Impact of climate change on France watersheds in 2050 : A comparison of dynamical and multivariate statistical methodologies By : Christian Pagé, CERFACS.

Impact of climate change on France watersheds in 2050 :

A comparison of dynamical and multivariate statistical

methodologies

By :By :

Christian Pagé, CERFACSChristian Pagé, CERFACSJulien Boé, CERFACSJulien Boé, CERFACS

Laurent Terray, CERFACSLaurent Terray, CERFACS

Florence Habets, UMR SisypheFlorence Habets, UMR Sisyphe

Éric Martin, CNRM, Météo-FranceÉric Martin, CNRM, Météo-France

Ouranos, 20 May 2008Ouranos, 20 May 2008

Problematic Problematic of Downscalingof Downscaling► Why use a statistical approach?Why use a statistical approach?

MethodologyMethodology► Statistical Downscaling & Weather TypesStatistical Downscaling & Weather Types

► Principles & HypothesisPrinciples & Hypothesis► Validation (also Hydrology)Validation (also Hydrology)

ApplicationApplication► Impact of climate change on France watershedsImpact of climate change on France watersheds

► UncertaintiesUncertainties► Comparisons against Quantile-QuantileComparisons against Quantile-Quantile

Summary & FutureSummary & Future

Outline

Ouranos, 20 May 2008Ouranos, 20 May 2008 22

How can we evaluate impacts of climate change?

Problematic: Generalities

Downscaling

Meteorological forcings <10km

Impact model

Precipitations (mm/day)

Perturbed climate meteorological fields ~ 250 km

Climate model

Precipitations (mm/day)


Statistical downscaling

Dynamicaldownscaling

Two main methodologies

Statistical relationship:

Local fields & Large-scale forcings

Resolve dynamics and physics:

Numerical model

Can be used separately or in combination

Downscaling

Problematic: Generalities


MCGOACNRM-CM3

GHG, Aerosols

F: Calibration Validation

Regional ModelARPEGE-VR

Bias correctionSpatialisation

Predictors

Boundary Conditions (also Oceanic)

Raw ForcingsOBS.

OBS.

Impact Model: ISBA-MODCOU

Local Forcing variables

Statistical Downscaling

Dynamical Downscaling

Downscaling Methodologies

Predictors

55

Weather Type: southerly winds

Arrows: 850 hPa WindLines: MSLP anomalies

Precipitation anomalies (%)

Dynamical downscaling

Observations

8 km

Regional Climate Model

60 km

Global Climate Model

280 km

66

Bias Correction

Several Methodologies (Déqué, 2007)

► Perturbation

► Quantile-Quantile


Δ

Obs. Scenario

Climate Change

Probability Density Functions

1)


2)

ModelPresent

ModelFuture

OBS.




Bias Correction


► Perturbation



CorrectedModelPresent

CorrectedModelFuture

3)



Bias Correction


► Perturbation


Statistical downscaling: General methodology

R = F (L, β)

Local ScaleClimate Variable R

10m wind, precipitation, temperature

Local Geographical Characteristicstopography, land-use, turbulence

Global ScaleClimate Variable L

(predictors) MSLP, geopotential,

upper-level wind

β such that║R – F(L, β)║ ~ MinF based on Weather Typing


Statistical downscaling: Current methodology

Based on:• NCEP re-analyses

• Weather typing► Mean Sea-Level Pressure

• Météo-France Mesoscale Meteorological Analysis (SAFRAN)

• France Coverage• 1970-2005• 8 km spatial resolution from coherent climatic zones• 7 parameters

• Precipitation (liquid and solid)• Temperature• Wind Module• Infra-Red and Visible Radiation• Specific Humidity

SAFRAN 8-km resolution orography



Boe J., L. Terray, F. Habets and E. Martin, 2006: A simple statistical-dynamical downscaling scheme based on weather types and conditional resampling J. Geophys. Res., 111, D23106.

For a given day j in which we know the Large-Scale Circulation

1. Find closest weather type (daily data)• Euclidian distance over first ten principal components• Select all Ri days of this type• MSLP and Temperature index

2. Reconstruct precipitation index: using regression of learning period and MSLP of climate model




• Look for analogs (15 days) among all Ri days

• Closest in terms of precipitation and temperature index

► Belonging to the same decile

• Randomly choose one day

• Use SAFRAN data for the chosen day

• Apply temperature correction if Tindex - TNCEP > 2 C

• Correct precipitation (solid/liquid) and IR radiation

• Applicable if having long enough observed data time series

Statistical downscaling: Validation

Is Climate Model simulating

correctly Weather Types ? YES

Precipitation mm/day

Period: 1981-2005

Downscaling:MSLP ARPEGE

A1B ScenarioRegional Simulation

SST fromCNRM-CM3 model

DJF

JJA

Safran Downscaling

0.6 7 0.6 7

0.5 5 0.5 5

1414

Statistical downscaling: Validation: Hypothesis

3 Main Hypothesis 1.Predictors

• Strong link with regional climate• Simulated correctly by model

2.Statistical relationship F still valid for perturbed climate.

• Cannot be validated or invalidated formally. Also true for physical parameterisations and bias correction.

3.Predictors encompass completely the climate change signal.


Hypothesis 1: predictors has strong link with regional climate

Precipitation: 8 weather types

Example for 2 winter type

MSLP AnomalyNDJFM

MSLP AnomalyNDJFM

-16 +16-16 +16

RatioPr(reg)/Pr(moy)

RatioPr(reg)/Pr(moy)

WT1 WT2

0 +3.5 0 +3.51616

Data courtesy of Météo-France

Hypothesis 1: predictors simulated correctly by model

Winter types 1950-1999:

WT5 (MSLP, composite anomaly in hPa)

NCEP Reanalyses ARPEGE GCM-VR

Spatial correlation > 0.96for all weather types

1717

Hypothesis 2 & 3: Predictors encompass completely climate change signalStatistical relationship still valid for perturbed climate

Perfect Model Validation

Precipitation mean over France

Reconstructed

Precipitation amountchange in %of current mean

(2100_2050) – (2000_1970)

A1B Scenario, Spring

-0.35 +0.35

SPRING

Pre

cip

itat

ion

mm

/day

1818

Tendencies ΣPr 1951-2000

Observationsvs

Reconstruction

Color: station latitudeSouthSouth North

Changes of weather type occurrence ► Precipitation Tendencies spatial structures (r=0.92)


Precipitation

1919


• Weather Type Occurrence changes cannot explain observed temperature tendencies► Mandatory to take into account temperature as a predictor

RATIO Temperature Tendencies [Reconstructed] / [Observed]

1951-2000 Period

Temperature

Data courtesy of Météo-France


Statistical downscaling: Validation: Hydrology

Flow Validation

Winter MeanOBSNCEP (0.85)SAFRAN (0.97)

Annual CycleOBSNCEP ARPEGE-VR

CDFOBSNCEP ARPEGE-VR

Jan to Dec Jan to Dec Jan to Dec

0 to 1 0 to 1 0 to 1

ARIEGE (Foix)

ARIEGE (Foix)

LOIRE(Blois)

LOIRE (Blois)

SEINE (Poses)

SEINE (Poses)

VIENNE (Ingrandes

0

2500

000

0 0

1200

2500250

150 800

20101960

500

0

Statistical downscaling: Validation: Summary

Predictors Strong link with regional climate Simulated correctly by model

Predictors encompass completely the climate change signal

Need to use Temperature as a predictor

Watersheds flows are correctly reproduced Annual Cycle Annual Variability Cumulative Density Function


Application: Impact of climate change on France watersheds

Multi-Model relative change ofDownscaled Precip. (%),

2046/2065

WINTER: DJF

Black-circled: at least 85% models has sign agreement

Multi-Model relative change of watershed

Flows (%), 2046/2065

Dispersion:Spatial Mean

σ = 18%

QuantifyingUncertainties

Ouranos, 20 May 2008Ouranos, 20 May 2008


Relative change precipitation2046/2065 vs 1970/1999 in Winter


DynamicalQuantile-Quantile

downscaling


-0.6 +0.6


Relative change watershed flows2046/2065 vs 1970/1999 in Winter



downscaling


-0.5 +0.5

Summary - 1

• Statistical downscaling methodology

• Validation is very good

• Hypothesis of stationarity (regression)

• Weather Typing Approach

• Low CPU demand

• Evaluate uncertainties with many scenarios

• Uncertainties of downscaling method are limited

• Those of numerical models are, in general, greater


Summary - 2

• Ensemble Mean of Watershed flows

• Decreases moderately in Winter (except Alps and SE Coast)

• 2050 : important decrease in Summer & Autumn

• Robust results, low uncertainty

• Strong increase of Low Water days

• Heavy flows decrease much less than overall mean


Down the Road…

• Whole Code Re-Engineering

• Modular approach

• Implement several statistical methodologies

• Configurable

• End-user parameters

• Core parameters

• Web Portal

• Climate-Change Spaghetti to Climate-Change Distribution

• Probability Density Function

• Re-sampled Ensemble Realisations

• M. Dettinger, U.S. Geological Survey (2004)Ouranos, 20 May 2008Ouranos, 20 May 2008 2828

Merci de votre attention!

Christian Pagé, CERFACSChristian Pagé, [email protected]

Julien Boé, CERFACSJulien Boé, CERFACSLaurent Terray, CERFACSLaurent Terray, CERFACS

Florence Habets, UMR SisypheFlorence Habets, UMR SisypheÉric Martin, CNRM, Météo-FranceÉric Martin, CNRM, Météo-France


Application: France watersheds: Snow Cover

• Water Equivalent (mm) of Snow Cover• Pyrenees• 2055• Grayed zones: min/max

FuturePresent

Aug Aug

AugAug

Jul Jul

JulJul

5 30

500250

3030

Application: France watersheds: Uncertainties

Winter Weather Type occurrence changes IPCC (2081/2100 - 1961/2000)

-20

-15

-10

-5

0

5

10

15

20

1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5

Models

Num

ber o

f day

s in

win

ter

Atl. Ridge Blocking NAO+ NAO-

~0+

+ -

-0.5 +0.5

3131

20 days

-20 days Models

Atlantic Ridge

NAO+ NAO-

Blocking

Correlation Weather Type Occurrence Precipitation


Relative change watershed flows2046/2065 vs 1970/1999 Perturbation method

WinterCorr 0.92

SpringCorr 0.38

SummerCorr 0.86

AutumnCorr 0.72

3232

-0.5 +0.5


Relative change precipitation2046/2065 vs 1970/1999 in Summer



downscaling


-0.6 +0.6


Relative change watershed flows2046/2065 vs 1970/1999 in Summer



downscaling


-0.5 +0.5

Régimes de temps et hydrologie (H1)

Domaine classification MSLP (D1)

* 310 stations pour les précipitations

• Définition de régimes/types de temps discriminants pour les précipitations en France

• Variable de circulation de grande échelle: Pression (MSLP), provenant du projet EMULATE (1850-2000, journalier, 5°x5°), précipitations SQR (Météo-France)

• Classification multi-variée Précipitations & MSLP, pas de temps journalier, espace EOF. On conserve ensuite uniquement la partie MSLP pour définir les types de temps.

8 à 10 régimes de temps !

Impact of climate change on France watersheds in 2050 : A comparison of dynamical and multivariate statistical methodologies By : Christian Pagé, CERFACS.

Documents

problematic of downscaling

regional climate model

global climate model

impact model

model present model

meteorological forcings

statistical approach

numerical model