MICE Modelling the Impact of Climate Extremes

MICEMICEModelling the Impact of Climate ExtremesModelling the Impact of Climate Extremes

www.cru.uea.ac.uk/projects/mice

The MICE study area

HadCM3 land gridbox

Selected landmass

Other land areas

MICE has six work packages:

(a) Analysis of extremes in model data

1. Extraction of climate extremes for analysis

2. Time series characteristics of extremes under climate change

3. Spatial patterns of extremes under climate change

(b) Impacts evaluation

4. Quantitatively modelling impacts of changes in climate extremes on activity sectors

5. Expert-judgement based approaches to understanding impacts of changes in climate extremes on activity sectors

6. Co-ordination and dissemination

Types of extremes to be analysed:

– exceedences of percentiles and fixed thresholds– highest TMAX and lowest TMIN

Temperature

Precipitation

– exceedences of percentiles and fixed thresholds

Wind– exceedences of percentiles and fixed thresholds– highest maximum and mean windspeed per year

MICE impact categories:

IMPACT CATEGORY CLIMATE EXTREME PARENT VARIABLEa.    Forestry

i.      Wind throw Windstorm Storm tracking

ii.     Forest fire Heat stress, drought Temperature, rainfall

iii.    Ecological damage Flood, drought, heat stress

b. Mediterranean agriculture Heat stress, drought Temperature, rainfall

c. Energy use Summer heat waves Temperature

Heat stress & human comfort

TemperatureDeficit or excess of snow Precipitation, temperature

e. Insurance & civil protection

i.      Property damage ) Wind storm Storm tracking

ii.     Loss of life ) Floods Rainfall, snow

d. Tourism

Core data currently being used in MICEModel Data

HadCM3 scenarios A1f, A2a, A2b, A2c, B2a, B2b

HadAM3H scenarios A2a, A2b, A2c, B2a (Early 2003)

HadRM3H scenarios A2a, A2b, A2c, B2a

variables include:

temp, tmax, tmin, mslp, prec, rhum, wind, wmax, tclw, dswf

“Observed” Data NCEP Reanalysis data for the above variables, 1961-1990

Indices of extremes have been extracted from the above, at all grid points, for further analysis.

Sample Analysis 1: PCA of monthly tmax from three models

• data - monthly tmax from 1961-2099, using HADCM3, CGCM2, and CSIRO Mk2 data (A2 and B2 scenarios)

• objectives – provide background climatology for extremes, using PCA to:

• identify the main modes of variability

• examine effects of different scenarios

• identify inconsistencies in the models

All models give similar results in this PCA.

(Results courtesy of Tom Holt)

A2 B2

Factor 6 of the HADCM3 PCA, explaining 5.9% of the overall variance.

This relatively unimportant factor would be expected to show differences between the scenarios. Yet the spatial patterns are almost identical.

Conversely, the time series are quite different. This is typical for all factors.

Sample Analysis 2: The behaviour of extremes in HadCM3 and HadRM3H

• data – daily 1961-1990, using HadCM3, HadAM3H (early 2003), and HadRM3H data (A2 scenario)

• objectives – provide a better understanding of the behaviour of extremes (fixed thresholds and percentiles) prior to impact modelling.

• results – models appear to simulate extreme temperatures relatively well in mid-latitude regions with mild winters and summers, but significant season-long biases are present at in polar and Mediterranean regions.

(Results courtesy of Matt Livermore)

Example 1: Extremes in Central England

Example 2: Extremes in Northern Norway

Example 2: Extremes in Southern Spain

Sample Analysis 3: Identification of Cyclones for the European Sector winter (=ONDJFM)

• data – daily 1961-1990 and 2070-99, using HadCM3 (A2 and B2 scenarios)

• objectives – provide a better understanding of cyclone behaviour and wind fields prior to impact modelling.

• results – model appear to simulate too many weak centres and not enough deep cyclones – a scale issue.

(Results courtesy of Uwe Ulbrich - UKoeln)

Example 1: “Closed Systems” withp < 1010hPa

Unit: Cyclone days per month per 5x5 degree grid box

1961-90 A2a (2070-99) – 1961-90

Sample Analysis 4: Forest Fires

Typical high values for the above codes and indices. (Results courtesy of Marco Bindi – FMA)

These results apply to a single HadRM3 grid cell in Tuscany.

MICE over the next few months: Climate Model Analysis (1)

Time series characteristics of extremes under climate change

• Complete validation of HadCM3 and HaDRM3• Changes in extremes in model time

• compare extremes for 1961-1990 with 2070-2099• highest and lowest per year• three/four independent extremes above a threshold per year

• analysis techniques for extremes• stationarity, trends, periodicity, return periods• clustering and Poisson processes• GEV analysis• GPD analysis

MICE over the next few months: Climate Model Analysis (2)

Spatial patterns of extremes under climate change

• Map event occurrence for overlapping 10-year periods• Use PCA to identify major modes of extreme events

• relate the modes to NAO and AO using regression• Use CCA to examine slp and 500 hPa height• Continue analysis of model and NCEP storm tracks

MICE over the next few months: Impact Modelling

Quantitatively modelling impacts of changes in climate extremes on activity sectors

• Energy• Forest Fires• Insured Losses• Mediterranean Agriculture• Scandinavian Forests