GPC CPTEC: Seasonal forecast activities update Caio Coelho caio.coelho@cptec.inpe.br CPTEC/INPE Presentation plan Brazilian multi-model seasonal forecast system EUROBRISA multi-model system Regionally downscaled forecast products Final remarks Joint WGSIP/ET-OPSLS meeting Exeter, UK, 10 March 2014
Brazilian multi-model seasonal forecast system CPTEC/AGCM: Boundary conditions: persisted and predicted SST anomalies Convective parameterization schemes: Kuo, RAS, Grell 90 ensemble members CPT/IRI: Predictors (SST, Geopotential 500hPa, vertical velocity 850 hPa) Time series models: Holt-Winters Arima ECHAM4.6/AGCM: Boundary conditions: persisted SST anomalies 20 ensemble members All models produced retrospective forecasts (hindcasts) for the period 1989-2008 using exactly the same procedure used in real-time forecasting Joint effort: CPTEC/INPE, INMET and FUNCEME
Multi-model ensemble mean forecast CPTEC/INMET/FUNCEME Multi-model ensemble mean forecast JJA precipitation anomaly forecast skill: 1989-2008 correlation Precipitation anomaly forecast for JJA 2013 (mm) Final probabilistic forecast derived from forecast anomaly and retrospective correlation skill Issued in May 2013
Precipitation anomaly forecast skill: CPTEC/INMET/FUNCEME Multi-model probabilistic forecast Precipitation anomaly forecast skill: 1989-2008 correlation Regression based probabilistic forecast obtained from forecast anomaly and past performance Issued: May 2013 Forecast probabilities determined objectively
… task 5.3.1 http://eurobrisa.cptec.inpe.br EUROBRISA Integrated (empirical-dynamical combined and calibrated) precipitation seasonal forecasting system for South America Collaborative effort: INPE/CPTEC, Univ. Exeter, ECMWF, UK Met Office, Météo-France, UFPR, USP and INMET Previously supported by: Currently supported by: http://eurobrisa.cptec.inpe.br Key idea: Why not combining all available state-of-the-art forecast information for improving seasonal forecasts in South America - a region where the forecasts have skill and useful value
Updated EUROBRISA integrated forecasting system for South America Combined and calibrated coupled + empirical precip. forecasts Hybrid multi-model probabilistic system (NEW) Integrated forecast (NEW) Updated empirical model (NEW) Predictors: Atlantic and Pacific SST Predictand: Precipitation Coelho et al. (2006) J. Climate, 19, 3704-3721 Produced with forecast assimilation Stephenson et al (2005) Tellus A . Vol. 57, 253-264 Hindcast period: 1981-2010 Implemented in Jul 2013 Both forecast and verification products available at http://eurobrisa.cptec.inpe.br
Probabilistic forecast verification products for updated EUROBRISA system How well calibrated are EUROBRISA forecasts? How well do EUROBRISA forecasts detect below normal category events?
Prob. of most likely precipitation tercile (%) EUROBRISA integrated fcst for SON 2013 issued in August Empirical ECMWF UKMO Obs. SST anomaly Jul 2013 Meteo-France Integrated Prob. of most likely precipitation tercile (%) Issued: Aug 2013
BRAMS MAM 1989 seasonal precipitation forecasts for South America BRAMS: Brazilian Developments on the Regional Atmospheric Modelling System GPCP MAM 1989 Total precipitation (mm) CPTEC AGCM MAM 1989 Total precipitation (mm) BRAMS MAM 1989 Total precipitation (mm) Issued in previous Feb T62L28 (~ 200km x 200km) Persisted SST of previous Jan Issued in previous Feb 30km x 30km Persisted SST of previous Jan
CPTEC ETA model seasonal forecast for FMA 2013 issued in January 40km x 40km Persisted SST of previous Dec
Final remarks CPTEC GPC forecast products, Brazilian Multi-model forecast system (CPTEC/INMET/FUNCEME), EUROBRISA and ETA forecast products routinely produced and used in national and regional climate outlook forums: demonstrated national and regional cooperation in providing access to forecast products Currently testing new version of CPTEC AGCM with new physics and dynamics Start investigating potential gain in predictive skill by increasing spatial forecast resolution when downscaling CPTEC AGCM forecasts with BRAMS
Thank you for your attention !