Design of a high-resolution ensemble prediction system for Argentina

MATSUDO, CYNTHIA; GARCIA SKABAR, YANINA; DILLON, MARIA EUGENIA; HOBOUCHIAN, PAULA; RUIZ, JUAN; VIDAL, LUCIANO; SALIO, PAOLA

Simposio; WMO WWRP 4th International Symposium on Nowcasting and Very-short-range Forecast 2016; 2016

Organización Meteorológica Mundial

p { margin-bottom: 0.25cm; line-height: 120%; }Numeroushigh impact weather events (HIWEs) are observed in Argentina (e.g.intense rainfall, hailstorms and tornadoes, among others) affectingpeople?s lifestyles, socio-economic activities and the environment.Efforts to improve the prediction accuracy of such events are ofvital importance. In this regard, a deterministic high resolutionnumerical weather prediction scheme was implemented operationallyusing a convection-allowing WRF model with 4-km resolution. Twice aday 48-hour forecasts are available at the forecaster offices at theNational Meteorological Service of Argentina. Some objectiveverification metrics related to precipitation forecasts are presentedto assess the skill of the system.Asthe high performance computing resources have been increasing, mainforecast centres around the globe tend to incorporate ensembleprediction systems (EPSs) to address the uncertainty associated withhigh resolution-short range weather forecasts. Given the fact thatforecast uncertainty is highly dependent on the meteorologicalsituation and model configuration it is important to implementmethods to estimate it. In the present work, preliminary resultsobtained with a high resolution EPS for Argentina are presented.Based on the experience and results derived from the deterministicscheme, the 4-km resolution WRF model is used to generate theensemble system. Different EPS versions are explored consideringdifferent physical parameterization schemes. Apreliminary evaluation of the EPS system is conducted using a casestudy from the 2015-2016 warm season in which HIWEs have beenreported. The EPS skill and performance are evaluated against surfaceconventional observations and remote sensing information. Resultswill be presented at the conference.