A Multivariate Autoregressive Forecast Model for Short-Term Predictions.

Abstract

Data from 3 years (1972-1974) of synoptic observations collected at seven German stations were used to determine multivariate autoregressive (MVAR) models to make short-term forecasts (3 to 12 hours) for six atmospheric variables (temperature, u-wind, v-wind, visibility, ceiling height, and height of first cloud layer). So that certain tactical constraints could be met, the order and number of predictor variables used by the MVAR models were limited. To emphasize the variance of low ceiling and visibility situations, a variable transformation was performed upon the observations of visibility and the cloud height variables. The best forecasts were obtained when six seven-parameter MVAR models were used. Each model produces a forecast for a particular variable, using the observations at the seven stations as parameters. The variables that can be forecast best are temperature and the u- and v-components of the wind with about 95, 75, and 60 percent of the variance, respectively, explained by the model. From 45 to 70 percent of the variance is explained by the model for visibility while from 30 to 60 percent is explained for the cloud height variables. Finally, data from observations collected in 1976 were used in testing the MVAR models, and the error statistics from these actual forecasts agreed with theory. (Author)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 1981
Accession Number
ADA100465

Entities

People

  • James S. Goerss

Tags

Communities of Interest

  • C4I
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Atmospheric Sciences
  • Classification
  • Coefficients
  • Computations
  • Computers
  • Contracts
  • Covariance
  • Data Analysis
  • Data Storage Systems
  • Errors
  • Information Science
  • Intervals
  • Observation
  • Security
  • Statistical Analysis
  • Statistics
  • Visibility

Fields of Study

  • Environmental science

Readers

  • Atmospheric Science/Meteorology
  • Climatology
  • Statistical inference.