Evaluation of the Model MM5 Performance in an Extreme Precipitation Event and a Deeper Analysis of its Meteorological Causes

The predictability of adverse meteorological events is one of the most challenging tasks of numerical weather forecast. A good representation of these events by the numerical models is essentially important to the meteorological warnings and the decision making, by the authorities.

An extreme precipitation event occurred in the city of Pelotas, Rio Grande do Sul state, Brazil, in January, 2009, leading to rainfall accumulate greater than 600 mm in less than 12 hours, four times greater than the climatic mean of January for this region. Some synoptic features were identified by Bainy and Teixeira (2009), among which are highlighted: low level mass convergence, related to winds coming from the Ocean and the Amazonic region and the approach of an mid-tropospheric trough. Satellite and radar imagery allowed the identification of an cloud cluster in sub-synoptic scale, which remained during all the 28th afternoon.

Even though synoptic conditions over the period in which the situation occurred were favourable to severe weather, they do not completely explain the magnitude of the rainfall that affected the region of Pelotas. Beyond this fact, Almeida and others (2009) verified a sub-estimative of the accumulate precipitation of the numerical models forecasts and hidroestimators based in satellite and meteorological radar, whose estimative did not crossed the 111 mm.

Aiming to obtain some more information about the event which took place in Pelotas, in January, 2009, experiences are being done with the meteorological model MM5, with the following goals: to verify the impact of improving the spatial scale on the representation of accumulate rainfall during the event, to identify with more accuracy the structure and development of the phenomenon and to evaluate the hability of the MM5 model in reproducing the observed rainfall in function of different convection and cloud microphysics parametrization .

The model is being run with Final NCEP Analisys, which spatial resolution is 1º and time resolution is 6h. The simulation period extend from the 2009 January 27th (the day before the event) to the 2009 January 29th (day after the event). All the simulations are being run with two nested domains: the mother domain, which spatial resolution is 36 km, and dimensions that encompasses most of South America, and a second domain, which spatial resolution of 12 km, and dimensions that encompasses the most southern Brazil, where Pelotas is located. Statistical indexes like BIAS, False Alarm Ratio (FAR) and Threat Score (TS) will be used to evaluate objectively the representativeness of rainfall by the model. What is more, spatial correlation between the simulates and observed rainfall fields will be obtained, toghether with the evaluation of the terms of equations of atmospheric dynamics, such as vorticity equation, thermodynamic equation and moist conservation equation. Those terms will be helpful in the evaluation of the most important dynamic processes that lead to the occurrence of such extreme event.