Skip to content
Back to search results

Catchment Area Analysis

Introduced by
David L. Huff (probabilistic formulation)
Year
1964
Type
Delineation of the geographic area served by a facility
Family
Service and trade area analysis
Font size

What does it do?

Catchment area analysis delineates the geographic area that a facility — a shop, hospital, school, or station — actually serves, turning the abstract question of 'who uses this place?' into a mapped polygon.

Methods range from the simplest fixed-radius buffer through nearest-facility (Voronoi) tessellation and network drive-time isochrones to David Huff's 1964 probabilistic model, in which patronage is shared among competing facilities by their relative attractiveness and distance. The choice of method reflects how strictly customers are tied to the nearest centre and how much competition and travel cost shape real behaviour.

The intuition behind the method

Every facility draws its users from some surrounding territory, but the shape of that territory depends on the assumptions you make. If you assume people simply travel up to a fixed distance, the catchment is a circle. If you assume everyone goes to their nearest centre, the catchments tile the map into Voronoi cells with no overlap. If travel happens on roads rather than as the crow flies, the catchment becomes a lumpy drive-time blob following the network. And if people choose among competing centres by weighing size against distance, no single boundary exists at all — instead each location belongs partly to several facilities according to a probability. Catchment analysis is the toolkit for drawing whichever of these pictures matches reality.

History

Catchment and trade-area analysis grew out of early retail geography, beginning with William Reilly's 1931 law of retail gravitation, which predicted a breaking point between two competing cities. Georgy Voronoi's tessellation and Alfred Thiessen's interpolation polygons provided the nearest-facility partition still used today. David Huff transformed the field in 1963–1964 with a probabilistic gravity model that distributed patronage across competing centres by relative attractiveness and distance, replacing rigid boundaries with continuous probability surfaces. The spread of GIS added network-based drive-time isochrones, and the logic fed directly into spatial-accessibility measures such as the two-step floating catchment area method for constrained services like healthcare.

When to use it

Use catchment area analysis whenever you need to know which population a facility serves — for retail site selection and sales forecasting, for hospital and school enrolment planning, for measuring how many people a transit stop reaches, or for assessing service coverage and gaps. Choose a fixed radius only for quick screening; use Voronoi tessellation when a clean, exhaustive partition under a nearest-centre assumption suffices; use drive-time isochrones when the road or transit network materially shapes access; and use the Huff probabilistic model when facilities compete for the same customers and patronage is shared by attractiveness and distance. It is less suitable when individual choice is driven by factors other than size and distance, or when no facility or population data are available.

Limitations and considerations

  • Fixed-radius and Voronoi methods ignore the road network, barriers, and competition, distorting real catchments.
  • Drive-time and Huff models require detailed network and attractiveness data that may be costly to assemble.
  • All methods are sensitive to the chosen radius, threshold, or decay parameter, which strongly shape results.
  • Catchments are usually static snapshots that ignore temporal variation in travel conditions and behaviour.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Spans a graded toolkit from quick circles to behaviourally rich probability surfaces, fitting many data and question types.
  • Turns facility and population data into directly mappable, decision-ready service areas.
  • Voronoi and drive-time methods produce exhaustive, non-overlapping or network-realistic partitions of space.
  • The Huff model represents competition and shared patronage that deterministic boundaries cannot capture.

Applications

  • Retail site selection and sales-potential forecasting for stores and shopping centres.
  • Planning hospital, clinic, and school service areas and projecting enrolment or caseload.
  • Measuring the population reached by transit stops, stations, and public facilities.
  • Defining business trade areas for marketing, franchising, and territory management.
  • Assessing service coverage, overlap, and gaps to guide new facility placement.

How it works

How it works

1

The simplest catchment is a buffer: a circle of radius r drawn around the facility f_j, capturing every location within that straight-line distance. It requires only the facility location and a chosen radius, and is fast and transparent, which makes it common for first-pass screening and reporting. Its weakness is that it ignores both the road network and the presence of competing facilities, so circles overlap and leave gaps and rarely match true service areas.

Formula
C_j = \{x : \lVert x - f_j\rVert \le r\}
2

Voronoi (Thiessen) tessellation partitions the plane so that every location is assigned to its single nearest facility, producing a set of non-overlapping polygons that exactly tile the study area. It encodes the assumption that consumers always patronize the closest centre and there is no competition or overlap. The cells are built efficiently from the facility points and give a clean, gap-free first model of market areas, but the nearest-facility assumption is often too strict for real choice behaviour.

Formula
V_j = \{x : \lVert x - f_j\rVert \le \lVert x - f_k\rVert\ \forall k\}
3

A drive-time (or walk- or transit-time) catchment replaces straight-line distance with travel time on the actual network, extracting all locations reachable from the facility within a threshold T to form an isochrone polygon. This follows roads, respects barriers, and reflects how people really move, giving far more realistic service areas than circles. It requires a routed network and is the standard GIS approach for accessibility-aware catchments.

Formula
I_j(T) = \{x : t_{net}(f_j, x) \le T\}
4

The Huff model abandons hard boundaries and instead assigns each demand location i a probability P_ij of patronizing each facility j, proportional to the facility's attractiveness S_j (size, floorspace) and inversely related to distance d_ij raised to a decay exponent β, normalized across all competing facilities. A facility's catchment becomes a continuous probability surface rather than a polygon, and expected patronage is obtained by multiplying these probabilities by demand. This captures competition and partial loyalty that the deterministic methods cannot.

Formula
P_{ij} = \frac{S_j\, d_{ij}^{-\beta}}{\sum_k S_k\, d_{ik}^{-\beta}}

Common pitfalls

  • Using straight-line buffers where the network drive-time catchment would look very different.
  • Treating Voronoi cells as true market areas when competition and multipurpose trips break the nearest-centre rule.
  • Borrowing a Huff distance-decay exponent from another context instead of calibrating it locally.
  • Letting overlapping fixed-radius circles double-count population when reporting coverage.

Frequently asked questions

When should I use a Voronoi catchment versus a drive-time catchment?

Voronoi (Thiessen) catchments assume every location goes to its nearest facility measured as the crow flies and partition space into clean, non-overlapping cells, which is useful for a quick exhaustive model. Drive-time catchments instead measure reachability along the actual road or transit network within a time threshold, producing realistic, possibly overlapping service areas. Use Voronoi for a fast first cut under a nearest-centre assumption, and drive-time when the network, barriers, and real travel times matter.

How is the Huff model different from drawing a single catchment boundary?

A single boundary assigns each location entirely to one facility. The Huff model recognizes that nearby competing centres of different sizes share customers, so it assigns each location a probability of visiting each facility, proportional to that facility's attractiveness and inversely to distance. The catchment becomes a continuous probability surface rather than a hard line, letting you estimate expected market share and patronage where deterministic boundaries would oversimplify.

How does catchment area analysis relate to the two-step floating catchment area method?

Catchment area analysis defines the territory a single facility serves; the two-step floating catchment area (2SFCA) method extends this idea to measure accessibility across a whole region when supply is limited. 2SFCA floats a catchment over each service to compute a supply-to-demand ratio, then floats one over each population site to sum reachable ratios. It is essentially catchment delineation applied twice and combined with competition, and is the standard tool for measuring access to constrained services such as healthcare.

Related methods

Sources

  1. Huff, D. L. (1964). Defining and estimating a trading area. Journal of Marketing, 28(3), 34–38.

    Open DOI reference
    Copied.
  2. Luo, W., & Wang, F. (2003). Measures of spatial accessibility to health care in a GIS environment. Environment and Planning B, 30(6), 865–884.

    Open DOI reference
    Copied.