Attributing Variation in a Regional Climate-change

Modelling Experiment

Chris Ferro

Centre for Global Atmospheric Modelling

Department of MeteorologyUniversity of Reading, UK

CRU Seminar, Norwich, 16 December 2004

PRUDENCEStatistical modelIllustrative exampleGrid-point analysisDiscussion

The PRUDENCE Project

European climate simulations

9 limited-area RCMs, drivenby various global GCMs

Control 1961–1990Scenarios 2071–2100

http://prudence.dmi.dk

Orography at 50km resolution

Experimental Design

ObservedSST/SICE

AGCM

RCM

ObservedEmissions

SRES A2Emissions

AOGCM AOGCM

RCM RCM

AGCM

RCM

Control Scenario

Aims

Characterise

projected climate change

effects of model differences

relative importance of different model components: emissions, GCM, RCM and initial conditions

HADCM3 / OPYC3 /HADAM3H ECHAM4

RCM C A C A GroupCHRM 1 1 ETHZCLM 1 1 GKSSHADRM 1 1 HCHIRHAM 1 1 1 1 DMIPROMES 1 1 UCMRACMO 1 1 KNMIRCAO 1 1 1 1 SMHIREGCM 1 1 ICTPREMO 1 1 MPI

Selected Experiments

Land-averaged annual mean 2m air temperature

interpolated to CRU gridECHAM HADAM

HIRHAMRCAO

HIRHAMRCAO

Tem

pera

ture

(°C

)

Tem

pera

ture

(°C

)

Year Year

General Linear Model

Ti j k l is temperature for GCM i, RCM j, Period k, Year l

Ti j k l = i j k + i j k t l + i j k t l2 + Zi j k l

Residuals Zi j k l ∼ N(0, 2i j k), cor(Zi j k l, Zi j k l)

= i k (j, j)

Model Components

Ti j k l = i j k + i j k t l + i j k t l2 + Zi j k l

i j k = + iG + j

R + kP + i j

GR + i kGP + j k

RP +

i j kGRP

overall mean

iG effect of GCM i

jR effect of RCM j

kP effect of Period k

i jGR effect of combining GCM i

with RCM ji k

GP effect of GCM i in Period k

j kRP effect of RCM j in Period k

i j kGRP effect of combining GCM i

with RCM j in Period k

Simplified Model

Ti j k l = + iG + j

R + kP + i j

GR + i kGP

+ ( + iG + j

R + kP) t l + ( + j

R ) t l

2 + Zi j k l

12.1(0.03

) -0.3

(0.15)

1G 0.07

(0.04)

1G 0.10

(0.04)

1R -

0.33(0.01

)1

R 0.11(0.05

)

1P -

2.22(0.04

)1

P -0.26

(0.04)

11GR -

0.13(0.02

) 0.21

(0.05)

11GP -

0.31(0.05

)1

R -0.03

(0.02)

i j k C AHIRHAM-

E0.56 0.62

RCAO-E0.55 0.51HIRHAM-

H0.50 0.43

RCAO-H0.46 0.52i k C A

ECHAM

0.92-

0.10HADA

M0.880.82

Diagnostic PlotsECHAM HADAM

HIRHAMRCAO QQ PP

Variance Partition

%

Trend 91.0

GCM 0.1

RCM 2.0

G x T 1.8

R x T 0.0

G x R 0.3

G x R x T

–

Z 4.7

ECHAM HADAM

HIRHAMRCAO

Temperature Contrasts (C)

HIRHAM

RCAO

HADAM

ECHAM

HIRHAM

RCAO

HADAM warmer than ECHAM at 1961 becomes progressively cooler– warm bias greater for HIRHAM

RCAO warmer than HIRHAM– warm bias greater for ECHAM

Response Contrasts

HADAM – ECHAM response-0.19C/decade (-0.37, -0.02)

HADAM response lower thanECHAM independently of RCM,Period, Year

RCAO response lower thanHIRHAM at start of integration,becomes progressively higherindependently of Period, GCM

Year

C /

decad

e

Grid-point Analysis

Fit model separately at each grid point and plot maps:

Proportion of variation explained by each model term

Evolution of GCM temperature contrasts for each RCM

Evolution of RCM temperature contrasts for each GCM

Evolution of GCM response contrasts for each RCM

Evolution of RCM response contrasts for each GCM

Variation Explained (%)model T Z

G R GR

GT RT GRT

Changes in Mean Temperature (°C)

HIR

HA

MR

CA

OECHAM HADAM

Temperature Contrasts: HADAM – ECHAM

19901961 2071 2100

HIR

HA

MR

CA

O

Response Contrasts: HADAM – ECHAM

19901961 2071 2100

HIR

HA

MR

CA

O

Temperature Contrasts: RCAO – HIRHAM19901961 2071 2100

EC

HA

MH

AD

AM

Response Contrasts: RCAO - HIRHAM19901961 2071 2100

EC

HA

MH

AD

AM

HADCM3 / OPYC3 /HADAM3H ECHAM4

RCM C A C A GroupCHRM 1 1 ETHZCLM 1 1 GKSSHADRM 1 1 HCHIRHAM 1 1 1 1 DMIPROMES 1 1 UCMRACMO 1 1 KNMIRCAO 1 1 1 1 SMHIREGCM 1 1 ICTPREMO 1 1 MPI

Selected Experiments

Residual Standard Deviation (⁰C)

CLM RACMOCHRMREMOPROMESHIRHAM

RCAO REGCM HADRM

0.38 0.39 0.39 0.40 0.44 0.45 0.46 0.48 0.48

Temperature Comparisons (⁰C)

Response Comparisons (⁰C/century)

Review

Method: statistical model synthesises output and enables inference on climate changes, model differences, and the relative importance of model components.

Results: warming signal dominates, choice of RCM affects temperature in mountainous regions, choice of GCM affects temperature and response in Atlantic and response over land, inter-annual variation is greater than GCM and RCM effects over land, effects of model differences change over space and time, other RCMs have different behaviour

Extensions: RCMs, annual cycle, model resolution, EOFs, multivariate, spatial, ensemble generation

Discussion

Danger of over-simplified summaries of model output

Features can be hidden by poor experimental design

If you were a politician…

http://www.met.rdg.ac.uk/~sws02caf

c.a.t.ferro@reading.ac.uk