Project 3: Analyzing Spatial Data in GIS
By Luminita Slevoaca
Initial data
Project 3 is a continuation of Project 2 and analyzes further the damages caused by the tornadoes that struck central Oklahoma on May 3rd, 1999. This project focuses on identifying high priority relief zones and institutions like school, churches and hospitals that are within these zones. Project 3 uses the same dataset like Project 2, plus one additional shapefile “OkPlaces” that contains the location and the name of potential relief sites.
Analysis
In this project we have been introduced to two new GIS analysis tools: buffering and intersection. The first requirement was to establish the relief sites and zones prioritized by counties. In order to create this map, the first step was to join the attribute table “Population”, which contains statistical information about the population, with the “Counties” shape file. We realized this join operation to be able to calculate the population density per county. Before we added and calculate the population density field, we exported the data into a new shapefile named “CountiesNew”. A join is not a permanent operation and can be easily removed. Exporting the data after a join makes the association of data permanent.
If in the previous project we represented the tornadoes proportional to their strength according to the Fujita Scale (F-Scale), in this project we are representing the width of their path. To accomplish this we were introduced to the buffer tool located in ArcToolbox->Proximity. The buffer tool will create a polygon around a specified geographic feature. The buffer tool gives us the option to create a standard width buffer or individual width for each feature, values that are stored in a field from the attribute table. Since the tornado table has a field that stores the width of each tornado, we will create a buffer based on that variable. First we need to create a field that contains the width of each tornado divided by two. The reason for this operation is that the buffer tool will use the distance stored in that field and apply it around each side of the centerline of the tornado, otherwise it will result in doubling the width of the tornado. After representing the width of the tornadoes on the map, we further applied a one mile buffer around them to identify the relief priority sites and zones. These relief sites located in the one mile buffer around the tornado path could have provided a shelter for the population that lives in the immediate vicinity. For our analysis we used attribute and spatial query to identify only the relief sites that are school, churches or hospitals that are in high priority zones.
Intersection was another new tool used in this lesson. The result from intersecting two shapefiles will contain all the features and portions of features from the overlapping areas. We intersected the emergency relief zones layer, which contains information about the tornadoes severity with the county layer, which contains demographic statistics and population density. The resulting layer contains both information about the tornadoes and population. This tool was especially useful in this case because the emergency relief zones layer has no spatial correlation with the counties layer and a join could not been performed here. The intersected layer we are able to map the relief priority zone, classified by relief priority field. The relief priority field was calculated (F_Scale + 1) * population density. F_Scale, represents a number from 0 to 5 associated with the intensity of the tornado and addition to 1 was necessary here to avoid multiplying by zero.
The map below shows the results of our analysis.
By Luminita Slevoaca
Initial data
Project 3 is a continuation of Project 2 and analyzes further the damages caused by the tornadoes that struck central Oklahoma on May 3rd, 1999. This project focuses on identifying high priority relief zones and institutions like school, churches and hospitals that are within these zones. Project 3 uses the same dataset like Project 2, plus one additional shapefile “OkPlaces” that contains the location and the name of potential relief sites.
Analysis
In this project we have been introduced to two new GIS analysis tools: buffering and intersection. The first requirement was to establish the relief sites and zones prioritized by counties. In order to create this map, the first step was to join the attribute table “Population”, which contains statistical information about the population, with the “Counties” shape file. We realized this join operation to be able to calculate the population density per county. Before we added and calculate the population density field, we exported the data into a new shapefile named “CountiesNew”. A join is not a permanent operation and can be easily removed. Exporting the data after a join makes the association of data permanent.
If in the previous project we represented the tornadoes proportional to their strength according to the Fujita Scale (F-Scale), in this project we are representing the width of their path. To accomplish this we were introduced to the buffer tool located in ArcToolbox->Proximity. The buffer tool will create a polygon around a specified geographic feature. The buffer tool gives us the option to create a standard width buffer or individual width for each feature, values that are stored in a field from the attribute table. Since the tornado table has a field that stores the width of each tornado, we will create a buffer based on that variable. First we need to create a field that contains the width of each tornado divided by two. The reason for this operation is that the buffer tool will use the distance stored in that field and apply it around each side of the centerline of the tornado, otherwise it will result in doubling the width of the tornado. After representing the width of the tornadoes on the map, we further applied a one mile buffer around them to identify the relief priority sites and zones. These relief sites located in the one mile buffer around the tornado path could have provided a shelter for the population that lives in the immediate vicinity. For our analysis we used attribute and spatial query to identify only the relief sites that are school, churches or hospitals that are in high priority zones.
Intersection was another new tool used in this lesson. The result from intersecting two shapefiles will contain all the features and portions of features from the overlapping areas. We intersected the emergency relief zones layer, which contains information about the tornadoes severity with the county layer, which contains demographic statistics and population density. The resulting layer contains both information about the tornadoes and population. This tool was especially useful in this case because the emergency relief zones layer has no spatial correlation with the counties layer and a join could not been performed here. The intersected layer we are able to map the relief priority zone, classified by relief priority field. The relief priority field was calculated (F_Scale + 1) * population density. F_Scale, represents a number from 0 to 5 associated with the intensity of the tornado and addition to 1 was necessary here to avoid multiplying by zero.
The map below shows the results of our analysis.
Click on the image to enlarge.
This map depicts relief priority zones and the selected relief sites located within these zones. In the legend of the map, 1- Highest Priority represents the zones that had the highest population density. As we notice, because the population density is calculated by county our map has a general representation of the priority zones. This map will not be very useful in a real world emergency action because representing the population density by county does not offer a very clear image of the population distribution. In a county, the majority of the population can be concentrated in a small urban area, while small towns would have a very low population density. In real world, for the relief efforts to be concentrated on important areas, we would need a more detailed representation of the population density.
The attribute table of the candidate sites represented on the map contains among other information their name, geographic coordinates and type. This table is arranged alphabetically after the “F_Type” field, using the AZ Advance Sorting command.
This map depicts relief priority zones and the selected relief sites located within these zones. In the legend of the map, 1- Highest Priority represents the zones that had the highest population density. As we notice, because the population density is calculated by county our map has a general representation of the priority zones. This map will not be very useful in a real world emergency action because representing the population density by county does not offer a very clear image of the population distribution. In a county, the majority of the population can be concentrated in a small urban area, while small towns would have a very low population density. In real world, for the relief efforts to be concentrated on important areas, we would need a more detailed representation of the population density.
The attribute table of the candidate sites represented on the map contains among other information their name, geographic coordinates and type. This table is arranged alphabetically after the “F_Type” field, using the AZ Advance Sorting command.
Click on the image to enlarge.
In the “Try This!” exercise we have been provided with census tract shapefile and a demography attribute table. I applied the same steps like in the above analysis and I obtained similar map, only instead of having the population density represented by county, it is represented by census tract. Census tracts offer a better pattern of the population distribution and the map would be much more useful for a real world emergency response team.
The map below represents the relief zone and candidate sites per census tract and classified by population density. To classify the data, I used the Natural Breaks classification, which breaks the features into classes where big differences in data values occur.
In the “Try This!” exercise we have been provided with census tract shapefile and a demography attribute table. I applied the same steps like in the above analysis and I obtained similar map, only instead of having the population density represented by county, it is represented by census tract. Census tracts offer a better pattern of the population distribution and the map would be much more useful for a real world emergency response team.
The map below represents the relief zone and candidate sites per census tract and classified by population density. To classify the data, I used the Natural Breaks classification, which breaks the features into classes where big differences in data values occur.
Click on the image to enlarge.
Because the highest population density occurs in Oklahoma County, below I will present a close-up of the county.
Because the highest population density occurs in Oklahoma County, below I will present a close-up of the county.
Click on the image to enlarge.
The above map, presented at this scale would prove to be of real use for a response emergency team.
To verify that Natural Breaks classification produced a good spatial pattern, I also tried using a quantile classification. The result is shown below.
The above map, presented at this scale would prove to be of real use for a response emergency team.
To verify that Natural Breaks classification produced a good spatial pattern, I also tried using a quantile classification. The result is shown below.
Click on the image to enlarge.
Analyzing the two maps, we can see that the Natural Breaks classification gives us a better spatial pattern distribution of the areas that are considered highest and higher priority. In a Natural Breaks classification, the relief priority zone 5 - Lowest Priority has a density of under 2,000 people per square mile and a relief priority zone 1 – Highest Priority between 24,522 and 45,758, while the quantile classification has relief priority zone 2 – Higher Priority with a population density between 139 and 2463, while relief priority zone 1 – Highest Priority has a population density between 2464 and 45758. Because quantile classification breaks the data into classes with equal number of feature, relief priority zone 1 – Highest Priority does not offering meaningful results.
Analysis limitations
Although this analysis provides with a basic map of relief priority zone and candidate relief sites, for a real world map other factors might taken in consideration. For a thorough analysis we have to take in consideration that the population density is not constant in a census tract and additional information should be used. Limiting the buffer zone to one mile may exclude important relief sites situated in the immediate vicinity. In an emergency situation, important information to consider would be the hospitals situated within a certain distance from the relief sites where possible victims can be transported to. I created a map that shows the hospitals that are in 10 miles buffer zones from the relief sites situated in central Oklahoma County.
Analyzing the two maps, we can see that the Natural Breaks classification gives us a better spatial pattern distribution of the areas that are considered highest and higher priority. In a Natural Breaks classification, the relief priority zone 5 - Lowest Priority has a density of under 2,000 people per square mile and a relief priority zone 1 – Highest Priority between 24,522 and 45,758, while the quantile classification has relief priority zone 2 – Higher Priority with a population density between 139 and 2463, while relief priority zone 1 – Highest Priority has a population density between 2464 and 45758. Because quantile classification breaks the data into classes with equal number of feature, relief priority zone 1 – Highest Priority does not offering meaningful results.
Analysis limitations
Although this analysis provides with a basic map of relief priority zone and candidate relief sites, for a real world map other factors might taken in consideration. For a thorough analysis we have to take in consideration that the population density is not constant in a census tract and additional information should be used. Limiting the buffer zone to one mile may exclude important relief sites situated in the immediate vicinity. In an emergency situation, important information to consider would be the hospitals situated within a certain distance from the relief sites where possible victims can be transported to. I created a map that shows the hospitals that are in 10 miles buffer zones from the relief sites situated in central Oklahoma County.
Click on the image to enlarge.
Also, a transportation layer containing the highways and the main roads would have been useful to calculate the shortest path to the nearby hospital. A hydrology layer might have been useful to avoid areas prone to flood when calculating routes to the near hospitals.
Even though our analysis for this project was basic, it proved that can be an important tool for an emergency response team to locate the areas that need intervention first.
Source
King, B., & Zeiders, M. (2009). Problem-Solving with GIS, Lesson 3, Part I,II. The Pennsylvania State University, World Campus. Retrieved February 2, 2010.
Also, a transportation layer containing the highways and the main roads would have been useful to calculate the shortest path to the nearby hospital. A hydrology layer might have been useful to avoid areas prone to flood when calculating routes to the near hospitals.
Even though our analysis for this project was basic, it proved that can be an important tool for an emergency response team to locate the areas that need intervention first.
Source
King, B., & Zeiders, M. (2009). Problem-Solving with GIS, Lesson 3, Part I,II. The Pennsylvania State University, World Campus. Retrieved February 2, 2010.