San Francisco: Where to build a new fire station?

Current Visualization:
How to use this visualization (more info)

Red areas of the map have low spatial accessibility. Practically, this means that the number of fire units available to respond per-incident is low. When there are simultaneous fire incidents, these areas may require units from much farther away to respond, delaying a potentially life-saving response. Naively, optimal placement of a new fire station would minimize the amount of red on the map (i.e. more even spatial accessibility). However, this must also be combined with domain-expertise about current challenges facing the fire department.

In placing a new station, one large consideration is the area that it can cover in less than 5 minutes. 85% of 2019 fire responses had a unit on scene in less than 5 minutes. When hovering over the map to place a new simulated station, the estimated area that can be covered in less than 5 minutes is shown. Note that proximity to highways, local traffic density, street conditions (e.g. single lane streets), and terrain (e.g. hills etc) will greatly impact the size of this area.

For existing fire stations, hovering over the map shows the number of fire stations within 5 minutes in Data for location at the top of the map (as estimated from 2019 data via the KDE).

:
  • This visualization shows the daily rates of fire department incidents in San Francisco. A "fire department incident" is any incident that a fire engine or fire truck responded to. Each grid square represents about 1/3rd of a mile squared.
  • This visualization allows you to simulate the placement of a new fire station in San Francisco. The visualization displays the results of the Two-Step Floating Catchment Area (2SFCA) algorithm, both on the observed data and after adding a simulated firestation at the specified location.
  • Hover over the map to see the estimated area the new fire station could respond to within 5 minutes, if a new station is placed in that area. The estimated area shown is sourced from the BingMaps API. Note that proximity to interstates and highways will greatly affect the size and shape of this area.
  • Click an empty location on the map to simulate placing a fire station at that location. The size of the impact (colors) is determined by the 2SFCA algorithm. The data is precomputed in the Python notebooks created for this project.
  • Click an existing fire station (●) to see the area that its fire units can reliabily reach within 5 minutes. The size and shape of this area is determined via Kernel Density Estimation. The size and shape of the area is greatly influenced by proximity to major highways and traffic conditions. Long and skinny protrusions from shapes show the impact that highways and interstates have. The estimates shown on the map are used as an important input into the 2SFCA algorithm, to determine whether the stations in an area are "spread thin" relative to the fire incident demand.
Map Legend
Existing Fire Station
Simulated Fire Station
Existing Fire Station Coverage
(<= 5 minute travel time)
click fire station to show
Simulated Fire Station Coverage
(<= 5 minute travel time)
Spatial Accessibility (2SFCA):
+00% Change in spatial accessibilty due to simulated firestation at location.
Data for location(based on 2019)
# Daily fire incidents (info) Average number of daily incidents at the location, which required a fire truck or fire engine to respond. : -
# Fire stations within 5 minutes (info) Number of fire stations that can respond with a truck or engine within 5 minutes. This is provided by kernel density estimates for each fire station to understand the geographic region each can respond to within 5 minutes (with high probability). This is not updated to include the simulation value, but if the grid square is affected by the simulation (and has a +00% icon), you can add 1 to see the effect. : -
# Fire stations per incident (info) Ratio of the number of fire stations with 5 minutes to daily incidents. : -
Spatial Accessibility (current) (info) Results of the 2SFCA algorithm. Unlike the ratio of fire stations to incidents, this estimate considers if nearby zones ALSO have high incident rates and therefore fewer fire stations are available to respond. : -
Spatial Accessibility (new) (info) Results of the 2SFCA algorithm applied to the simulation, when a new fire station is placed in the selected square. : -
% increase spatial accessibility (info) The increase in spatial accessibilty under the simulation : -