Geographic datasets often describe the same city using boundaries that refuse to line up.
Population may be published by census zone. Schools serve catchments with different borders. Health services use another geography. Flood models use drainage basins. Planning uses administrative areas. If we want to ask how many residents of one census system fall inside another system, the numbers cannot simply be copied across.
Areal interpolation is the process of estimating how values recorded for one set of geographic zones should be transferred to a different set of zones.
Areal interpolation is what Geography does when the data and the question were drawn with different boundaries.
Quick Read: The Areal-Interpolation Mechanism
SOURCE ZONES + SOURCE VALUES → OVERLAP / AUXILIARY INFORMATION → ESTIMATION → TARGET ZONES + UNCERTAINTY
1. Source and Target Zones
The source zones are the areas for which values are known. The target zones are the areas for which values are needed. The task is to move information from source to target without inventing more certainty than the evidence allows.
2. Simple Area Weighting
A basic method allocates a source-zone value according to the proportion of its area overlapping each target zone. If forty per cent of a source polygon lies inside a target polygon, forty per cent of an extensive variable such as population might be assigned there.
3. The Uniformity Assumption Is Strong
Area weighting assumes the variable is distributed evenly inside the source zone. That may be poor for population if half the zone is a reservoir, industrial estate or park.
4. Extensive and Intensive Variables Behave Differently
Counts such as population can be redistributed while preserving totals. Rates, percentages and densities cannot simply be split the same way. The mathematics must respect what the variable means.
5. Dasymetric Mapping Adds Better Geographic Knowledge
If land-use or building data show where people can plausibly live, estimates can be allocated more intelligently than by raw area. A large forest or reservoir can be excluded from the residential distribution.
6. Auxiliary Data Can Improve the Transfer
Building footprints, addresses, land cover, night lights or parcel data can help reveal where the source-zone value is likely concentrated. Better auxiliary information can reduce, though never entirely eliminate, uncertainty.
7. Areal Interpolation Is Not Point Interpolation
Spatial Interpolation owns prediction between measured points or locations. Areal interpolation owns transfer between incompatible polygon systems.
8. MAUP Is Upstream of the Problem
MAUP warns that zone construction changes statistics. Areal interpolation deals with the practical consequence: analysts often need to reconcile data already published under incompatible zone systems.
9. Totals Should Be Conserved When Appropriate
If 10,000 residents exist in the source zones, a population interpolation should not quietly create 10,500 residents in the target zones. Pycnophylactic and related approaches emphasise preserving totals while redistributing them spatially.
10. Boundary Slivers Can Create Noise
Tiny polygon overlaps caused by geometry precision can receive tiny allocations that are mathematically valid but practically meaningless. GIS preprocessing and topology therefore matter.
11. Primary Geography: Two Different Grids
Place ten counters inside four large boxes, then overlay a different set of boxes. Ask how many counters belong to each new box. If the counters are hidden and only old box totals remain, children immediately see why estimation is required.
12. Secondary Geography: Population Into Catchments
Students can ask how population data published for one geography might be estimated for school or clinic service areas. They should state which assumptions are necessary and where those assumptions are weakest.
13. Advanced Geography: Validate Against Finer Data
Where finer-resolution observations exist, they can test the interpolation. Comparing estimated target-zone values with known values reveals whether area weighting, dasymetric methods or other models perform better.
14. Singapore Example: Planning and Service Geographies
Population, transport, school, health and environmental questions may use different spatial units. When data must be translated across those systems, the transfer method becomes part of the evidence rather than a clerical afterthought.
15. Flood Example
Population data reported by administrative zones may need to be estimated within flood-risk polygons. Simple area weighting can overassign people to uninhabited land; building or land-use information can improve exposure estimates.
16. Healthcare Example
Health statistics may be reported under one boundary system while service planning uses another. Transferring rates carelessly can create false precision, especially when populations differ strongly inside source zones.
17. Historical Geography Example
Administrative boundaries change through time. Comparing a modern district with a historical census may require translating old zones into new ones. The apparent trend can depend partly on how that translation is performed.
18. Hostile Test: “The GIS Software Gave Me a Number”
A number is not a measurement merely because software produced it. Ask which source values were known, which assumptions redistributed them and how much uncertainty the boundary mismatch introduced.
19. Where Areal-Interpolation Reasoning Breaks
- Uniformity fantasy: assuming values are evenly distributed inside source zones.
- Rate-count confusion: redistributing percentages as though they were counts.
- Boundary innocence: treating zone mismatch as a harmless technical detail.
- Auxiliary-data worship: assuming more layers automatically make the estimate correct.
- False precision: reporting highly exact target values from coarse source zones.
- Total leakage: failing to preserve counts when the variable requires conservation.
20. Ten Questions for Areal Interpolation
- What are the source zones?
- What are the target zones?
- Is the variable a count, rate, density or average?
- What internal distribution is assumed?
- Can land use or buildings improve the estimate?
- Should totals be preserved?
- How much do boundaries overlap?
- Where are source zones internally heterogeneous?
- Can the method be validated?
- How should uncertainty be communicated?
21. Where This Fits
MAUP owns aggregation instability. Spatial Interpolation owns prediction across unmeasured space from locations. This article owns the translation of data from one polygon geography to another.
The Idea to Keep
When boundaries do not match, the transferred number is an estimate—and the boundary problem belongs inside the answer.