Perhaps the best-known use for GIS is the mapping of different features, whether man-made (roads or fire hydrants) or natural (lakes and landforms). As an example, in Yellowstone National Park, here is an image of roads, trails, buildings, and bear management areas in the Park:

While this is often useful, why not do something really cool and interesting with these data? For this project, I set out to determine the distribution of water resources in Yellowstone National Park. It is first helpful to define just what makes a water resource. In general, we will consider significant water resources to be those that have scenic, fishing, historic, or scientific value. Putting these into five major categories, we have:

• Lakes
• Streams and Rivers
• Waterfalls
• Geysers
• Other Thermal Features

The overall goal is to take shapefiles representing each of these features and end up with a raster grid depicting relative intensity across the park. This is depicted below.

Beginning with shapefiles of each feature, the plan is to use distance decay to assign a relative value to every location in the park. Those closer to water features will receive higher values, while those farther will be valued lower. An example of this on a stream/river shapefile is shown. Lighter areas are higher in water resources, darker areas are lower.

This is a good start, but you may notice that every river, stream, and tributary has the exact same lightness pattern around it. Some rivers are more important or valuable than others, so it makes sense that these should have the effect of increasing the lightness (or water resource richness) of areas directly around them more than smaller streams would. This is called Feature Weighting, since features considered more important have greater "weight" than others. In the example below, the bottom lake is weighted more heavily than the other.

A natural next question is, how do we determine weighting of the different features? This will be different for each feature type (lakes, streams, etc.), so here's a quick look at gathering data and determining weighting for Yellowstone waterfalls.

The data source for this feature type is the excellent book The Guide to Yellowstone Waterfalls and Their Discovery, by Paul Rubinstein, Lee Whittlesey, and Mike Stevens (Westcliffe Publishers). The first step is to actually enter locations for each waterfall, made easy by the book's inclusion of UTM coordinates for each fall.

For weighting, it will be necessary to have attributes to work with for each waterfall. What makes a waterfall impressive? Height of the fall is obviously very important, perhaps the most important factor. Another to consider is the volume of water flowing over the falls, since a roaring waterfall is more impressive than a mere trickle. And finally, waterfalls that are plunge type, where the water is generally free-falling, seem to be more impressive than cascades, where the water flows down a very steep slope. Height and plunge/cascade information are both included in the book, so we enter these into a spreadsheet. Stream order, which can be a rough indicator of how major a stream is, comes from a spatial join of the waterfalls shapefile with the stream/river shapefile. This also goes into the spreadsheet

From this set of data, it's a simple matter of developing a formula to use the three waterfall factors to determine waterfall impressiveness score. The final equation gets a bit complicated and involves a fun Natural Logarithm but, as evidenced below, it produces reasonable results.

Applying this formula to all 110 major Yellowstone waterfalls, the final result is a shapefile that has the locations and Weight attribute for the waterfalls. A final step is to use a custom tool, Weighpoints, to create a grid indicating relative waterfall richness in each cell. Developed using the ArcObjects framework in ArcGIS 8.3, Weighpoints takes the weight of all waterfalls, and distance to them, into consideration in order to assign a score to each cell in the grid.

That's the basic process, and this is the result, showing the distribution of waterfalls across Yellowstone:

We apply the same process to other types of water resources and come up with similar images:

The combination of these five grids produces the final output showing distribution of water resources in Yellowstone:

This image helps to clarify locations a bit. Roads, lakes, and rivers are shown, along with waterfalls as blue dots and thermal features as red dots

There you have it, an analysis of water resource distribution in Yellowstone National Park. It's interesting to note how the road pattern generally follows this distribution. Now, wasn't that a lot more interesting than just showing locations of lakes and rivers?

If you like Yellowstone, especially the backcountry, take a look at my adventures over two summers in the park: 500 Miles in Yellowstone