models are wrong; some models are useful".
George Box
“Downscaling”.
Future time-series
Historical time-series
Global Climate Model outputs
Downscaling algorithms
• Bias Correction• Historical modification
What is the first-guess future climate?
Modelled historical
Observed historical
Modelled scenario
Bias Downscaled scenario
Bias Correction
Modelled historical
Observed historical
Climate Change
Downscaled scenario
Historical ModificationModelled scenario
Bias correction methods
Bias–correction:Distribution-based scaling
rådata
observationer
DBSdata
rådataobservationer
DBSdata
24
Climate change impact modeling
Hydraulisk modell
Source - SMHI
Bias-correction methods
• Advantages:– Better physical consistency.
• Disadvantages: – One time-series/climate model run.
• Example methods:– Bias-corrected RCM– Empirical-statistical downscaling.
Historical modification example.
Historical modification.• Advantages:
– Unmodelled parameters/time-periods.– Integrate information from models.
• Disadvantages:– Uncertain assumptions– Loose physical coherence
• Examples:– Weather-generators, delta-change– Analogue-resampling
One time-series containsClimate Change signal from 20 GCM runs!
Global Temp from CCSM3 SRESa1b run1.
Historical modification example.
When downscaling, think:
• What do you want to know?• How realistic must inputs be?• What data are available?