Overview of Fuzzy Logic Site Selection in GIS

BY AMANDA BRINEY




Site selection, also called suitability analysis, is a type of GIS analysis that is used to determine the best site for something. Potential sites used in suitability analysis can include businesses such as a store or city facilities like a hospital or school. Site selection can also be used to determine ideal habitat for a specific plant or animal species. When performing site selection analysis in GIS users must set various criteria so that the best or ideal sites can be rated based on this criteria.

Fuzzy logic is one type of commonly used type of site selection. It assigns membership values to locations that range from 0 to 1 (ESRI). 0 indicates non-membership or an unsuitable site, while 1 indicates membership or a suitable site. Fuzzy logic site selection is different from other site selection methods because it represents a possibility of an ideal site, rather than a probability and it is commonly used to find ideal habitat for plants and animals or other sites that are not specifically chosen by a user or developer (ESRI).

How to Use Fuzzy Logic Site Selection
Like other site selection methods, fuzzy logic uses a standard workflow to ensure that all necessary steps are followed. It is different from other methods however because it is much more complex and uses a continuum of values between 0 (completely false or unsuitable) and 1 (completely true or suitable) rather than a simple yes or no (ESRI). Fuzzy logic is capable of examining conditions that can be both true and false at the same time.

The standard workflow for fuzzy logic is as follows:

1. Define the problem and site selection criteria
2. Collect criteria layers
3. Assign fuzzy membership values
4. Perform fuzzy overlay
5. Verify and apply results

Defining the problem and selection criteria is the most important step in fuzzy logic site selection because it helps the user to determine the type of data needed for the analysis. Fuzzy logic membership (discussed further below) is an important reclassification step. Reclassification is used to simplify the interpretation of raster data by changing a single input value into a new output value (ESRI). Fuzzy overlay allows the user to overlay the various reclassified layers to analyze the possibility of a specific occurrence. This can then be used to verify the results and use them to choose the best site.

Fuzzy Logic Membership
Fuzzy logic membership helps the user to determine the likelihood that a site is suitable or unsuitable. This step assigns values from 0 to 1 with 0 being not likely or unsuitable and 1 being most likely or suitable (ESRI). Thus, the higher the fuzzy membership value, the more ideal the site. When assigning fuzzy membership values it is important to understand the four types of membership and choose the one that best fits the analysis criteria. These membership types are as follows:

1. Linear – High fuzzy membership is assigned to large or small values and fuzzy membership decreases at a constant rate.
2. Small – High fuzzy membership is assigned to small values.
3. Large – High fuzzy membership is assigned to large values.
4. MS Small – High fuzzy membership is assigned to values less than the mean.
5. MS Large – High fuzzy membership is assigned to values more than the mean.
6. Near – High fuzzy membership is assigned to mid-range values.

GRAPHIC EXAMPLE OF THE MEMBERSHIP FUNCTION TALLNESS. SOURCE: RAINES, SAWATZKY, AND BONHAM-CARTER, ESRI.

Fuzzy Logic Overlay
Once the appropriate fuzzy membership value for data criteria is assigned several reclassified surfaces showing a value from 0 to 1 are generated (ESRI). The next step in applying fuzzy logic is to overlay these surfaces. This step is similar to weighted site selection (a site selection type that allows users to rank raster cells and assign a relative importance value to each layer) because the different reclassified surfaces are compared to each other (ESRI). To complete this step, one of several fuzzy overlay types must be chosen. The fuzzy overlay types are as follows:

1. And – This type is best used for finding the locations that meet all criteria.
2. Or – This type is best used for finding the locations that meet any of the criteria.
3. Product – This type is best used for finding the best locations with combined input fuzzy membership values (ESRI).
4. Sum – This type is best used for finding all possibly suitable locations with combined input fuzzy membership values (ESRI).
5. Gamma – This is a complex fuzzy overlay type that requires expert knowledge and a combination of various sub-models.

PREDICTING THE POST APOCALYPTIC GEOGRAPHY OF ZOMBIES USING FUZZY LOGIC. SOURCE: THE UNDEAD LIVEN UP THE CLASSROOM, EDWARD GONZÁLEZ-TENNANT.

When to Use Fuzzy Logic
Because there are several different site selection methods it is important to understand when to use a complex method like fuzzy logic. Fuzzy logic site selection is most commonly in projects that have an element of uncertainty or where the user cannot state specifically where a site would be as would be the case of an ideal site found with weighted site selection (ESRI).

Fuzzy logic site selection is also ideal for analyzing data that does not have discrete polygons and boundaries as would be the case with a new sporting venue for instance (ESRI). Instead, fuzzy logic can be used to look at areas of deer habitat based on a factor such as elevation. Potential habitat types could be classified based on elevation levels. In this example, low elevations would be considered suitable habitat and given values close to 1 while high elevations would be unsuitable and have values closer to 0.

References

ESRI. (n.d.). “Using Raster Data for Site Selection.” ESRI Virtual Campus. Personal Notes. (Course Taken 2 April 2014).

Further Reading
Incorporating Expert Knowledge – New fuzzy logic tools in ArcGIS 10 By Gary L. Raines, Don L. Sawatzky, and Graeme F. Bonham-Carter

Source