Regression Methods in Five-Day Mean Height Prediction at 500MB Using Mean NIMBUS II Composited for Identical Periods in the Update Procedure.

Abstract

A series of 15 successive 5-day mean 500mb height fields in gridprint form covering the quasi-hemispheric area between 14.8 - 70.2N latitude was used in this experiment. The time period extended between 15 May 1966 through 28 July 1966. The stepwise regression procedure was employed to derive statistical estimator equations for each mean map field Z(5)(I,J,t+1) in terms of the preceding field data Z(5)(I,J,t) as the primary predictor, (t+1 = 2, ..., 15). As additional predictors, the 5-day mean composited gridpoint values of the NIMBUS II MRIR equivalent black body temperatures in the water-vapor channel (denoted T(1)) and the window channel (denoted T(2)) were introduced bor both analyses times t(1) and t+1. The five-predictor regression equations thus developed for Z(5)(I,J, t+1) proved to have substantial statistical significance, both from the standpoint of the specification and of prediction of Z(5)(I,J,t+2) valid five days later (t+2 = 3, ..., 15). (Author)

Document Details

Document Type
Technical Report
Publication Date
Mar 31, 1972
Accession Number
AD0739358

Entities

People

  • Frank L. Martin

Organizations

  • Naval Postgraduate School

Tags

DTIC Thesaurus Topics

  • Body Temperature
  • Coverings
  • Equations
  • Estimators
  • Grids
  • Latitude
  • Mathematics
  • Specifications
  • Vapors
  • Water Vapor

Readers

  • Atmospheric Science/Meteorology
  • Calculus or Mathematical Analysis
  • Computational Modeling and Simulation