From Climate Data to Adaptation Large-ensemble GCM Information and an Operational Policy-Support Model Mark New Ana Lopez, Fai Fung, Milena Cuellar Funded.

From Climate Data to Adaptation Large-ensemble GCM Information and an

Operational Policy-Support Model

Mark NewAna Lopez, Fai Fung, Milena CuellarFunded by Tyndall and Environment Agency

Adaptation Challenges

1. Uncertainty in climate information

2. Interactions with other uncertain changes

3. Integrated assessment

Wimbleball Water Resource Zone

Route Map

• Large ensemble climate data

• River flow ensemble

• Water resource system modelling

Large GCM Ensemble: CPdN• Explore model uncertainty by varying settings of poorly

constrained model parameters• HADCM3L model: standard atmosphere & low resolution

ocean.• 26 perturbed parameters (radiation, large scale cloud

formation, ocean circulation, sulphate cycle, sea ice formation and energy convection)

• Initial condition ensembles.• Transient runs:

– 1920-2000 forced with historical CO2, solar and volcanic forcing.– 2000-2080 forced with different possible scenarios

First 246 Completed Simulations

IPCC 4AR models

CPDN model runs

Global Mean Temperature: SRES A2

An

om

aly

fro

m 1

96

1-1

99

0

Data Available• 10-year seasonal mean fields• Monthly mean (time series):

– Large regions (Giorgi)– Selected grid-boxes (including UK)

• Variables include– Total precipitation rate– Convective cloud amount– Surface air temperature (1.5m)– Relative humidity (1.5m)

Modelling Set-up

• Downscale climate in space and time– SW England -> River Exe– Monthly -> Daily

• Generate ensemble of daily river flows– CATCHMOD rainfall-runoff model

• Run flow-ensemble through water resource model

Downscaling: Precipitation

• Gamma transform method– Remove GCM monthly biases– Select daily values from observations

August 1930-1985

Fre

quen

cy

Monthly Precip (mm/d)

ModelObserved

August 2020-2060

Fre

quen

cy

Monthly Precip (mm/d)

ModelObserved

Downscaling: Precipitation

Downscaling: PET

• Calculate GCM PET from– Temperature, RH & cloud-cover (radiation)– Adjust for climatological bias– No daily downscaling

Downscaling: PET

River Flows

River Flows

Month

% C

hang

e

Mean Flow Change: 2020-2039 from 1961-1990

Wimbleball Water Resource Model

• Supplies:– Somerset & Devon (Exeter, Tiverton)

• River & reservoir dominated• 50 ML/d Groundwater• Lancmod WR model

Wimbleball Reservoir: Historic

Monthly Storage, 1930-2005

Month

Sto

rage

(M

l x 1

04 )

Wimbleball Reservoir: 2040 Ensemble

Monthly Storage, 2040

Month

Sto

rage

(M

l x 1

04 )

Wimbleball Reservoir: Changing Risk

September Storage

Year

Sto

rage

(M

l x 1

04 )

Failure to Meet Demand

Devon Demand Failure

Year

No.

Sim

ulat

ions

Failure to Meet Demand

Devon Demand Failure

No.

Sim

ulat

ions

Year

Ave

. Sho

rtfa

ll

Outstanding Issues / Future Work

• Biases in runoff simulations• Simplistic downscaling• Higher multiple year failures in simulations• Scenarios / ensembles of changing demand• Incorporating adaptation options• Staged methodology• Relative likelihoods• Comparison with UKCIP08 / ENSEMBLES