recent developments in dynamical climate seasonal forecasting

CLIMAG Meeting Geneva, 11 May 2005

Recent Developments in Dynamical Climate Seasonal

Forecasting

Francisco J. Doblas-Reyes, Renate Hagedorn, Tim N. [email protected]

European Centre for Medium-Range Weather Forecasts


CLIMAG objective

“To utilize the ability to predict climate variability on the scale of months to a year to improve management and decision making in respect to crop production at local, regional, and national scales.”


CLIMAG objective

“To utilize the ability to predict climate variability on the scale of months to a year to improve management and decision making in respect to crop production at local, regional, and national scales.”

Requirements by the end user:

• predict climate variability: skilfully deal with uncertainties in climate prediction

• seasonal-to-interannual time scales: coupled ocean-atmosphere general circulation models

• variable spatial scale: downscaling


• Research project funded by the Vth FP of the EC, with 11 partners.

• Integrated multi-model ensemble prediction system for seasonal time scales.

• More than a multi-model exercise: seasonal hindcasts used to assess the skill, reliability and value of end-user predictions.

• Applications in crop yield and tropical infectious disease forecasting.

• Officially finished in September 2003, but with an operational follow up.

End-to-end: DEMETER

http://www.ecmwf.int/research/demeter/


Multi-model ensemble approach

Uncertainty

initial conditionsmodel formulation

Estimation

ensemblemulti-model

multi-model ensemble forecast multi-model ensemble forecast systemsystem

N models x M ensemble membersN models x M ensemble members


Multi-model ensemble system

• DEMETER system: 7 coupled global circulation models

• Hindcast production for: 1980-2001 (1958-2001)

9 member ensembles

ERA-40 initial conditions

SST and wind perturbations

4 start dates per year

6 months hindcasts

Partner Atmosphere Ocean

ECMWF IFS HOPE

LODYC IFS OPA

CNRM ARPEGE OPA

CERFACS ARPEGE OPA

INGV ECHAM-4 OPA

MPI ECHAM-5 MPI-OM1

UKMO HadCM3 HadCM3



Feb 87 May 87 Aug 87 Nov 87 Feb 88 ...

7 models x 9 ensemble members

63 member multi-model ensemble

DEMETER system: 7 coupled global circulation models

CNRM (FR) ECMWF (INT) INGV (IT) LODYC (FR) MPI (DE) UKMO (UK) CERFACS (FR)




63 member multi-model ensemble

= 1 hindcast




Forecast quality assessment

Forecast quality assessment is a basic component of the prediction process

Information about the quality and the uncertainty of the predictions is as important as the prediction

itself


ENSO predictionsMulti-model seasonal (MAM) predictions for Niño3.4 SSTs


River basin predictionsMulti-model predictions of precipitation over river basins and many other

verification diagnosticshttp://www.ecmwf.int/research/demeter/d/charts/verification/


DEMETER end-to-end methodolgy

63………… 624321Seasonal forecast

………… 63624321 Downscaling

63………… 624321Application

model

0

Probability of Precipitation Probability of Future Crop Yield

0

non-linear transformation


Downscaling for s2d predictions

• Use dynamical and empirical/statistical methods.

• Correct systematic errors of global models and obtain reliable (statistical properties similar to the observed data) probabilistic predictions (with only relatively short, i.e., 15-30 years, training samples).

• Deal with full ensembles, not a deterministic prediction or the ensemble mean, maximising the benefit of limited simulations with regional models.

• Consider model and initial condition uncertainty.

• Generate high-resolution (e.g., daily) time series of surface variables (using, e.g., weather generators with statistical methods).


http://www.ecmwf.int/research/EU-projects/ENSEMBLES/news/index.html

Downscaling for s2d predictions


France Germany

Denmark Greece

Wheat yield predictions for Europe

From P. Cantelaube and J.-M. Terres, JRC

SIMULATION WEIGHTED YIELD ERROR (%)

± STANDARD ERROR

JRC February 7.1 ± 0.9

JRC April 7.7 ± 0.5

JRC June 7.0 ± 0.6

JRC August 5.4 ± 0.5

DEMETER (Feb. start)

6.0 ± 0.4

DEMETER multi-model predictions (7 models, 63 members, Feb starts) of average wheat yield for four European countries (box-

and-whiskers) compared to Eurostat official yields (black horizontal lines) and crop results from a simulation forced with downscaled

ERA40 data (red dots).


DEMETER Special Issue 2005

Tellus 57A, No. 3, 21 contributions


ECMWF public data server

A service that gives researchers immediate and free access to datasets hosted at ECMWF

• DEMETER• ERA-40• ERA-15• ENACT- Monthly and daily data- Select area- GRIB or NetCDF- Plotting facility

http://data.ecmwf.int/data/


Future developments

• Integration of weather and climate predictions at different time scales.

• Interaction between different climate-related end-user systems. User-oriented verification.

• Optimisation of the a-posteriori multi-model information through single-model weighting depending on past performance.

• Anthropogenic impact on seasonal climate predictions.

• The ENSEMBLES project: probabilistic climate prediction at seasonal, interannual and longer time scales.


1) Prediction of different time scales

Probabilistic seamless forecast system at ECMWF:

1-10 days: medium range EPS (TL399L60)

10 days-1 month: monthly forecast system (TL255L60)

1 month-12 months: seasonal forecast system (TL159L40)

10d

1mth

12mth

01/01 01/02 01/0315/01 29/01 12/02 26/02


2) Interacting factors: tropical malaria• Tropical disease incidence is a major factor

affecting food security in tropical/semi-arid areas (socio-economic interaction).

• The following example deals with uncertainty in malaria prediction using a probabilistic approach to reduce forecast error and can easily be extended to prediction of climate-related yields (uncertainty).

• The predictions are designed to be included in an early warning system (decision making).

• Seasonal prediction allows users to become familiar with the use of climate information and understand methods to mitigate the impact of and adapt to future global change (climate change).


2) Malaria warning: seasonal predictionRelationship between DJF CMAP precipitation and

Botswana standardised log malaria incidence for 1982-2002


-- high malaria years

-- low malaria years

2) Malaria warning: seasonal predictionProbabilistic predictions of standardised malaria incidence

in Botswana five months in advance of the epidemic

Very low malaria

Very high malaria

0.840.940.52Very high

1.001.000.95Very low

DEMETERCMAPDEMETEREvent

IncidencePrecipitationROC Score

0.840.940.52Very high

1.001.000.95Very low

DEMETERCMAPDEMETEREvent

IncidencePrecipitationROC Score

Available in November

Available in March

CLIMAG Meeting Geneva, 11 May 2005From Coelho et al. (2005)

3) Calibrated downscaled predictionsPAGE agricultural extent

PAGE agroclimatic zones


Northern box

Forecast Correlation

BSS

Multi-model

0.57 0.12

Forecast Assimilation

0.74 0.32

3) Calibrated downscaled predictions

From Coelho et al. (2005)

Southern box

Forecast Correlation

BSS

Multi-model

0.62 0.16

Forecast Assimilation

0.63 0.28


Constant GHGCorrelation = 0.52

4) Anthropogenic effect: T2m predictions

Variable GHGCorrelation = 0.77

1-month lead, summer (JJA) predictions of global T2m


5) The future: ENSEMBLES project

• Integrated Project funded by the EC within the VIth FP, 69 partners.

• Start date: 1 September 2004, Duration: 5 years

• Integrated probabilistic prediction system for time scales from seasons to decades, and beyond.

• Seasonal-to-decadal hindcasts will be used to assess the reliability of forecast systems used for scenario runs.

• Comparison of the benefits of the multi-model, perturbed parameters and stochastic physics approaches to assess forecast uncertainty.

• Great diversity of applications: health, crop yield, energy production, river streamflow, etc.


Summary

• The multi-model has proven to be an effective approach to reduce forecast error by tackling both initial condition and model uncertainty.

• The end-to-end approach has shown promising results in seasonal forecasting.

• There is a clear need to link the research and development carried out about climate variability at different time scales.

• Seasonal-to-interannual forecasting can evolve into a field where end-users learn to use (and verify) climate information before developing adaptation/mitigation strategies for environmental global change.


Questions?


Generalized ensemble approach

Uncertainty


Estimation

ensemble perturbed parameters

perturbed parameters ensembleperturbed parameters ensemble

N versions x M ensemble membersN versions x M ensemble members


Generalized ensemble approach

Uncertainty


Estimation

ensemble with stochastic physics

Ensemble with stochastic physicsEnsemble with stochastic physics

M ensemble membersM ensemble members


Multi-model benefits: Reliability

0.0390.8990.141

BSSRel-ScRes-Sc

0.0390.8990.140

0.0950.9260.169

-0.001 0.877 0.123

0.0650.9180.147

-0.064 0.838 0.099

0.0470.8930.153

0.2040.9900.213

Reliability for T2m>0, 1-month lead, May start, 1980-2001


River basin predictionsMulti-model predictions of precipitation over the Nile basin


JRC’s CGMS in DEMETER

Crop Growth Indicator

Jan Feb Aug

Meteo data

Yield

Statistical model

Meteo data ERA / DEMETER data


gathering cumulative evidence for early and focused response . . .

case surveillance alone = late warning

geographic/community focus

Malaria early warning systems


Malaria warning: seasonal predictionPrecipitation composites for the five years with the highest

(top row) and lowest (bottom row) standardised malaria incidence for NDJ DEMETER (left) and DJF CMAP (right)

Areas with

epidemic malaria


Bayesian procedure:

• Climate model ensembles give

• But we are interested in , not !!!

• Bayes’ theorem updates and obtain

Forecast assimilation

Bayes’ theorem

tXtY : Obs

: Forecasts

Bayes’ theorem

tXtY : Obs

: Forecasts

)X(p t

)X|Y(p tt)X(p t

)Y(p t)X|Y(p tt


ObservationsMulti-modelForecast

Assimilation

(mm/day)

r=0.51

r=0.28

r=0.97

r=0.82

Calibrated South American Precipitation

From Coelho (2005)

• 3 DEMETER coupled models

• 1-month lead time DJF precipitation

• ENSO composites for 1959-2001

• 16 warm events• 13 cold events


ENSEMBLES: General information• Integrated Project funded by the VI FP of the EC

• Integrated probabilistic prediction system for time scales from seasons to decades and beyond

• 69 partners

• Seasonal-to-decadal hindcasts will be used to assess the reliability of model systems used for climate change experiments

• Great diversity of climate applications

• 2 consultants @ ECMWF

• Start date: 1 September 2004, Duration: 5 years

• http://ensembles-eu.metoffice.com


OrganizationThe project is organized in ten Research

Themes (RT), ECMWF involvement in red:• RT0: Management

• RT1: Development of the EPS

• RT2A: Global model engine

• RT2B: Production of regional climate scenarios

• RT3: High resolution regional ensembles

• RT4: Analysis of processes

• RT5: Evaluation

• RT6: Assessment of impacts

• RT7: Scenarios and policy implications

• RT8: Dissemination and training

recent developments in dynamical climate seasonal forecasting

Documents