ADVANCED METHODS FOR ATMOSPHERIC MODELING

ATMOSPHERIC SCIENCE INVOLVES STUDIES OF VERY COMPLEX PHYSICAL PROCESSES. OFTEN, SIMPLIFIED VERSIONS OF THESE PROCESSES ARE SIMULATED IN A LABORATORY OR ON A COMPUTER. COMPUTATIONAL FLUID DYNAMICS (CFD) METHODS PROVIDE A GOOD WAY TO PERFORM THESE SIMULATIONS OF ATMOSPHERIC PHENOMENA, SINCE THEY ALLOW FOR THE EASY VARIATION OF INDEPENDENT VARIABLES. HOWEVER, AS THE NUMBER OF POSSIBLE COMBINATIONS OF VARIABLES INCREASES, THE AMOUNT OF GENERATED DATA GROWS TO THE POINT THAT IT IS DIFFICULT TO DISTILL SPECIFIC RESULTS FROM THE SOLUTIONS. AN INNOVATIVE METHOD IS PROPOSED HERE WHICH WILL APPLY AN "INTELLIGENT" POSTPROCESSOR TO THE RESULTS OF A SERIES OF RELATED COMPUTATIONS TO GIVE INSIGHT INTO THE CAUSE-AND-EFFECT RELATIONSHIPS FOR VARIOUS FLOW QUANTITIES. THE POSTPROCESSOR DETERMINES HOW CERTAIN FLOW RESULTS, FOR EXAMPLE, THE TIME HISTORY OF A TURBULENT VELOCITY FIELD, ARE DEPENDENT UPON THE VARIABLES OF THE CALCULATION, SUCH AS THE DEGREE OF STRATIFICATION. THE CODE ALSO DETERMINES IF A FLOW RESULT IS DEPENDENT UPON MORE THAN ONE VARIABLE. IN ADDITION, THE POSTPROCESSOR CHECKS FOR CORRELATIONS AMONG SEVERAL FLOW QUANTITIES OF INTEREST OVER THECOURSE OF A CALCULATION. THE RESULT OF THE POSTPROCESSING IS THE DETERMINATION OF FUNCTIONAL RELATIONSHIPS BETWEEN THE COMPUTED FLOW QUANTITIES AND THE INPUT VARIABLES, IN ADDITION TO THE RELATIONSHIPS AMONG THE FLOW QUANTITIES THEMSELVES.