Use Existing Weights/Variances Tables...

This option allows some flexibility in the application of the Weights of Evidence model. Once a set of weights and variances tables have been calculated by running the Calculate Response Theme... function, these tables can be ‘reused’ as a source of weights for application to a different area (which may or may not include the original study area.)

This menu item will be enabled if there is at least one weights table and one variances table in the format output by Arc-WofE.

The general idea

After calculating a response theme, the study area my be changed (in Set Analysis Parameters). If a new Response Theme is then calculated by running the ‘Use Existing Weights/Variances Tables...’, it must take as its inputs the same evidential themes in order to compute a the unique conditions grid and set of weights and variances calculated based on a previous study area.

An examples

This example is a general description of a user’s application of this Arc-WofE option.

Suppose that the evidential themes available to the user had a much greater geographic extent than the set of known mineral occurrences or training points. An assumption was made that the same geological model applies to the entire area. Two study area grids were created: one that defined the extent of the evidential themes and a second that defined an area roughly encompassing the extents of the training points, about 25% of the area of the first study area. After calculating weights and then generalizing individual evidential themes, one response theme was calculated using the smaller study area. (Using the 'Calculate Response Theme...' menu item.)

The study area theme was then changed (in the parameters) to the larger geographic area and a second response theme was calculated using the weights and variances tables calculated for the smaller extent. (Using the 'Use Existing Weight/Variances Tables...' menu item.) In this way, weights were calculated for a relatively confined area that contained many known occurrences, but then those weights were applied to adjacent areas with no training points.

Note that the prior probability calculated for the first study area is used in the calculations for the second study area.