Which site for the new station maximizes fire protection coverage?

Note:

This example presents an approach to using ArcGIS to find the best location for a new facility (a fire station, in this case). The process uses location-allocation analysis to identify the site that maximizes coverage. While the data is real, the scenario, analysis, and resulting decisions are hypothetical. This is just one approach to using spatial analysis tools for siting a new facility. Your criteria and decision-making process may differ.

Mike Price of Entrada/San Juan Inc. and Elise Fisher of EF Geographic contributed to this case study.

An increasing need for fire protection

A recent article in the local paper reported on fire response times in the City. The article claimed that while much of the City is within a four-minute travel time of one of the City's three fire stations, portions of the City are six, eight, or even twelve minutes from a station. The four-minute travel time standard used by the reporter who wrote the story was based on one developed by the National Fire Protection Association. It sets a goal for urban areas that units respond to incidents within a four-minute travel time 90 percent of the time. (An additional minute is allocated up front for turnout—taking the call from dispatch, putting on protective gear, and boarding the truck.)

Existing fire station service areas using a four-minute travel time

Read the case study about the fire station response time the reporter did for her story.

At the City Council meeting following the publication of the article, the Director of Public Safety briefed the council members on options for reducing emergency response travel times in the City. The Director noted that it has been over a decade since the last fire station was constructed. During that time, the population has increased slowly but steadily. The Director indicated that the most effective action for the City to take to reduce overall emergency response times would be the construction of a new fire station in an area that is currently underserved. At the meeting, the City Manager noted that all capital improvement projects have been on hold for the past several years due to decreased tax revenues and steep budget cuts. Several City Council members pointed out that there are a number of City-owned parcels that are vacant, and building a new station on one of these parcels could help mitigate the cost.

Following the meeting, the City's Planning Director, anticipating that a new station would eventually be built, had her staff undertake an analysis to see if any of the vacant, City-owned parcels would be suitable for a new fire station. The analysis identified eight potential sites. This was a first step in the site selection process—additional spatial analysis, along with input from the citizenry, will help the City Council decide on the final site (if indeed a new fire station is approved).

The site-selection process identified eight potential sites for the new station (the labels indicate the site ID).

Read the case study about the analysis used by the planners to identify suitable sites.

At the next council meeting, the members did in fact vote to budget for the construction of a new fire station in the coming fiscal year, despite the City Manager's concerns about the cost. The members cited the safety of the City's residents and a recent increase in City revenues. After the vote, the Planning Director presented the potential sites her staff had identified. The council members indicated they would review the potential sites but, in the meantime, tasked the Planning Department with determining which of the sites will result in the greatest increase in the portion of the City within the four-minute travel time. The Director of Public Safety noted that the four-minute travel time standard is generally applied to dense urban areas and that in rural areas, such as the northern and eastern parts of the City, a common standard is to respond to incidents within a fifteen-minute travel time. Nonetheless, the City Council decided to stick with the four-minute travel time so they can demonstrate to residents that the City is using the highest possible standards.

Establishing a baseline

Given that the locations of future incidents cannot be known with certainty, the planners need to identify a way to measure how much of the City will be covered within four minutes by four fire stations—the three existing stations and the one new station. The City Council did not provide guidance on how to measure coverage of the City within the four-minute travel time. After discussions with the City's Public Works Department, the Fire and Police Departments, and the City Manager, the planners settle on three measures:

• The total population served
• The total value of the structures served
• The miles of streets covered

The primary mission of the Fire Department is the protection of lives, hence the first measure is population. The department is also tasked with preventing, or limiting, monetary losses to structures. Streets represent access to emergencies at homes and businesses; in addition, some emergencies (such as traffic accidents) occur along streets.

The City Council will likely use some combination of these measures to make their decision on the best location for the new station.

First, the planners want to establish the current coverage of the City given the existing fire stations—this will become the baseline and will allow the City Council to see the improvement in coverage when the new station is included.

To get numbers for the baseline, the planners use the ArcGIS Location-Allocation tool. This tool assigns locations to their closest facility in a way that maximizes the number of locations covered while minimizing overall travel time, using drive time over streets. The tool can also measure an attribute value associated with the locations. For example, it can assign census blocks to their closest fire station while ensuring that the greatest number of people are covered.

The planners know that the key to getting meaningful results is having a reliable street network dataset. Starting with their existing layer of streets, they create a network dataset that includes over and underpasses. For this analysis the planners will be calculating travel times, so they make sure the network dataset accounts for the time required to make a turn at an intersection. Since portions of the City are hilly, they decide to also include a factor to account for the increase in travel time on steep streets.

Read white paper for more on creating a street network dataset for emergency response.

With the street network dataset complete, the planners are ready to start calculating the baseline measurements. The first measure the planners calculate is population covered. They run the location-allocation analysis using the existing fire stations and the locations of census blocks, represented by points with an attribute for the number of people living in each block (in location-allocation analysis, both facilities and locations are represented by point features). The Location-Allocation tool assigns each census block to its closest fire station (within a four-minute travel time over the street network) and sums the population of the assigned blocks.

GIS assigns each census block to its closest fire station, as indicated by the red lines (keeping in mind that actual drive time is calculated over the street network). Blocks that are more than four minutes from a station or that have no population, such as in a business park, are not assigned. The resulting table shows the number of census blocks assigned to each station (DemandCount) and the summed population in those blocks (DemandWeight).

The planners then run the location-allocation analysis for the other two measures—the total value of properties covered and total miles of street—to complete the baseline figures.

To calculate the total value of structures covered, the location-allocation analysis uses the locations of individual land parcels (again represented as point features), summing the assessed value of structures. The DemandCount is the number of parcels assigned to each station and the DemandWeight is the total assessed value of structures for those parcels.

To calculate the total miles of streets covered, the location-allocation analysis uses the locations of street segments, summing the length of segments (since locations must be point features, each street segment is represented as a point, with the length of the segment as an attribute). The DemandCount value is the number of street segments assigned to each station and the DemandWeight value is the total miles of street.

The planners enter the results of the three location-allocation analyses into a spreadsheet and create a summary table that calculates the totals for all three stations and the percent coverage for the City. (They calculate the total population, total value of structures, and total miles of streets using the attribute tables for the census blocks, parcels, and streets layers, respectively.)

According to the analysis, the existing stations cover 57 percent of the population and just over 50 percent of the value of the structures and miles of streets in the City, within a four-minute travel time.

Finding the optimum site

The next step in finding the optimum location for the station is to determine which of the eight sites—in conjunction with the existing stations—will provide the most coverage of the City. They run the same location-allocation analysis they used earlier for the baseline numbers, but this time with the three existing stations and the eight candidates. The analysis identifies the optimum solution for each measure—that is, the combination of four station locations (three existing plus one candidate) that provides the most coverage.

For total population covered, the location-allocation analysis identifies site 26 as the best location. The planners find that locating the new station at this site would increase the coverage of the population within four minutes to 67 percent.

The location-allocation analysis identifies site 26 as the location that, in conjunction with the existing fire stations, covers the most people within four minutes of a fire station. As before, the DemandCount is the number of census blocks within four minutes of a station and the DemandWeight is the population in those blocks.

For the other two measures, though, the location-allocation analysis identifies site 6 as the best location.

The location-allocation analysis identifies site 6 as the best location when the value of structures is used as the measure of coverage, as shown in the map and table. The same is true when the measure is miles of streets.

As it turns out, site 6 is across the street from site 26. The planners postulate that the totals for the two measures (value of structures and miles of streets) with site 26 should be only slightly less than with site 6. To confirm this they remove site 6 from the list of potential sites and rerun the location-allocation analysis for these measures. The analysis does indeed identify site 26 as the next best location. When they calculate the percent coverage, they find that the numbers are virtually the same for both sites on both measures: 60percent for the value of structures and 62 percent for miles of streets covered.

The planners surmise that the City Council and the residents are likely to support the site that results in the greatest number of people being within four minutes of a fire station, with coverage of buildings and streets secondary considerations. Since site 26 covers more people in conjunction with the existing stations, they decide to recommend that site for the location of the new station and reserve site 6 as a backup location.

Evaluating the results

The planners modify their spreadsheet to show the percentages with the new station located at site 26.

The table shows the improvement in percentage covered for each of the three measures, assuming the new station is built at site 26.

They create maps and a chart for the Planning Director, City Manager, and Director of Public Safety to review along with the table.

The maps show the area currently within a four-minute travel time of a fire station (left) and the area that would be covered, assuming a new station is built at site 26 (right), while the chart shows the improvement in percentage covered for each of the three measures with the new station.

The public review

The City Manager presents the results of the analysis to the City Council. The Council members agree to pursue site 26 as the likely location for the new station.

The decision is backed up by the Fire Department, which reports that their historical incident data shows that the highest risk area is the high population density area between Station 51 and Station 53. Building the new station at site 26 will reduce the load on Station 51. The new station will also be close enough to the higher risk area to provide backup to the two existing stations.

The City Council direct the City Manager to solicit feedback from the community regarding the proposed site. To help with that effort, the City Planners post a story map on the City's website showing the preliminary results of the analysis. The maps will help community members understand the site-selection process to date.

After monitoring the website for several weeks, the City Manager finds that feedback from residents in the southern part of the City is generally positive. In their comments, people refer to the population density map and also point out that there are several schools that are currently beyond the four-minute travel time that would now be within four minutes of the new station if it were built at the proposed site.

Residents in the central and eastern parts of the City, however, express concern that, with the new station built at the City's preferred site, there will be three fire stations in the southern half of the City, one in the north, and none in the central area. They point out that many people will still be relatively far from a fire station. Further, they argue that houses in their neighborhood have high values (compared to houses in neighborhoods closer to the proposed site) and a fire in their area would lead to greater monetary losses.

Given the feedback from the residents in the central and eastern parts of the City, the City Council directs the Planning Department to look into possible sites for the new fire station in those areas.

Back to the drawing board

Since the original site-selection process turned up no suitable City-owned parcels in those areas, the new site would have to be on private property. To speed up the search process and keep the residents involved, the City staff follow up with residents and homeowners associations in the area to solicit suggestions for possible sites for a new station. The staff compile a list of the most frequently mentioned sites. Several vacant parcels in the area are at the top of the list.

First, the planners evaluate the privately owned parcels to see if they meet the City's criteria for a location for a fire station, as they did earlier for the City-owned sites. As it turns out, two of the sites are in fact suitable.

Two privately owned sites in the central part of the City meet the suitability criteria. The planners assign labels to the sites using the names of adjacent streets: Northcrest and Ridgeview.

If either of the sites improve coverage more so than City-owned site 26, the City might consider purchasing the site outright or possibly obtaining the property through eminent domain. To find out, the planners run the location-allocation analysis with the two privately owned sites (along with the existing fire stations) to find out how well they do. One of the sites, dubbed the Ridgeview site, does the best on all three measures. To make it easier for the City Council to assess the results, the planners add the Ridgeview site to their chart to show the percentage coverage for the Ridgeview site, along with the percentages for the baseline and site 26.

The revised chart shows the improvement in percentage covered for each of the three measures with the new station at site 6 or at the Ridgeview site.

The chart shows that the Ridgeview site does, in fact, cover a slightly higher percentage of the value of structures in the City than site 26. This supports the claim made by the residents from the central and eastern parts of the City. However, it falls short on total population and barely matches site 26 on miles of streets covered.

The planners create a map of the four-minute travel times from the Ridgeview site with the three existing stations. The City Council will be able to compare this map to the travel time maps for site 26 that they received earlier.

The map shows the area that would be within a four-minute travel time, assuming a new station is built at the Ridgeview site.

The final decision

The charts and maps are sent to the Director of Public Safety, the City Manager, and City Council members for review. The City Manager looks into the ownership and cost of the Ridgeview site. The assessed value of the Ridgeview site is $163,000 (the market price could be considerably higher).

Given the cost and time involved in obtaining the property, even through eminent domain, and the fact that fewer people would be covered in the four-minute travel time than with the proposed City-owned site—site 26—the City Council decides not to build the new station at the privately owned Ridgeview site. However, they will keep the site as an additional backup location.

The council decides to continue the assessment of site 26. They direct the City Manager to hire a consultant to conduct a site analysis and prepare an environmental impact report.

Anticipating pushback from some segments of the community, the City Manager has his staff work with the City Planners to create a story map for the City's website, explaining the project background, the decision-making process, and the next steps the City will take to improve emergency response coverage.

The City Manager posts a story map to the City's website that explains the decision-making process for selecting the fire station location.

Given that the City Council members are elected at large from across the City and voter turnout is always high in the wealthier central and eastern portions of the City, the City Council passes a nonbinding resolution that the next new fire station to be built will be located in the central part of the City. They direct the City manager to continue analyzing possible sites for a station in that area—starting with the Ridgeview site—and given the fact that property values are starting to rise again, to look into sources of funding and the possibility of purchasing land in the near term.

Workflow using ArcGIS Desktop

In ArcGIS Desktop you use the Network Analyst location-allocation solver to assign demand (in the form of population, value of structures, or miles of streets) to facilities (that is, existing fire stations and potential sites). You will need to have ArcGIS Network Analyst extension installed and a street network dataset that contains travel times. Learn how to configure the Network Analyst environment

In Network Analyst, the demand features and the facilities are represented as point features. In this example, the facilities—the existing fire stations and the potential sites for the new station—are already stored as point features. However, the census blocks and parcels are polygon features and the streets are line features. Before continuing, use the Feature to Point tool.

Calculate the baseline values

1. First, add the Network Analyst toolbar and create a New Location-Allocation.

   

2. Open the Network Analyst window. Right-click Facilities, click Load Locations, and select Existing fire station as the layer to load from. Set the FacilityType to Required so all three stations will be included in the solution. (This is not strictly necessary at this point since there are three facilities and you'll specify that all three be chosen; however, when you include the potential sites in the next part of the analysis, the existing stations need to be set as Required; otherwise, four candidate stations may be chosen and the existing stations not included.)

   

3. In the Network Analyst window, right-click Demand Points and select Load Locations. The first baseline will be for population, so choose Census block point location as the layer to load from. Specify FID as the Name and TOTPOP as the Weight field.

   

4. Right-click Location-Allocation in the Table Of Contents and select Properties. On the Analysis Settings tab, set the Impedance to MIN_SLOPE (Minutes). This will calculate travel times using minutes adjusted for the slope of the street.

   

5. On the Advanced Settings tab, set the Problem Type to Maximize Coverage, the Facilities To Choose to 3 (for the three existing fire stations), and the Impedance Cutoff to 4 (to calculate the population within a four-minute drive time of a station). Click OK to save the settings.

   

6. Click Solve on the Network Analyst toolbar to run the location-allocation analysis. (A warning message will list any demand points with no weight value. In this example, these are census blocks with no population—such as those in an industrial area.)

   

7. In the Table Of Contents, open the attribute table for the Facilities to display the population within four minutes of each fire station, indicated by the DemandWeight field.

   

8. You'll probably want to save the values, so export the table before continuing on to calculate the location-allocation values for the other two measures. (When you do the export, be sure to specify the location where the project data is stored.)
9. To find the total value of structures within four minutes of each fire station, delete the Demand Points (the census blocks) and load the Parcel point location layer. Specify APN as the Name and ASR_IMPR (assessed value of improvements) as the Weight field. All the other analysis parameters remain the same. Click the Solve button on the Network Analyst toolbar and open and export the resulting table for the Facilities layer.

   

10. To find the total miles of streets within four minutes of each fire station, delete the Demand Points (the parcel point locations), as shown above, and load the street point locations, using MILES as the Weight field. Once again, click the Solve button and open and export the resulting table.

    

Find the optimum location for the new fire station

1. To find the location that increases the coverage of the City within four minutes to the greatest extent, add the potential sites to the list of Facilities. Specify Facilities_Name as the Name, and accept the default FacilityType of Candidate.

   

2. Open the Location-Allocation Advanced Settings tab (as you did earlier) and increase the Facilities To Choose to 4 (the three existing stations plus one new location to be chosen).

   

3. Now repeat the analysis as described above for calculating the baseline values for each of the measures, by loading the census block point locations, parcel point locations, and street point locations as Demand Points, in turn, setting the appropriate name and weight field values for each and solving the location-allocation. (You can start with the street point locations since they're still loaded as Demand Points.) This time the resulting Facilities table for each measure will include the three existing fire stations along with the chosen candidate site that results in the best coverage for that measure. The other candidate sites are also listed.

   

4. To find the next best candidate for a given measure, remove the chosen facility from the Facilities list by right-clicking and deleting it. Click the Solve button to rerun the location-allocation analysis.

   

Create the drive-time areas

1. To create the four-minute drive-time area for the existing fire stations, on the Network Analyst toolbar, create a New Service Area.

   

2. In the Network Analyst window, right-click Facilities and select Load Locations to load the fire stations.

   

3. Open the Layers Properties for the Service Area, and on the Analysis Settings tab, specify the Impedance as MIN_SLOPE and the Default Breaks as 4 (minutes).

   

4. On the Polygon Generation tab, specify detailed polygons trimmed at 100 meters from streets and not overlapping.

   

5. Click Solve on the Network Analyst toolbar to display the drive-time areas. You'll want to save the polygons by exporting them before continuing.

   

6. To create the drive-time areas, including site 26 (the optimal location identified earlier), you need to load the site as a facility along with the existing fire stations. To do this, select site 26 on the map, save it as a layer file, and add it to the Facilities using Load Locations. Click Solve on the Network Analyst toolbar to display the new drive-time areas. Once again, export the polygons to save them.

Evaluate the privately owned sites

1. If you want to evaluate the privately owned sites on the three coverage measures, follow the same steps as for the City-owned sites. Load the privately owned sites along with the existing fire stations into the Location-Allocation solver. Load each of the demand point layers representing the measures, in turn, and solve the location-allocation for each.
2. Once the optimum site has been identified (the Ridgeview site), select and save it as a layer and load it into the Create Service Areas solver, along with the existing fire stations, to create the drive-time area.

Workflow using ArcGIS Pro

In ArcGIS Pro you use Location-Allocation (under Network Analysis on the Analysis tab) to assign demand (in the form of population, value of structures, or miles of streets) to facilities (that is, existing fire stations and potential sites). You will need to have the ArcGIS Network Analyst extension installed and a street network dataset that contains travel time (the case study data available for download includes such a dataset).

In Location-Allocation analysis, the demand features and the facilities are represented as point features. In this example, the facilities—the existing fire stations and the potential sites for the new station—are already stored as point features. However, the census blocks and parcels are polygon features and the streets are line features. Before continuing, use the Feature to Point tool to create point features from each of these three layers, in turn.

Calculate the baseline values

1. First, from the Network Analysis dropdown (on the Analysis tab) create a Location-Allocation layer.

   

2. On the Location-Allocation tab click Import Facilities to open the Add Locations tool.

   

   Select Existing fire station as the Input Locations. Under Field Mappings, select FacilityType (under Property) and set the Default Value (under Field) to Required so all three stations will be included in the solution. This is not strictly necessary at this point since there are three facilities and you'll specify that all three be chosen. However, when you include the potential sites in the next part of the analysis, the existing stations need to be set as Required. Otherwise, four candidate stations may be chosen and the existing stations not included.

   

3. On the Location-Allocation tab click Import Demand Points. The first baseline will be for population, so choose Census block point location as the Input Locations. Under Field Mappings, specify ORIG_FID as the Name and TOTPOP as the Weight field.

   

4. On the Location-Allocation tab, set the Mode to Time in Minutes Weighted by Slope (this will calculate travel times using minutes adjusted for the slope of the street). Set the Cutoff to 4 (to calculate the population within a four-minute drive time of a station), the Facilities to 3 (for the three existing fire stations), and the Type to Maximize Coverage.

   

5. Click Run.

   

6. Open the attribute table for the Facilities to display the population within four minutes of each fire station, indicated by the DemandWeight field. You'll probably want to save the values, so export the table before continuing on to calculate the location-allocation values for the other two measures (select Export from the menu in the upper right corner of the table). Be sure to specify the location where the project data is stored.

   

7. To find the total value of structures within four minutes of each fire station, click Import Demand Points and load the Parcel point location layer. Specify APN as the Name and ASR_IMPR (assessed value of improvements) as the Weight field. Be sure to uncheck the Append to Existing Locations box so the parcel point locations replace the census block point locations you imported earlier.

   

   Once the parcel point locations have been imported click Run (all the analysis settings you specified earlier remain the same). Then open and export the resulting table for the Facilities layer.

8. To find the total miles of streets within four minutes of each fire station, import the street point locations as demand points (make sure Append to Existing Locations is unchecked). Specify MILES as the Weight field. Once again, click the Run button and open and export the resulting table.

Find the optimum location for the new fire station

1. To find the location that increases the coverage of the City within four minutes to the greatest extent, import the candidate sites as Facilities. Specify Facilities_Name as the Name, and accept the default FacilityType of Candidate. Make sure the Append to Existing Locations box is checked so the candidate sites are added to the existing fire stations.

   

2. On the Location-Allocation tab, increase Facilities to 4 (the three existing stations plus one new location to be chosen).

   

3. Since the street point locations are currently loaded as demand points you can find the optimum location using miles of streets first. Click Run to run the Location-Allocation analysis. This time the resulting Facilities table includes the three existing fire stations along with the chosen candidate site that results in the best coverage for this measure. The other candidate sites are also listed.

   

4. To find the next best candidate for a given measure, temporarily delete the chosen facility from the Facilities table by right-clicking it and selecting Delete Row. Do not save your edits at this point.

   

   Click the Run button to rerun the analysis. Location-Allocation will find the optimum location from among the remaining candidates—the second best choice.

   Before continuing, click the Edit tab and click Discard. That will restore the candidate you deleted back into the Facilities table so you can run the analysis for the next measure with the full set of candidates.

   

5. Now repeat the analysis for the other two measures: total population and value of structures. Import the census block point locations as demand points (setting the appropriate name and weight field values) and run the analysis to find the best and next best candidates. Then do the same for the parcel point locations. When importing the demand points make sure the Append to Existing Locations box is unchecked so the current demand points are replaced each time.

Create the drive-time areas

1. To create the four-minute drive-time area for the existing fire stations, use the Network Analysis dropdown and choose Service Area.

   

2. On the Service Area tab click Import Facilities and add the existing fire stations.

   

3. Now set the analysis parameters. For Mode, specify Time in Minutes Weighted by Slope. Specify a single Cutoff of 4 (minutes) and specify that the output polygons be Split.

   

   Click Run to create and display the drive-time areas. Before continuing, you'll want to save the polygons by exporting them.

   

4. To create the drive-time areas including site 26 (the optimal location identified earlier), you need to load the site as a facility along with the existing fire stations. To do this, select site 26 on the map, save it as a layer file, and add it to the facilities using Import Facilities. Make sure to check the Append to Existing Locations box. Click Run on the Service Area tab to display the new drive-time areas. Once again, export the polygons to save them.

Evaluate the privately owned sites

1. If you want to evaluate the privately owned sites on the three coverage measures, follow the same steps as for the City-owned sites. Import the privately owned sites along with the existing fire stations into Location-Allocation. Load each of the demand point layers representing the measures, in turn, and solve the location-allocation for each.
2. Once the optimum site has been identified (the Ridgeview site), select and save it as a layer and import it as a Service Area facility, along with the existing fire stations, to create the drive-time area.