Modeling Potential for Bald Eagle Habitat using Fuzzy Logic
PROBLEM
A wildlife biologist is trying to locate the most suitable habitat for Bald Eagles. Fuzzy logic is a method of computation that involves degrees of inclusion, rather than true or false situations. This type of reasoning is well-suited to modeling wildlife habitat and the following criteria will be used to locate suitable habitat using a fuzzy logic method: 1) Far from human disturbances, such as roads, power lines, cities. 2) Close to water. 3) Not too densely or sparsely covered by forest—ideally between 40 percent and 70 percent tree cover.
ANALYSIS
The data are in raster format and include layers for land-cover and human disturbance. The first step is to use ArcMap’s Euclidean Distance tool to produce layers for the distance from human disturbance and the distance to water. The land-cover raster is reclassified on the basis of tree cover on a scale of 1 to 9. I assigned fuzzy membership for the distance to water layer so that small values (close to the water) are assigned high values of membership and I assigned fuzzy membership for the distance to human disturbance such that areas near humans had low values. Areas with a mid-range of tree cover were assigned high fuzzy membership values and both extremes of cover were assigned low values. The Fuzzy Overlay analysis resulted in high values for areas meeting all the criteria.
RESULTS
The output of the fuzzy membership tool is a raster that assigns each cell a value based on the values specified by the user (e.g., high values for cells close to water). The fuzzy overlay tool then combines these inputs into a new raster, showing the combined effects of all the values. This output raster can be mapped to show areas that meet all of the input criteria, in this case, areas most suitable for Bald Eagles (Fig 1).
REFLECTION
This assignment focused on a geoprocessing technique that is highly applicable to analysis of wildlife habitat because it allows for uncertainties, or degrees or inclusion, rather than Boolean, or true/false inclusion in the model. In this example, I found suitable habitat for Bald Eagles based on three variables, but any number of variables may be used. One reason this tool is so powerful is it allows the model’s parameters to be defined by the user based on expert input, allowing a great deal of customization and flexibility. To master this technique, I had to reclassify raster data, calculate distance (Euclidean Distance) or membership (Fuzzy Membership), and perform an overlay that combines several rasters into one. Each of these skills can be used separately for different purposes, and this particular type of analysis is widely applicable to the field of ecology.
Using Weighted Overlay to Locate Suitable Habitat for Black Bears in the Great Smoky Mountains National Park
PROBLEM
Wildlife Biologists in the Great Smoky Mountains National Park relocate Black Bears that come into contact with humans in an effort to reduce conflict and future interactions between humans and bears. In order to relocate the bears, the biologists need to locate suitable habitat. A weighted overlay is another overlay technique that uses data that has been reclassified to the same scale.
ANALYSIS
The data include a raster of elevation and vector files of vegetation, roads, trails, and streams. The relationship between each variable and the preference that Black Bears have shown towards that variable has been provided. I used ArcMap to reclassify each of the variables in terms of its favorability for bear habitat, and then used the Weighted Overlay tool to determine which areas within the park boundaries would be favorable bear habitat. The analysis was performed using the Model Builder so the tool can be re-run using different settings if desired in the future (Fig. 1).
RESULTS
A raster representing slope was produced from the elevation raster. The vector data was transformed into raster format via directly transforming the data or creating a Euclidean Distance raster. Each of these rasters were reclassified to the same scale and then overlaid according to the weight specified by the user to produce a raster representing all of the variables combined (Fig. 2). Here, the raster represented areas with varying degrees of suitability for Black Bears.
REFLECTION
This is another powerful tool in terms of customizing the input. The user can specify the scale using as many different levels as appropriate, and then give each variable its own weight in the analysis. So, this is a great tool if not all variables are equally important. This analysis can also handle variables with different scales. For example, up to 360 circular degrees for aspect, up to 90 angular degrees of slope, a high of 41 inches of rain, and up to 100 percent tree cover are variables that could be combined into one analysis. Like fuzzy overlay, weighted overlay and its component skills (reclassify and rescale) are widely applicable to ecological studies.
While both of these overlay techniques sound very similar, they differ in their founding principles. Fuzzy overlay is designed to handle uncertainty or inaccuracies in the data in terms of the boundary of each class. Weighted overlay is based on clear (Boolean) membership in a class or no membership. Fuzzy overlay results in a representation of the possibility the area is ideal while weighted overlay results in a probability of membership. Depending on the type of data available and the questions you are asking it might be possible to use both types of analysis and then determine which result is more accurate. Because of my interest in modeling wildlife habitat, I expect these skills to be used frequently in my future.
Using Census Data to Identify Areas with High Rates of Uninsured Women in Pennsylvania
PROBLEM
Lack of health insurance is a problem that plagues far too many Americans. As someone who spent most of my adult life in the uninsured category, I was interested in how many others there were like me from the area in which I grew up. I chose to create a map showing the number of females aged 45-54 without health insurance in each county in Pennsylvania. I will use data collected through the U.S. Census Bureau, and available on American Fact Finder, to create this map.
ANALYSIS
The appropriate data are downloaded from the American Fact Finder website (Census data and TIGER shapefile). The tabular data are examined and cleaned, a field is added and populated with the unique identification code so that types of data match in the fields being used to join the spatial and tabular files. Now the layer can be symbolized based on the desired attribute, in this case, the number of females without health insurance in the targeted age group (Fig. 1).
RESULTS
The results show that both sides of Pennsylvania have areas with very high populations of uninsured women in their late 40s and early 50s (Fig. 2). This pattern corresponds to the population centers of Pittsburgh and Philadelphia, but also shows moderately high rates of uninsured women scattered across the state.
REFLECTION
There is a plethora of census information that is available that could be used to answer many questions about socio-economic factors, so understanding how to search these records for the desired data and how to find the appropriate shapefile to join with them is an important skill. Cleaning the data and making the fields match for a join is a skill that can be applied to many problems. For example, I have used this skill to join information about trash removed to the spatial location where the trash eventually was located. Any research project that has data stored in tabular form and has spatial locations associated with these data could apply this technique to analyze such data spatially.