Is This Content Helpful?
We're glad to know this article was helpful.
What is the sensitivity of the weighted overlay process to changes in the percent influence parameters?
This information is not specific to Spatial Analyst in either ArcView GIS 3.x or ArcGIS Desktop 9.x; it can be applied to any version of the Spatial Analyst weighted overlay process. Although ArcGIS 8.x does not have a tool to do weighted overlays, the process can be replicated in steps using the Spatial Analyst analysis tools and Military Analyst.
The sensitivity of the output suitability model to the percent influence is related to the evaluation scale applied to the input rasters in the weighted overlay process. The output suitability model must be an integer raster, so that the percent influence can be changed without seeing a direct change in the output.
For example, a model has three input rasters that are reclassified to an evaluation scale of 1-5. In the three rasters, each cell in the raster has a value of 2, 4, and 2; each raster has a percent influence of 50%, 30%, 20% respectively. The value in the suitability output would be 3. This is calculated as (2*.5) + (4*.3) + (2*.2) = 2.6. This figure must be rounded to the closest integer, which is 3. If the percent influence is changed to 55%, 25%, and 20% but the suitability value of this cell has not been changed, it is still 3. This is calculated as (2*.55) + (4*.25) + (2*.2) = 2.5. This figure must be rounded to the closest integer which is 3.
If the evaluation scale were increased by a power of 10, the above change in the percent influence would affect the output. So, in the first case above the value would be 26, and in the second case above it would be 25.