Analyzing Spatial Data in GIS
Geog 483 – Lesson 3
January 30, 2010
The objective of lesson 3 is to supplement from the skills (table manipulations) learned in
lesson 2 by completing an overlay analysis (intersection) and proximity analysis (buffer)
to some of the previous tornado data. At the end of this analysis, we will be able to find
potential emergency relief sites and see which affected areas need the most emergency
aid. In this exercise, we will be completing the following operations:
Executing attribute and spatial queries
Updating and creating new attribute fields in the data (tables) provided.
Joining a feature attribute table together with a stand alone table.
Creating a buffer zone or proximity analysis around a line and area layers.
Completing an overlay analysis such as an intersection.
Creating a thematic map using specific features.
High population areas affected by these tornadoes probably need the most help.
Therefore, the population density for each county was calculated. In order to determine
the population density for each county, the population stand alone table was joined to the
counties layer. Next, some modifications to the Tornadoes layer had to be completed.
First we needed to add some extra data that was missing in the original layer. To add the
extra information, the tornado attributes table was joined with the tornado paths layer.
We used some of the added attributes in some of the next upcoming steps.
Another issue with the tornado paths layer is that the layer represents the centerline of the
tornado path. We wanted to see what areas were damaged by the tornado. Using half
the path width of each tornado as a buffer distance, an area damaged polygon layer was
created. Next we wanted to find a safe area or zone to set up aid stations. This had to be
within one mile of the damaged area but not in it. A one mile buffer around the damaged
area layer was completed to create an emergency relief zone layer. Now we can use this
emergency relief zone to locate potential emergency relief sites.
Hospitals, schools and churches are great places for setting up emergency help sites. The
Oklahoma place layer has extra features (i.e. cemeteries) that do not meet our criteria. To
filter out the features not needed, an attribute query was completed. Now I have a list of
hospitals, schools, and churches. In addition, we want to makes sure that the sites are
close (within one mile) to the damaged areas. The emergency relief zone created earlier
was used in the next step. Using a spatial query, I wanted to find the number of
emergency sites within the emergency relief zone. Now that there is a list of 33
potential sites (Fig 1), we had to prioritize what affected areas need the most help. Our
relief effort was prioritized by the following criteria: areas that were hit hard by the
tornadoes (intensity) and areas with high population densities. In the next step, an
overlap analysis was performed.
Figure 1. List of Potential Sites for Relief Efforts.
Definition of an Overlap Analysis
According to the ArcGIS Desktop Help (2010), an overlay analysis is the question of
what layer is on top of another layer. An example of this would be: What is the land use
around the Mississippi River?? Most of the time the overlay analysis is used in
conjunction of another analysis such as the proximity analysis (buffer) which we
completed in this exercise. An overlay analysis can also be used with a surface analysis
(i.e. slope) which hopefully we’ll learn in a future class.
An overlay analysis has three parts: the input layer, the overlay layer and the output
layer. An overlay analysis can join, erase, or update attributes or features from the input
layer and overlay layer to create the new output layer. Although there are several
different types of overlay analysis, we focused on an intersection in this exercise. When
an intersection has been completed, the output layer has features that are common to both
the input and overlay layer. For this exercise, we wanted only the emergency relief zone
area to have both the population density and tornado intensity in one final layer.
Therefore, I intersected the emergency relief zones layer and counties new layer to create
a Relief Priority Zones layer. Since I wanted to prioritize the relief effort by the
population density and tornado intensity, I used the following equation to populate a
relief priority measure:
ReliefPrty = (Right ([F_Scale + 1), +1) * PopDense
To display the emergency relief zones by priority in my thematic map, I used a graduated
color ramp (light to dark) with five classes.
Most of the relief efforts should occur in Oklahoma County (Fig 1). This area has the
highest population density and the strongest tornadoes. In a real world analysis, I think I
would want to know where the nearest hospital is from these emergency relief areas. The
within 1 mile criteria wouldn’t work here because looking at this map the nearest hospital
is in Canadian County which may be 30 miles away. So I decided to change my criteria
just for kicks.
Figure 1. Candidate Relief Sites & Priority Relief Zone for the May 9, 1999 tornadoes in
I decided I wanted to know where are the hospitals within a distance of 10 miles from the
relief priority zones. In Figure 2, a few more hospitals popped up on the map using the
new criteria. However Logan County and Kingfisher County, for example, does not have
any hospitals near by. In order to get better results, I would have to increase my distance.
In addition to hospitals, I may want to know what major roads have been affected by
these tornadoes. These roads could be closed and could slow down the transportation of
supplies to the relief sites.
Figure 2. Hospitals with 10 miles of Relief Priority Zones.
Similar to the lesson 2 try this section, we were asked to use census data in our analysis.
The results we will get from using the census tracts are more accurate than the county
data used earlier. Census data is updated every 10 years; where as county data may not
be updated as frequently. Census tracts are subdivisions within a county (DiBiase, 2009).
Each subdivision or tract can have a population range of 2,500-8,000 individuals.
In addition, the size of the subdivision or tract varies depending on the area. When we
use the county population data, it is assumed that the population is evenly distributed
throughout the county. I can see this in my results above. Using only the county data, all
relief priority zones in Oklahoma County are considered to be at the highest level of
priority (Fig 2). In the real world, we know that populations are not evenly distributed
throughout a county. When we used the census data in the try this exercise, I can see this
difference in my results. For example, in Oklahoma County the relief priority zones are
now broken down to five levels of priority (Fig 3).
Figure 3. Relief Sites & Relief Priority Zones by Census Tracts with a close up of
Oklahoma & Cleveland Counties.
DiBiase, David (1999-2009). The Nature of Geographic Data, Chapter 3. Pennsylvania
State University World Campus Certificate Program in GIS. Retrieved in 2009. from
ESRI’s ArcGIS Desktop Help (2010). Overlay Analysis Retrieved January 29, 2010.
King, E., & Walrath, D (1999-2010). Problem-Solving with GIS, Lesson 3, The
Pennsylvania State University World Campus Certificate/MGIS Programs in GIS.
Retrieved January, 27, 2010, from https://www.e-education.psu.edu/geog483/l3_p1.html