The most important of steps in model design, application, and testing are to define the goals of a modeling study, select appropriate algorithms, obtain sufficient input and emissions data, compare model predictions with data, and analyze results. To illus trate these steps, the design of an existing model is briefly discussed, and statistical and graphical comparisons of predictions from the model to data are shown.

Model design, application, and testing require several steps. These include (1) defining and understanding the problem of interest, (2) determining the spatial and temporal scale of the problem, (3) determining the dimension of the model, (4) selecting the physical, chemical, and/or dynamical processes to simulate, (5) selecting variables, (6) selecting a computer architecture, (7) codifying and implementing algorithms, (8) optimizing the model on a computer architecture, (9) selecting time steps and intervals (10) setting initial conditions, (11) setting boundary conditions, (12) obtaining input data, (13) obtaining ambient data for comparison, (14) interpolating input data and model predictions, (15) developing statistical and graphical techniques, (16) comparing results with data, (17) running sensitivity tests and analyzing the results, and (18) improving algorithms. Each of these is discussed below.

The steps 6, 7, 8, 14, 15, 18 are more of the scientific and computational topics beyond the air quality management, thus are not specifically descript here.