Scales, Levels, and Analysis of the Test Cases

The Initial Test Cases were designed to try to maximize their applicability across the country and across typical model uses.  Effort has been made to cross temporal and spatial scales, and to maximize the amount of observational data that can be utilized for evaluation and validation.  The resulting choices have implications for the analysis scales that can be applied, modeling levels to examine, and observational data available.

Spatial and temporal scales

Each test case produces output on a number of scales - information associated with each fire contained within it, and likely for each day of each fire, if not for each hour, as well as spatial coverages of not only the fire itself, but also its plume's vertical structure and horizontal diffusion.

Because of this, the test cases are most usefully classified by their maximum spatial and temporal scales.  This is generally defined by the horizontal scale of the smoke impacts (dispersion) and the temporal scale of the collected fires in the test case.  In the figure below the test cases are plotted on both maximum spatial scale (x) and maximum time scale (y) axes.  The Fires Everywhere case utilizes climatological fuel moisture and winds and so is classified at the "climatology" scale.

While information is produced that in aggregate covers these scales, the analysis of each test case is done by computing statistics on the information.  The maximum scales define only the maximum possible extent over which the statistic is calculated, but there is nothing to prevent choosing a smaller scale.  For example, a single fire can be extracted from the National Emissions Inventory calculations.

The figure below describes the useful spatial ranges for computing statistics over for each of the test cases.  Initially, these statistics will be computed at the high end of each range, but as the analysis deepens to analyze cause, it is likely that the statistic scales will need to be broken down, for example by region or vegetation type.

The prescribed burns cases (excluding the Naches case which had a specific observational campaign) are shown as not extending down to the local in part because individual prescribed burns do not lend themselves to standard remote sensing and downstream sampling observations due to their more limited size.  It is possible that some of the prescribed burns will be big enough to potentially examine individually, but this is the exception, not the rule.

Modeling steps and observational data

The test cases lend themselves to somewhat different modeling steps, as indicated on the figure below.

The colored bars indicate the modeling steps for which output will be examined.  Some of the cases will use given inputs of fire information and fuel loading;  in this case the level is indicated with an arrow.  Check marks indicate where data is likely to exist that will allow for comparison of least some of the data with observations. We are actively working to gather this data, but it is likely that some areas marked will not be available for model-to-observation comparisons.  In this case model-to-model comparisons can be done instead.

Analysis

The analysis of the test cases will proceed according to the Analysis protocols, with an initial focus on model-to-model comparisons as discussed in the Analysis status page.