Development and evaluation of Global Eta Framework (GEF) model at medium and seasonal ranges

• Ano de defesa: 2018
• Autor(a) principal: Dragan Latinovic
• Orientador(a): Chou Sin Chan
• Banca de defesa: Silvio Nilo Figueroa Rivero, Fedor Mesinger, Miodrag Rancic
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Instituto Nacional de Pesquisas Espaciais (INPE)
• Programa de Pós-Graduação: Programa de Pós-Graduação do INPE em Meteorologia
• País: BR
• Link de acesso: http://urlib.net/sid.inpe.br/mtc-m21c/2018/03.15.18.01

The objective of this research is to implement the Global Eta Framework (GEF) model and evaluate its performance in medium and seasonal range integrations. GEF is a global atmospheric model on cubed-sphere grid, constructed as a combination of the technique of quasi-uniform gridding of the sphere and the numerical structure of the regional Eta model. Six regional models, interconnected through the cubed-sphere framework are integrated simultaneously, one on each side of the cube, to provide a global coverage and to create unique globalized version of the regional Eta model. Two model setups of the model are developed and configured, one at 25-km horizontal resolution for seasonal range runs and another at 8-km horizontal resolution for medium-range runs. Total of 10 seasonal integrations of approximately 4 months are performed creating 5 member ensembles for the period September- November (SON) of 2011 and 2013 with the objective to evaluate the model skill to simulate the onset of the rainy season in Western-Central Brazil (WCB). Comparative assessment of daily means of global model output fields against appropriate reanalysis and observations for the austral spring indicates high level of agreement, both in spatial distribution and intensity for most of the variables. The lowest skill is shown for precipitation which is overestimated over some tropical oceanic regions and underestimated over tropical continental regions, including South America. Other fields, evaluated at different levels include 200-hPa wind from upper-troposphere, 500-hPa geopotential height from mid-troposphere, 850-hPa temperature and wind representing lowertroposphere and mean sea level pressure (MSLP) at the surface. The onset of the rainy season is determined by using methods based on precipitation and outgoing long-wave radiation (OLR). Comparison of 5-day averaged values (pentads) of precipitation and OLR of the ensemble members and the ensemble mean against observed data shows the ability of GEF to reproduce the typical pattern of transition from dry to wet season in WCB almost in the same pentad determined by both methods. However, most of ensemble members tend to underestimate precipitation and overestimate OLR. The other set of integrations viii is performed at 8-km horizontal resolution, for the length of 10 days. The total of 22 integrations is performed with the lead times of 24 h, 48 h and 72 h to analyze 8 selected cases of extreme rainfall over the city of Manaus, Amazonas. The model demonstrates the ability to simulate well the large-scale patterns for the period of up to 7 days. The Equitable Threat Score (ETS) for the time ranges of 36 h and 60 h shows that the model simulates well the areas with precipitation, while for the higher thresholds, the results are on the lower limit of referential values. The BIAS score (BIAS) shows almost perfect score for the lowest rain-no rain threshold, and decreases rapidly for the other thresholds. The areas with precipitation over South America are well simulated by the model, with comparable intensity in extratropical regions and mostly underestimated values in the tropical regions. The model simulated rain for Manaus in almost every simulation, with underestimated values in most of the cases, in addition to showing the low skill in the simulation of the positions of the precipitation maxima in Central Amazon. The computational efficiency of GEF and the results presented in this study show that continuous efforts in the development of the GEF model can give significant contribution to the improvement of medium-range and seasonal forecasts at CPTEC. Improvements mostly related to the parameterization of convection and radiation, and the ongoing development of nonhydrostatic version should position GEF as a good candidate for future unified model of the centre, capable of running across a range of scales.