Avaliação da previsão numérica sazonal de precipitação para o Rio Grande do Sul

• Autor(a) principal: Gonçalves, Jéssica Stobienia
• Data de Publicação: 2015
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Manancial - Repositório Digital da UFSM
• dARK ID: ark:/26339/001300000c3h7
• Texto Completo: http://repositorio.ufsm.br/handle/1/10281
• topic: Previsão climática; Precipitação; Parametrização cumulus; Climate forecasting; Precipitation; Cumulus parameterization; CNPQ::CIENCIAS EXATAS E DA TERRA::GEOCIENCIAS::METEOROLOGIA

In order to obtain an increase of quality seasonal climate forecast of precipitation, for the state of Rio Grande do Sul, were implemented and evaluated nine types of simulations that uses different cumulus parameterization schemes available in the Regional Climate Model version 4 RegCM4.The tested parameterizations were Grell with closure Arakawa and Schubert - AS and Fritsch and Chappell - FC, MIT-Emanuel and mixed convection that is the use of different convection schemes on the land and the sea. The evaluation method consisted of analysis qualitative and quantitative statistics of seasonal precipitation climate forecasts of five regions of Rio Grande do Sul, from August 2013 to August 2014.The statistics applied were Taylor diagram, random and systematic error analysis, concordance index and contingency table. The forecasts were evaluated using observed data from meteorological stations of the Instituto Nacional de Meteorogia (INMET). The analysis showed the RegCM4 had higher correlations and lower errors compared to the Global Model. The best results were observed in the northern and western part of the state with the parameterizations Grell FC, Grell AS and the combination of Emanuel simulated ocean and Grell AS on land. Although some regions were not adequately represented by the Regional Climate Model RegCM4, yet it performed well reducing the overestimation of precipitation observed in the simulation of the Global Model and improved temporal distribution of the same.