SupplInfo Seneviratne etal - iacweb.ethz.chiacweb.ethz.ch/staff/sonia/download/Seneviratne_etal06...present in most RCM simulations but less consistent than the increase in temperature

1

T2M (K), mean, SCEN-CTL T2M (K), stdev, SCEN-CTL 876543210-1-2-3-4-5-6-7-8

1.210.80.60.40.20-0.2-0.4-0.6-0.8-1-1.2

SM (mm), mean, SCEN-CTL SM (mm), stdev, SCEN-CTL

P (mm/d), mean, SCEN-CTL P (mm/d), stdev, SCEN-CTL

ET (mm/d), mean, SCEN-CTL ET (mm/d), stdev, SCEN-CTL

2001501251007550250-25-50-75-100-125-150-200

302520151050-5-10-15-20-25-30

32.521.510.750.50.250-0.25-0.5-0.75-1-1.5-2-2.5-3

0.90.70.50.30.1-0.1-0.3-0.5-0.7-0.9

1.51.2510.750.50.250-0.25-0.5-0.75-1-1.25-1.5

0.50.40.30.20.10-0.1-0.2-0.3-0.4-0.5

Supplementary Figure 1: Changes in mean (left) and interannual variability(standard deviation, right) of JJA temperature (a,b), soil moisture (c,d), precipitation(e,f), and evapotranspiration (g,h) between the CTL and SCEN experiments (SCEN-CTL). The underlying scenario is the SRES A2 and the periods correspond to CTL(1970-1989) and SCEN (2080-2099).

a b

c d

e f

g h

2Supplementary Figure 2: As SF1, but for the mean of the following GCMs:ECHAM5, HADGEM1, and GFDL (see Supplementary Discussion 1).

ET (mm/d), mean, SCEN-CTL ET (mm/d), stdev, SCEN-CTL

P (mm/d), mean, SCEN-CTL P (mm/d), stdev, SCEN-CTL

SM (mm), mean, SCEN-CTL SM (mm), stdev, SCEN-CTL

T2M (K), mean, SCEN-CTL T2M (K), stdev, SCEN-CTL b

d

f

h

a

c

e

g

876543210-1-2-3-4-5-6-7-8

2001501251007550250-25-50-75-100-125-150-200

32.521.510.750.50.250-0.25-0.5-0.75-1-1.5-2-2.5-3

1.51.2510.750.50.250-0.25-0.5-0.75-1-1.25-1.5

1.210.80.60.40.20-0.2-0.4-0.6-0.8-1-1.2

302520151050-5-10-15-20-25-30

0.90.70.50.30.10.050-0.05-0.1-0.3-0.5-0.7-0.9

0.50.40.30.20.10-0.1-0.2-0.3-0.4-0.5

3Supplementary Figure 3: As SF1 and SF2, but for the mean of all 12 consideredGCMs (see Supplementary Discussion 1).

T2M (K), mean, SCEN-CTL T2M (K), stdev, SCEN-CTL

SM (mm), mean, SCEN-CTL SM (mm), stdev, SCEN-CTL

P (mm/d), mean, SCEN-CTL P (mm/d), stdev, SCEN-CTL

ET (mm/d), mean, SCEN-CTL ET (mm/d), stdev, SCEN-CTL

0.60.50.40.30.20.10-0.1-0.2-0.3-0.4-0.5-0.6

1512.5107.552.50-2.5-5-7.5-10-12.5-15

0.90.70.50.30.10.050-0.05-0.1-0.3-0.5-0.7-0.9

0.30.250.20.150.10.050-0.05-0.1-0.15-0.2-0.25-0.3

876543210-1-2-3-4-5-6-7-8

2001501251007550250-25-50-75-100-125-150-200

32.521.510.750.50.250-0.25-0.5-0.75-1-1.5-2-2.5-3

1.51.2510.750.50.250-0.25-0.5-0.75-1-1.25-1.5

f

h

d

ba

c

e

g

4

SCEN-CTL

SCENUNCOUPLED-CTLUNCOUPLED (SCEN-SCENUNC)-(CTL-CTLUNC)

CTL-CTLUNCOUPLED SCEN-SCENUNCOUPLED

T2M (K), stdev

1.210.80.60.40.20-0.2-0.4-0.6-0.8-1-1.2

1.210.80.60.40.20-0.2-0.4-0.6-0.8-1-1.2

1.210.80.60.40.20-0.2-0.4-0.6-0.8-1-1.2

1.210.80.60.40.20-0.2-0.4-0.6-0.8-1-1.2

1.210.80.60.40.20-0.2-0.4-0.6-0.8-1-1.2

Supplementary Figure 4: Analysis of factors contributing to change in JJAtemperature variability [K] between the CTL and SCEN simulations: (a) SCEN-CTL;(b) SCENUNCOUPLED- CTLUNCOUPLED; (c) (SCEN-SCENUNCOUPLED) - (CTL-CTLUNCOUPLED);(d) CTL-CTLUNCOUPLED; (e) SCEN-SCENUNCOUPLED. (See Supplementary Discussion 2).

a

b c

d e

5Supplementary Figure 5: As SF4, but for precipitation [mm/d].

SCEN-CTL0.9

0.70.5

0.3

0.1

-0.1-0.3

-0.5

-0.7

-0.9

SCENUNCOUPLED-CTLUNCOUPLED (SCEN-SCENUNC)-(CTL-CTLUNC)

CTL-CTLUNCOUPLED SCEN-SCENUNCOUPLED

Precipitation (mm/d), stdev

0.90.7

0.5

0.3

0.1

-0.1-0.3

-0.5

-0.7

-0.9

0.90.7

0.5

0.3

0.1

-0.1-0.3

-0.5

-0.7

-0.9

0.90.7

0.5

0.3

0.1

-0.1-0.3

-0.5

-0.7

-0.9

0.90.7

0.5

0.3

0.1

-0.1-0.3

-0.5

-0.7

-0.9

a

b c

d e

6

Supplementary Table 1: Set-up of simulations

Simulations Driving GCM

simulation

Simulation

period

Analysis

period*

Soil moisture

state

CTL HadAM3_CTL 1960-1989 1970-1989 Interactive

SCEN HadAM3_A2 2070-2099 2080-2099 Interactive

CTLUNCOUPLED HadAM3_CTL 1960-1989 1970-1989 CTL

climatology

SCENUNCOUPLED HadAM3_A2 2070-2099 2080-2099 SCEN

climatology

* Whenever mentioned, the “CTL time period” and “SCEN time period” refer to the analysis period.

7

Supplementary Discussion 1:Consistency of mean climate and interannual variability of CTL and SCENsimulations with multi-model RCM and GCM experiments

Within the framework of the European project PRUDENCE, the unperturbedsimulations CTL and SCEN were compared with a number of state-of-the-art RCMswith regard to changes in summer climate variability (Vidale et al. 2006, hereafterreferred to as V06). It was found that the identified increase of summer temperaturevariability in Central Europe is a very consistent feature in all RCMs, though themagnitude, exact spatial distribution and timing of the effect can somewhat differ.Moreover, V06 also showed that the decrease in mean soil moisture content andincrease in soil moisture variability found in Central Europe was present in six RCMsanalyzed in deeper detail (V06, Fig. 10). The increase in precipitation variability ispresent in most RCM simulations but less consistent than the increase in temperaturevariability (V06, Fig. 7).

In the Supplementary Figures 1-3 (hereafter referred to as SF1-3), we extend thisanalysis to IPCC AR4 GCM simulations. SF1-3 display changes in mean and standarddeviation (see Methods) of the JJA temperature, soil moisture, precipitation, andevapotranspiration in the CTL and SCEN simulations (SF1), in the ECHAM5,HADGEM1, and GFDL GCMs (SF2), and in all 12 analyzed GCMs (SF3). For detailsconcerning the GCM simulations, please refer to the Methods section. The choice of 3GCMs displayed in SF2 corresponds to models characterized by high-qualitycirculation patterns in the northern mid- and high latitudes and in Europe (van Uldenand van Oldenborgh, 2006).

The comparison of SF1-3 shows that the analyzed GCMs present similar changes inmean climate and climate variability as the unperturbed RCM experiments (CTL,SCEN). They thus appear overall consistent with the results obtained in our modellingframework. Note that our experiments display a particularly close agreement with thethree high-quality circulation GCMs concerning the exact magnitude of the changes ininterannual summer variability (which are more damped in the 12-GCMs meanvalues).

References:

Vidale, P.L., Lüthi, D., Wegmann, R. & Schär, C. European climate variability in aheterogeneous multi-model ensemble. Clim. Change, conditionally accepted (2006).

van Ulden, A.O. & van Oldenborgh, G.J. Large-scale atmospheric circulation biasesand changes in global climate simulations and their importance for climate change inCentral Europe. Atmos. Chem. Phys., 6, 863-881 (2006).

8

Supplementary Discussion 2:Analysis of factors contributing to changes in summer variability of temperatureand precipitation

We present here a more detailed analysis of the factors contributing to changes insummer temperature and precipitation variability between the CTL and SCENsimulations. The relative contribution of changes in land-atmosphere coupling can beexactly defined using the following equation:

SCEN-CTL = (SCENUNCOUPLED- CTLUNCOUPLED) (1)

+ [(SCEN - SCENUNCOUPLED) - (CTL - CTLUNCOUPLED)]

Following (1), we find two main contributions to the change intemperature/precipitation variability:

• [(SCEN - SCENUNCOUPLED) - (CTL - CTLUNCOUPLED)]: Change in land-atmospherecoupling contribution to temperature/precipitation variability between the CTL andSCEN climate conditions

• (SCENUNCOUPLED- CTLUNCOUPLED): Change in other factors (e.g. - but not exclusively -circulation patterns, sea surface temperatures)

The relative contributions of these two terms to the changes in summer temperaturevariability are displayed in Figure 1g,h as well as in combination with the terms(SCEN-SCENUNCOUPLED) and (CTL-CTLUNCOUPLED) in the Supplementary Figure 4(hereafter referred to as SF4). These figures show that the effect of the change incoupling is mainly located in Central and Eastern Europe, while effects of externalfactors appear stronger in France. Note that Fig. 1h (respectively, SF4c) is consistentwith the analysis of changes in land-atmosphere coupling strength displayed in Fig. 2and Fig. 3a,b.

The same analyses for changes in summer precipitation variability are displayed inFigure 4c,d and SF5. These figures show that the overall patterns of changes (decreasein the Mediterranean, increase in Central and Eastern Europe) appear related to externalfactors, while the particularly high increase of variability in the Alpine region is linkedto changes in land-atmosphere coupling characteristics in the simulations.