Water Mass Classification in the North Atlantic Using IR Digital Data and Bayesian Decision Theory,

Abstract

A method is described which utilizes Bayesian decision theory and historical statistics of sea surface temperature to classify surface water masses and ocean fronts from satellite-derived infrared data. Probabilities that certain features occur are determined from the normal distributions of specific statistical characteristics, known a priori, and the same characteristics computed from satellite data. The better the match between the a priori information associated with a feature and the computed statistics, the higher the probability that the feature exists. The maximum probability determined by Baye's theory is subjected to two tests, based on absolute and relative threshold values, to reduce the chance of incorrect classification. The method was used for classifying satellite IR data to locate the major water masses in the Gulf Stream region. Results were compared to frontal positions obtained by conventional, subjective means. (Author)

Document Details

Document Type
Technical Report
Publication Date
Oct 01, 1983
Accession Number
ADP003125

Entities

People

  • R. E. Coulter

Organizations

  • Naval Oceanographic Office

Tags

DTIC Thesaurus Topics

  • Artificial Satellites
  • Classification
  • Decision Theory
  • Digital Data
  • Gulf Stream
  • Normal Distribution
  • Oceans
  • Probability
  • Sea Surface Temperature
  • Statistics
  • Surface Temperature
  • Surface Waters
  • Water
  • Water Masses

Readers

  • Computer Vision.
  • Fluid Dynamics.
  • Geodesy

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • Space