Método para detecção automática de vórtices
Nenhuma Miniatura disponível
Data
2016-10-20
Autores
Freguete, Larissa Marques
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Federal do Espírito Santo
Resumo
In the ocean, the vortices are related to western boundary currents, as the Brazil Current, that are confessedly the currents with the highest intensity. These currents have a quite vortex generation potential either being caused by barotropic or baroclinic instabilities. Hence, the goal of this work was the development of a vortex automatic detection technique by using the 3D velocity fields from de computational model HYCOM (HYbrid Coordinate Ocean Model). Without a lack of generality, the study region is the area delimited by the latitudes of 19oS - 26oS and the longitudes of 35oO - 50oO as being a site of big occurrence of vortices. The HYCOM velocity field is analyzed by layers and the localization of a existing singularity in the de velocity field is accomplished by the Poincaré Index (IP). The type of singularity is identified by using a linear dynamic model that takes in consideration the singularity’s neighborhood informations. In this work are presented three different techniques to quantify the IP. The first technique uses only the orientation field values from the velocity field, the second uses a line integral in a closed curve, and the third one uses a surface intengral. A numeric code was built and the three techniques compared by using a diagnostic assessment method. The techniques performances was evaluated considering a HYCOM velocity database where the occurring singularities was labelled manually. The measurement of the techniques performance is based on the sensibility (REC) and precision (PRE) metrics. The metrics results show the technique 3 has the best performance to identify the vortices (PRE = 73.96%, REC = 96.01%). A quasi-stationary was identified in the Cabo de São Tomé area. The vortices presented a well-developed vertical profile that reaches depths higher than 2000 m and a time length of 11 days.
Descrição
Palavras-chave
automatic detection , vortices , singularities classification , Poincaré Index , detecção automática , vórtices , classificação de singularidades , Índice de Poincaré , HYCOM , Singularidades (matemática) , Reconhecimento de padrõe