A Study to Determine the Relative Skill of Four Model Output Statistics Prediction Methods Using Simulated Data Fields.

Abstract

This thesis concerns the testing of three MOS prediction methods exercised by the previous NPS investigations (the Maximum-Probability Method II, the Multiple Linear Regression Method, and the Principal Discriminant Method), plus one additional method (Discriminant Analysis Method), on statistically-derived simulated (i.e., controlled) predictor/predict and data sets for the purpose of determining their relative skills in forecasting a generic weather parameter (predicthand). Of the four methods, three use Bayes Law of Inverse Probability to discriminate, while the other method uses conditional Probability. The simulated data sets, models and observers necessary to accomplish this goal are created according to a uniquely developed simulation design. The results indicate that there is a definite diffusing conditional probability, to forcecast the weather parameter. Through the use of Analysis of Variance (ANOVA) technique, this difference is found to be significant with respect to chance.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 1986
Accession Number
ADA167931

Entities

People

  • Steve J. Fatjo

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Air Platforms
  • Autonomy
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Air Force
  • Analysis Of Variance
  • Computational Science
  • Data Science
  • Data Sets
  • Dimensionality Reduction
  • Discriminant Analysis
  • Information Science
  • Meteorology
  • Observers
  • Oceanography
  • Probability
  • Research Facilities
  • Simulations
  • Statistics
  • United States
  • Weather Forecasting

Readers

  • Computational Modeling and Simulation
  • Naval Personnel Management
  • Statistical inference.