This project involved the creation of model in ArcGIS’ Model Builder application for the purpose of modeling mountain goat habitat along the North Cascades according to a series of winter and summer habitat criteria. Initially, a vector model was created, which took over four-and-a-half hours to run for a chosen study area of fifty townships totaling approximately 1710 square miles. This vector model proved extremely cumbersome, taking almost forty-four processes to complete due to a desire to maintain accuracy.
A second model was then created using entirely raster processes. This model, using the ‘Maximum of Inputs’ for a cell size value in the model’s environment settings, took approximately six minutes to run for either the summer or winter habitat modeling. The downfall of this approach was extreme lack of precision, with cell sizes ballooning to 190×190 meters. A compromise between models one and two was created, using ‘Minimum of Inputs’ for the cell size setting. This produced a model with 25×25 meter cells, with processing taking about twenty minutes. Using a Model Builder model meant that changing of environmental settings was all that was necessary, as the processes remained the same. This allowed for great time-saving, not having to re-create each process by hand just to change one setting. Model size could have been halved if the ECW and WCW rasters were mosaicked from the onset.
Depicted below is the final map output using 25×25 meter cells for raster analysis:
Modeled Habitat Area Stats:
Winter Habitat – 7,360 Acres
Summer Habitat – 2,389 Acres
Total Habitat – 9,749 Acres
Percentage of Goat Points in Habitat Area:
Winter Habitat – 0.53%
Summer Habitat – 0.86%
Total Habitat – 1.39%
When I got these low results, I decided to find the percentage of goat points in my previously created 190×190m cell size habitat outputs. This mimics a buffer effect, with noticibly higher percentages of points in the habitat area.
Percentage of Goat Points in 190×190m cell size Habitat Area:
Winter Habitat – 20.15%
Summer Habitat – 12.56%
Total Habitat – 32.71%
The Gap Analysis Program data was overlaid onto the final cartographic output, which showed that modeled habitat was predicted to occur in all GAP data HABZONEs. These zones are labeled as follows:
0 – Habitat not suitable
1 – Habitat Suitable and occurring in a primary vegetation zone
2 – Habitat Suitable and occurring in a peripheral vegetation zone.
A comparison of GAP data between the output for this GIS Lab must start with an understanding of the nature the GAP study itself. It was meant to primarily be a generalized habitat model for many different species in Washington State. This being said, our input criteria for mountain goat habitat appears to lend itself to a more constrained output than the GAP data. GAP data for example, was derived using vegetation types as a primary determinant for habitat suitability, where as our analysis took into account topological factors such as elevation, slope, and aspect in addition to percentage of vegetation cover. Using vegetation types as a primary input for the GAP analysis model naturally leads to different outputs than we see with our analysis in the lab.
The spatial extent of our study is also much smaller than the extent of the GAP study. While we working specifically in the Cascades, the GAP study strives to model habitat throughout the state. It is not expected then, that the GAP data should be as fine grain as the outputs for our lab.
Furthermore, it appears that the GAP data is designed to be used in identifying critical habitat areas under the criteria that they are habitat for multiple species rather than just a single species. Knowing this, one can better understand why vegetation is used as a primary input for the GAP analysis model rather than species specific criteria such as those used as inputs in the lab.
Overall, both the GAP study and our analysis, while somewhat similar topics, have much different purposes. The sheer scope of the GAP project both spatially as well as in the number of species that it encompasses is far greater than our little lab. It is because of this that I feel that comparing the two together can quickly lead to unfair comparisons if one does not understand the purpose of each study.