Use of Correlation Techniques in Automatic Detection of Thermal Plumes in the Air Layer Adjacent to Natural Water Waves.
NAVAL POSTGRADUATE SCHOOL MONTEREY CALIF
Pagination or Media Count:
Pattern recognition of temperature fluctuations, representing plumes, occurring in data recorded in the atmospheric boundary layer was performed utilizing predefined characteristics of these phenomena. An averaged, normalized plume was computed based on the characteristics of ten previously identified plumes. This standard plume consisted of two components or curves and a component-by-component correlation was made of this standard plume with various data sets, each set containing temperature information recorded at two different heights. A product curve was then formed from the two correlation curves and used to identify plumes automatically. A plume was said to have been detected if the product curve exceeded a previously selected threshold. The accuracy of this detection method was determined by visual inspection of the same records by a meteorologist. Approximately 73 per cent of all plume occurrences in the data sets were detected which indicates the appropriateness of this method of pattern recognition for plume detection. Author
- Atmospheric Physics