From DEMETER to EUROSIP The European Seasonal to

From DEMETER to EURO-SIP: The European Seasonal to Inter-annual Prediction project Richard Graham acknowledgements to DEMETER and EURO-SIP partners (Francisco Doblas-Reyes, Laura Ferranti (ECMWF)) NOAA 30 th Climate Annual Climate Diagnostics and Prediction Workshop, October 24 -28 th, State College, Pennsylvania © Crown copyright 2004 Page 1

Talk Outline § DEMETER project (EU FP 5) – completed September 2003 §European seasonal multi-model system (hindcasts) §Application/impact models (crops, health) §forecast system sensitivity studies § EURO-SIP § operational real-time implementation of DEMETER § ENSEMBLES (EU FP 6) § prediction/impacts seasonal to decadal (and longer) timescales § new representations of model and initial condition uncertainty § 5 year project, now one year in § Met Office European winter forecast 2005/06 © Crown copyright 2004 Page 2

The DEMETER multi-model ensemble system § DEMETER system: 7 coupled global circulation models § 6 installed at ECMWF § 9 member ensembles § § ERA-40 initial conditions SST and wind perturbations 4 start dates per year 6 months hindcasts (Glo. Sea) § Hindcast periods: § 1980 -2001 (22 yrs) – all models § © 1959 -2001 (43 yrs) – 3 models Crown copyright 2004 Page 3

Multi-model ensemble system: hindcasts DEMETER system: 7 coupled global circulation models CNRM (FR) ECMWF (INT) INGV (IT) LODYC (FR) MPI (DE) UKMO (UK) CERFACS (FR) 7 models x 9 ensemble members 63 member multi-model ensemble Feb 87 May 87 © Crown copyright 2004 Aug 87 Nov 87 Feb 88. . . Page 4

Multi-model ensemble system: hindcasts DEMETER system: 7 coupled global circulation models CNRM (FR) ECMWF (INT) INGV (IT) LODYC (FR) MPI (DE) UKMO (UK) CERFACS (FR) Feb 87 May 87 © Crown copyright 2004 Aug 87 Nov 87 Feb 88. . . Page 5

Multi-model ensemble system: hindcasts DEMETER system: 7 coupled global circulation models CNRM (FR) ECMWF (INT) INGV (IT) LODYC (FR) MPI (DE) UKMO (UK) CERFACS (FR) Feb 87 May 87 © Crown copyright 2004 Aug 87 Nov 87 Feb 88. . . Page 6

Multi-model ensemble system: hindcasts 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 Feb 87 May 87 © Crown copyright 2004 Aug 87 Nov 87 Feb 88. . . Page 7

Results DEMETER Special Issue 2005 § Evaluation of DEMETER multimodel hindcast skill § Applications/impact models § Crop yield § malaria § Downscaling studies § Statistical & dynamical § Sensitivity studies § e. g. CGCM Vs AGCM Tell , A 7 us 5 © Crown copyright 2004 1 2 , 3 No. s n o i t u b i r cont Page 8

Single vs. multi-model: Brier skill score ‘basket’ of regions, start dates and leadtimes Brier skill score Reliability skill score=1 -Rel/BSc © Crown copyright 2004 Resolution skill score=Res/BSc Page 9

Impact of number of models on multi-model skill: Tropics, JJA precip, 1 m lead: RPSS Multi-model realizations Single-model realizations © Crown copyright 2004 Page 10

Application models: wheat yield predictions for Europe 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 ERA 40 data (red dots). France Denmark Germany SIMULATION WEIGHTED YIELD ERROR (%) ± STANDARD ERROR JRC February 7. 1 ± 0. 9 JRC April 7. 7 ± 0. 5 JRC June 7. 0 ± 0. 6 Greece JRC August 5. 4 ± 0. 5 DEMETER (Feb. start) 6. 0 ± 0. 4 From P. Cantelaube and J. -M. Terres, JRC © Crown copyright 2004 Page 11

Summary § § DEMETER: proof of multi-model benefit using 7 coupled ocean-atmosphere models More than a multi-model exercise…encouraging results for integration with application models Mutual understanding between prediction and application communities – ‘end-to-end’ approach Infrastructure and standardisation developed § § § coordination of model output content (with application partners) model output formats and units (archiving at ECMWF). downscaling techniques/experience centralised model diagnosis/verification system at ECMWF data download and plotting facility http: //data. ecmwf. int/data/ © Crown copyright 2004 Page 12

EURO-SIP real-time multi-model ensemble § Three models running at ECMWF: § Met Office: Glo. Sea (version of Had. CM 3) (41 -ensemble) § ECMWF: system 2 (40 -ensemble) § Météo-France – Arpège (41 -ensemble) § Unified system with common development strategy § Products in development: § 2 -model products available on Met Office website § ECMWF release of web products is expected soon © Crown copyright 2004 Page 13

Probability of precipitation – outer-quintile categories, GHACOF region, SON 2005 Glo. Sea EURO-SIP (80 ensemble) P(well-above) Yellow/red: enhanced chance of category (relative to climatology) Blue: diminished chance White: near climatological probabilities P(well-below) based on 1987 -2001 climatology © Crown copyright 2004 Page 14

Probability of lower temperature tercile category: NDJ 2005 (from September) Met Office (Glo. Sea) ECMWF EURO-SIP multi-model Meteo-France Forecast is made available on the 15 th of each month. © Crown copyright 2004 Page 15

Tropical Storms Forecasts starting on 1 st June 2005: JASON ECMWF Met Office Obs 1 st May- mid-Sep Météo-France Multimodel W-Pac E-Pac Atl Forecast No. = 18 Climatology = 11. 6 © Crown copyright 2004 Page 16

After DEMETER…ENACT: enhanced ocean assimilation EN 2 QC’d dataset and objective analysis, 1956 -2004 http: //www. hadobs. org © Crown copyright 2004 Page 17

ENSEMBLES: seasonal to decadal & improved representing model uncertainties 1. 7 model multi-model installed at ECMWF 2. - includes Met Office decadal forecast system (anomaly assimilation) 3. - realistic GHGs, aerosols Met Office decadal forecast (De. Pre. Sys) © Crown copyright 2004 Page 18

ENSEMBLES: perturbed physics 2. Many models from perturbing physics parameters, eg. RHcrit for cloud formation (ref Murphy et al, 2004; and climateprediction. net) From Murphy et al, Nature 2004 © Crown copyright 2004 Page 19

ENSEMBLES: stochastic physics 3. Stochastic ‘backscatter’ experiment (ECMWF) With CASBS Without CASBS E(n) n= spherical harmonic order missing energy Stochastic physics experiment with ECMWF system: Cellular Automaton Stochastic Backscatter Scheme (CASBS) © Crown copyright 2004 From Tim Palmer, ECMWF Page 20

Summary and plans § EURO-SIP: Real-time follow up to DEMETER § products becoming available § Meteo-France model fully implemented early 2006 § plans for model upgrades, convergence on the periods and ensemble size of hindcasts § ENACT: § enhanced ocean data assimilation – ocean ‘reanalyses’ § QC’d ocean data and objective analyses § ENSEMBLES: § § builds on DEMETER and ENACT seasonal to decadal timescales (and longer) further investigations of model uncertainties calibrating and combining model ensembles © Crown copyright 2004 Page 21

Met Office forecast: European winter 2005/06 Empirical NAO prediction Rodwell & Folland, 2002 NAO index: difference between normalised pressure anomaly Azores - Iceland Predicts correct sign in 2 out of 3 winters N. Atlantic May SST anomaly associated with +ve NAO N. Atlantic May SST anomaly 2005 © Crown copyright 2004 Page 22

September surface and sub-surface anomalies and Glo. Sea prediction SST anomalies actual normalised Glo. Sea prediction from September 60 -90 m anomalies © Crown copyright 2004 Page 23

Thank you! additional slides follow © Crown copyright 2004 Page 24

EURO-SIP real-time multi-model ensemble § Met Office (Glo. Sea 2) § § § Modified Had. CM 3 Atmos. 2. 5° x 3. 75° x 19 L Ocean (1. 25° to 0. 3°) x 1. 25° x 40 L 41 ensemble members (from mix of windstress and SST perturbations) Hindcast: 1987 -2001, 15 members § ECMWF (system 2) § § Atmos. TL 95 L 40 Ocean (1. 0° to 0. 3°) x 1. 0° x 29 L 40 ensemble members Hindcast: 1987 -2001, 5 members § Meteo-France § § Atmos. Arpège TL 63 L 31 Ocean: ORCA (2. 0° to 0. 5°) 31 L 41 ensemble members (mix of 5 ocean analyses and 8 lagged atmos) Hindcast: 1993 -2003, 5 members © Crown copyright 2004 Page 25

Parameter Perturbations Large Scale Cloud • Ice fall speed • Critical relative humidity formation • Cloud droplet to rain: conversion rate and threshold • Cloud fraction calculation Boundary layer • Turbulent mixing coefficients: stability-dependence, neutral mixing length • Roughness length over sea: Charnock constant, free convective value Convection • Entrainment rate • Intensity of mass flux • Shape of cloud (anvils) • Cloud water seen by radiation Radiation • Ice particle size/shape • Sulphur cycle • Water vapour continuum absorption Sea ice • Albedo dependence on temperature • Ocean-ice heat transfer © Crown copyright 2004 Dynamics • Diffusion: order and e-folding time • Gravity wave drag: surface and trapped lee wave constants • Gravity wave drag start level Land surface processes • Root depths • Forest roughness lengths • Surface-canopy coupling • CO 2 dependence of stomatal conductance Page 26

Met Office seasonal/multi-annual/decadal runs *Later Had. GEM Model Glo. Sea (Had. CM 3*) De. Pre. Sys (Had. CM 3) Current oper. range Seasonal (6 months) decadal Hindcasts pre-production assimilation method Conventional (OI type) calibrated anomalies pert. ODA 9 -ensemble experiments 1991 -2001 pert. phys. Glo. Sea: ->7 m: 1 st/15 th May/Nov 1 st June/Dec ->14 m: 1 st May/June/Nov/Dec -> 10 y: 1 st May 1964, 1994 pert. phys. De. Pre. Sys: lagged avge © Crown copyright 2004 period: 1991 - 2001 lagged avge -> 10 y: 1 st May/Nov (all years) Page 27

Multi-model evaluation: Niño 3. 4 SST anomaly: ASO at 3 m lead: Correlation: Multi-model = 0. 83 Best single model = 0. 81 http: //www. ecmwf. int/research/demeter/d/charts/verification/ © Crown copyright 2004 Page 28

Glo. Sea development and response to North Atlantic SST ‘tri-pole’ pattern: spring 1989 MAM 1989 Jan 89 Had. AM 3 (+ pers Jan SSTA) Feb 89 Glo. Sea May 89 ERA 40 © Crown copyright 2004 Had. ISST Glo. Sea Page 29

ENSEMBLES: seasonal to interannual timescales (and longer); representing model uncertainties § ENSEMBLES system: 7 coupled GCMs running at ECMWF § upgraded models from DEMETER (eg. using ENACT developments etc) § 9 -member ensembles § ERA-40 atmosphere and soil initial conditions § ENACT-based ocean initial conditions with SST and wind perturbations § 2 seasonal (7 months), 1 annual (12 -14 months) runs per year § Two multi-annual runs (1965, 1994) except 2 per year for De. Pre. Sys) § Realistic boundary forcings: GHGs, aerosols, solar forcing, etc. Crown copyright 2004 § ©Hindcast pre-production period: 1991 -2001; later 1960 -2001 Page 30

Sensitivity studies: AGCM Vs CGCM Nov. start for FMA (3 -month lead) ROC Skill for 2 m temperature, upper tercile Had. AM 3 (2 -tier) Glo. Sea (1 -tier) Much improved Glo. Sea skill in tropical (south) Atlantic and Indian ocean, from representation of lagged responses to ENSO. © Crown copyright 2004 Page 31

Met Office decadal forecast system (De. Pre. Sys): Forecast of global annual mean Ts §Predict 0. 36+/-0. 23 o C (5 -95%) warming for the coming decade §Unless a major volcanic eruption occurs! §Uses ‘anomaly assimilation’ (seasonal-range skill very competitive!) © Crown copyright 2004 Page 32