Shelter Site Analysis and Accessibility by Travel Time

Shelter site analysis and accessibility: where should shelters go?

Shelter site analysis is the work of finding places that avoid hazard areas, can be reached on foot and can hold enough people. If you measure by travel time instead of distance, vulnerable areas that a map alone does not show come to light.

In short

  • For a shelter, look at three conditions together: avoiding hazard areas, walking access and capacity.
  • Accessibility is measured more accurately by the actual travel time than by straight-line distance.
  • GIS network analysis can find vulnerable areas that cannot reach a shelter.

A disaster alert says, "Move to the nearest shelter." But when you open a map app, the nearest shelter is 400 meters away in a straight line, and in reality you have to go around a stream, so it takes 15 minutes on foot. Meanwhile, an older person at the senior center who walks with a cane may need more than twice as long on the same route.

Siting a shelter is not just placing a dot on a map. It has to avoid dangerous places, people have to be able to actually walk to it, and it has to hold everyone who arrives. If even one of these three conditions fails, the shelter exists but cannot be used. Today we look at how to check these conditions with GIS (geographic information systems).

Illuminated green emergency exit sign with an arrow pointing right in a dark setting.
Stock photo. Photo: Damir K / Pexels

What condition should you screen out first when choosing a shelter site?

The first thing to screen is whether the site itself is dangerous. Overlay existing hazard maps, such as a landslide hazard map, a flood hazard map (inundation map) and a steep slope failure risk zone map, and check first whether a candidate site falls within them. Areas where the inundation map shows deep flooding, and areas with a high landslide risk grade, are removed from the candidates from the start. Even if a building looks sound, if the ground is weak or the site is low-lying, it can become a place of secondary damage in a disaster. If you skip this step and decide only by capacity or accessibility, the shelter may be flooded or cut off when a disaster actually happens. In practice, it is safer to work in the order of switching on the hazard map layers before drawing up the list of candidates.

Why measure by travel time instead of distance to a shelter?

Straight-line distance is a value measured with a ruler between two points on a map, so it cannot reflect real obstacles such as roads, streams, slopes and traffic lights. The same 500 meters is quick on a flat, straight road, but it feels much farther if you have to go around a stream or climb a steep alley. So GIS uses network analysis, which calculates time along the route you actually have to walk, based on road network data. A map that uses conditions such as walking speed, slope and road width to show "how many minutes from here to the shelter" in color across an area is called a service area map. In situations where evacuation time is tied to lives, this travel time map is a much more realistic basis for judgment than a straight-line distance map.

How do you find areas that cannot reach a shelter with GIS?

The procedure is relatively simple. First, load all shelter locations as a point layer and build a network from road network data. Then, with a network analysis tool, calculate the area reachable on foot from each shelter within a set time (for example, 10 or 20 minutes) and draw it as a polygon (an area-shaped extent). When you merge these polygons and place them over the map of the whole region, the empty areas that fall within the reach of no shelter appear. These empty areas are the "vulnerable areas," and if you overlay them with population density or the share of older residents, the priorities become clear. This is something you learn only by trying: these ranges can change with actual road works or traffic restrictions, so recalculating periodically with the latest road network data is the way to stay accurate. Vulnerable areas found this way are used as evidence for choosing new shelter candidates or for placing mobile evacuation means.

Why does the result change when you consider older adults and people with limited mobility?

The "walking speed" used in network analysis is often set by default to that of an ordinary adult. But older adults, wheelchair users and caregivers with small children take longer to cover the same distance, and stairs or steep slopes may be impassable altogether. To reflect these conditions, you either lower the walking speed and recalculate, or exclude sections where the slope is above a certain level from the network and analyze again. Then it is not rare for an area that was "reachable" on the general standard to turn into a vulnerable area on the older adult standard. In fact, the higher the share of older residents, the bigger this difference, so overlaying population statistics by area is essential. Shelter capacity planning is also adjusted to this result, for example by designating facilities with flat approach routes first near areas with many people who have limited mobility.

How do you check and reflect shelter capacity?

Capacity is how many people a facility can take at one time, and it is usually calculated from the building's gross floor area or from the number of people per designated standard area. However good the accessibility, if the number of arrivals exceeds capacity, people have to wait outside the shelter, so you need to look at travel time analysis and capacity data together. In GIS, you add a capacity attribute to each shelter polygon and check for overload by comparing it with the population living inside that shelter's reach area. If an area has a population far above capacity, you need to add a supporting shelter nearby or reassign the areas. Base this calculation on the usual population, but in places where people gather, such as tourist spots and commercial districts, also consider changes by time of day to get closer to a real disaster.

Step by step on screen

Steps to follow

  1. Load the layers
    In your mapping program, load the shelter point data and the road network line data.
  2. Overlay the hazard areas
    In the Layers panel, switch on the inundation map and the landslide hazard map together and check which candidate sites overlap them.
  3. Build the network
    In your network analysis tool, convert the road network layer into a network dataset.
  4. Calculate travel time
    In the Service Area analysis menu, set the shelters as starting points and set 10-minute and 20-minute intervals. In QGIS, look in the Processing Toolbox under Network analysis.
  5. Adjust the speed condition
    In the analysis settings, lower the walking speed value and recalculate for older adults.
  6. Overlay and check the results
    Overlay the calculated reach polygons and the population distribution layer to check the vulnerable areas.

Menu names and positions may differ depending on the program and version you use. If you have trouble finding one, try the search function for "network analysis" or "service area." The figure above is a diagram of the sequence.

Prompts you can copy

A prompt for organizing shelter travel time analysis results into a report

You are an assistant who writes GIS analysis reports in the disaster safety field.
Reflect the conditions below and organize the shelter accessibility analysis result for [area name] into a draft report.
Input information: number of shelters [count], share of population reachable within 10 minutes [ratio], vulnerable areas beyond 20 minutes [list]
Describe areas with a high share of older residents separately.
Format: in the order 1) purpose of analysis 2) main results 3) status of vulnerable areas 4) improvement proposals, with concise sentences
Do not make up speculative figures or unconfirmed statistics; mark them [needs confirmation].

What to check: compare against the original map to see whether the figures and area names the prompt produced match the actual analysis results, and whether any vulnerable areas are missing from the list.

Replace the bracketed parts for your situation. Do not use the answer as it is; check the facts first.

Try it yourself

  • Search for the shelters in your neighborhood on a public map service and compare the straight-line distance with the actual walking route.
  • Walk and time the difference in how long it feels over the same distance on a route with a slope or a stream and on a flat route.
  • Open the inundation map or landslide hazard map of your area and check that the nearest shelter does not overlap a hazard zone.

Common questions

Is the nearest shelter always the best?
Being close is not enough. You also need to see whether the route is actually safe and quick to walk, and whether the shelter itself is outside a hazard area.

Where can I get the road network data used in travel time analysis?
In many cases you can get a road network layer from open map data or from public data such as a national spatial data portal. However, how current and accurate it is depends on the source and its update cycle, so it is good to check it yourself after you download it.

Do I really need to run a separate analysis for older adults?
If the share of older residents in the area is high, the result can change a lot, so it is safer to run it separately. Judging only from the general-standard analysis can miss areas that are in fact hard to reach.

What comes next

A shelter site shows its real shape only when you overlay the three conditions of avoiding hazard areas, travel time and capacity, and if you recalculate for older adults and people with limited mobility in particular, the map looks completely different. A later post in this series covers how to newly place shelters or evacuation routes in the vulnerable areas found this way.

This is an English adaptation of a post in the Korean GeoAI basics series. The original is here, and more learning materials are on GeoAI for Disaster Safety.

This post was drafted with the help of AI tools and reviewed by the author. Tool features and data portals change over time, so check the official documentation when you practice.

Comments

Popular posts from this blog

What Is GeoAI, and How Does It Help Disaster Safety?

Getting Started with QGIS: What to Do First

Whitebox QGIS Practice 26 — Elevation Statistics of Two Basins: Means of 2,000.1 m and 1,924.3 m, a 76 m Gap