Advanced Methods for Atmospheric Modeling
Final rept. 1 Jul-31 Dec 1990,
NIELSEN ENGINEERING AND RESEARCH INC MOUNTAIN VIEW CA
Pagination or Media Count:
The evolution of physical processes in a numerical simulation of the atmosphere and the effects of their interactions on a forecast are very complex and difficult to isolate. An analysis was developed which uses artificial intelligence techniques to study the time-evolving quantities in a numerical simulation in order to determine the relationships among various flow parameters. This procedure was divided into two parts. First, the data was manipulated into a form that allowed the relationships of interest to be isolated. In particular, it was necessary to be able to see how several quantities may interact to affect the solution. The next step used an expert system to analyze the data based on the relationships. The test problem chosen was the numerical simulation of the decay of turbulence in a stratified fluid. Three different analyses were used to determine sensitivities, comparisons, and correlations among the flow quantities. The automation of this technique allowed the processing of the large amounts of data that were generated by a numerical simulation. For the test case, this method was able to draw similar conclusions about the nature of the flow solution compared with those given in the literature.
- Atmospheric Physics