Evaluation of subseasonal forecast models for a strong heat wave over Southern South America

MARISOL OSMAN; MARIANO SEBASTIÁN ALVAREZ

Quingdao

Conferencia; CLIVAR Open Science Conference; 2016

World Climate Research Programme

The prediction skill of subseasonal forecast models is evaluated for a strong and long-lasting heatwave occurred in December 2013 over Southern South America. Reforecasts from two modelsparticipating in the WCRP/WWRP Subseasonal to Seasonal project, the Bureau of MeteorologyPoama and Beijing Climate Center model (CMA), were considered to evaluate their skill inforecasting temperature and circulation anomalies during that event. The Poama reforecast of 33member ensemble size, initialized every five days, and CMA reforecast of 4 members ensemblesize for each day were considered for the 1994-2013 period. CMA ensemble was constructedusing the forecast of the same date of Poama plus the previous 4 days resulting in a 20 ensemble-members forecast. Weekly forecasts were computed with leadtimes from 2 days up to 22 daysevery five days. For each model, 2 meter temperature, 200-hPa and 850-hPa geopotential heightand outgoing longwave radiation were used to describe the regional circulation. Daily ERA Interimreanalysis of 2 meter temperature and 200-hPa and 850-hPa geopotential height from 1994 to2013 were used as observations while outgoing longwave radiation was taken from NOAAinterpolated satellite estimates. Weekly anomalies were calculated for observations from 13 th ofDecember to 31 st of December 2013. Anomalies for both observations and reforecast werecalculated with respect to their own climatology.Results show that both models forecast lower temperature anomalies than those observed duringweek 1 and week 2 of the heat wave while Poama model overestimate temperature anomalies forweek 3. Poama model forecasts the first two weeks of the heat wave at long and short leadtimesbetter than CMA although week 3 anomalies were better forecasted by CMA, especially for longleadtimes. Strong positive circulation anomalies over Southern South America at high and lowlevels during week 1 and 2 were well forecasted by Poama more than 15 days in advanced whileCMA produced skillful forecast just two days prior to the beginning of the event. Week 3 negativecirculation anomalies were of the opposite sing in long lead forecast of both Poama and CMA.In summary, for this case, models skill in forecasting surface temperature resulted moderate atlong lead times. However, their performance for regional upper and low level circulation leads tothe question whether the used of these forecasted variables in combination to other dynamic andstatistic forecast tools could improve skill.