Markov Simulations of One- and Two-Dimensional Weather Data Bases.
Abstract
In many cases, weather is only one of many inputs to a simulation model, so a realistic but simple weather simulation method should be included in the model. This paper has three major areas of concern: (1) A fairly extensive review of applications of one-dimensional Markov and seim-Markov chains to weather data simulations, (2) A consideration of factors involved in and methods that are appropriate for extending Markov concepts to simulating gridded data in two or more dimensions, and (3) Evaluation of the proposed methods in terms of realism and simplicity of application. A discussion of the general characteristics of real weather variables and observed weather data in the context of simulating weather as a stochastic process is also given. The data base used for the example consisted of gridded weekly maps of temperature departures from normal in an area of the United States. For most analyses, the data was converted to five states, from state 1 (coldest) to state 5 (warmest). In the real data, it was rare to have an occurrence of unequal and nonconsecutive states in adjacent grid points. Such occurrences were called 'unusual transitions, ' and one criterion for evaluating the realism of a weather simulation scheme was the frequency of generating such transitions.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1979
- Accession Number
- ADA080217
Entities
People
- Steven R. Schroeder
Organizations
- Air Force Institute of Technology