EXERCISE 11 – Cost Surface Analysis

1. Cost Surface Analysis

Today we're going to take a quick look at a modeling technique known as cost surface analysis. It's often used to model the potential movement of some entity (an animal or pollutants, for instance) through a study area. The cost surface is a representation of the study area that describes how easy or difficult it is for our entity to travel through each cell (yes, it's raster based) in the study area. For example, if we're modeling salamander movement, a 4 lane highway would have a rather high cost of movement, since it's a very difficult feature for a salamander to cross, while a forested wetland would have a low cost of movement.

Let's say we have an area with two disjoint populations of salamanders in the northern Barnegat Bay watershed. A local land conservation group is interested in preserving land that will help these two populations of salamanders remain connected. We need to find out the areas that make the most sense for them to preserve.

First we need to generate a cost surface. A cost surface can be derived from any combination of layers, such as wetlands, major highways and land use. To keep things simple we're using a land use/land cover data layer that sufficiently delineates high and low travel cost areas from a salamander's point of view. Load \\ad-rsc\data\teach\intgeo\ClassWork\costpath\lu06 and open the attribute table. Notice there is a cost field already here. The higher the value the more difficult a salamander will find the area to cross (the value must always be higher than 0). We're going to use this cost data to determine the least cost path (i.e. the easiest path) that connects the two salamander populations.

In order to do this we need a raster layer that has the cost data in its value field. We can create this by using the Spatial Analyst>Reclass>Lookup tool on the lu06 layer, using the cost field as the lookup field. Once we have this we need to designate a source of salamanders that we can calculate a cost distance surface to. The cost distance surface adds up the cost of moving from the source to every other cell in the cost surface raster. Create this cost distance surface by using the Spatial Analyst>Distance>Cost Distance tool. The input source will be \\ad-rsc\data\teach\intgeo\ClassWork\costpath\vernsource and the cost raster is the cost surface we just created. You'll want to create an output backlink raster (we'll need it later). This raster identifies the direction of easiest movement from the cell to the source.

The cost distance raster shows the minimum cost of movement from the source to each cell. Now that we've created it, we can use it and the backlink raster to determine the least cost path to our destination population of salamanders. Open the Spatial Analyst>Distance>Cost Path tool. The destination data layer is \\ad-rsc\data\teach\intgeo\ClassWork\costpath\verndest, the cost distance and backlink rasters are those we just created, and we can accept the defaults for the rest. The output is a raster that shows the least cost path that connects the populations of salamanders. This shows where preservation activities should be concentrated in order to ensure connectivity between the populations. If you're not satisfied with the single path, you can use the Spatial Analyst>Distance>Corridor command to create a swath of least cost cells. To do this you first have to make a cost distance surface using the salamander destination point as the feature source. Use the Corridor tool to combine the two cost distance layers. The low cost values in the corridor output represents a low-cost corridor that connects the source and destination points.

Assignment: Make a copy of the lu06 layer in your directory. Edit the cost values to values of your own choosing (higher is more difficult and they must be greater than zero) and turn this into a cost surface raster. Then create the least cost path as above and make a nice map using the land use data and path. You'll have to convert the cost path raster to a line (Raster to Features (use type polyline)) under the Spatial Analyst>Convert menu).

Hand it in Monday, April 28th.

2. Select by location

One potentially useful tool you should take a look at is the Select by Location tool, found under the Selection menu. This tool allows us to select features of one layer using the location of features in another layer. We could, for example, select land cover polygons that intersect roads or a set of sample points. We aren't limited to using intersections, either. There is a relatively comprehensive list of geographic relationships we can use, including within a distance of. We can also buffer the features in the layer we're using to select the other features.

3. Labeling and Legends

By now you should be advanced enough in your map making skills to find the labels and legends defaults in ArcGIS a bit limiting. What can you do about them?

Well, the engine that ArcGIS uses to place labels on maps is powerful, with many options that can be adjusted under the Label properties for any layer. I'm not going to talk about them because I don't think they're very useful for most of the labelling issues you'll encounter in class. Instead I'll tell you the nuclear option for getting the labels to look the way you want them. If you have a labeled layer and you're just not happy with the positioning, you can right-click on the layer name and Convert the Labels to Annotation. This turns every label into its own free floating text box that you can then manipulate to your heart's content. Why is this the nuclear option? Because, you delink the labels from the label properties for the layer, meaning the label properties of the layer will no longer change the appearance of the layer. You can select all the labels and alter their appearance by using the Select All Elements tool under the Edit menu.

There's a nuclear options for legends too. If you select a legend and right-click, you'll see one of the options is Convert to Graphics. This converts the layer elements to graphic objects that can be manipulated independently (you need to Ungroup them before you can work on them). This is a nuclear option because it delinks the legend from the Symbology properties of the layer. Any changes you make in the layer's symbology after you convert its legend to graphics will no longer be automatically changed in the legend. Again, this means make sure you have your Symbology set before you convert your legend. Once you convert you can then alter the placement and look of the legend elements as much as you'd like.