What Is a Heat Map?

In geographic mapping, a heat map usually turns many point observations into a continuous-looking surface of intensity. Areas where points are close together appear stronger or “hotter”; areas with few nearby points appear weaker.

The method is useful when the pattern of concentration matters more than the identity of each individual point.

From points to intensity

A heat map assigns influence around each point, and as nearby influences overlap, clusters accumulate more intensity than isolated observations.

A radius or bandwidth controls how far each point contributes. Small radii preserve local hotspots but can create a noisy surface, whereas large radii produce smoother regional patterns but can merge distinct clusters.

That parameter is part of the interpretation. The heat map does not simply reveal a pattern that existed independently of the method.

Weighted and unweighted heat maps

An unweighted heat map treats each point equally, while a weighted heat map gives points different influence based on a numeric value.

For example, one map might show the concentration of hospitals, where each hospital contributes one observation. Another might weight hospitals by bed capacity, creating an intensity surface of capacity rather than facility count.

Those two maps answer different questions even though the source point locations are identical.

Heat map does not mean “any map with hot colours”

The term is sometimes used loosely for any red-yellow-green map. In geographic analysis, that is unhelpful.

A choropleth shaded from cool to warm colours is still a choropleth, while a raster of temperature is a measured or modelled surface. A geographic heat map specifically refers to spatial intensity generated from point observations, unless the author defines another meaning clearly.

Heat map vs kernel density

Many web-mapping libraries implement visually useful heat layers whose purpose is exploratory display, whereas formal kernel-density estimation is an analytical method with explicit statistical choices about kernel function, bandwidth, units and output density.

The two can look similar, but a visually smoothed heat layer should not automatically be described as a statistically estimated density surface.

If the output supports a substantive quantitative claim, document the method rather than relying on the visual label “heat map”.

Where heat maps can mislead

A heat map can make sparse observations look like continuous evidence between sampled locations. A broad radius can create apparent hotspots that depend more on smoothing than on local geography, while a narrow radius can make random point noise appear meaningful.

Point collection bias also remains. If observations are more likely to be recorded near roads, clinics or populated areas, the heat map can visualise the recording process as much as the underlying phenomenon.

The map cannot distinguish those causes on its own.

When should you use one?

Use a heat map when:

the source consists of many point observations;
concentration or clustering is the main question;
exact point identity is secondary;
smoothing parameters can be explained or treated as exploratory;
the audience will not mistake the surface for direct continuous measurement.

If you need explicit aggregation units and reproducible counts per cell, a hexbin map may be clearer. If exact locations matter, keep the points. If the values belong to administrative areas, use an area-based representation instead.

Scale changes the visible hotspot pattern

A heat map is highly scale-dependent. A bandwidth that reveals meaningful neighbourhood clusters at city scale can merge them into one broad hotspot at metropolitan scale. The same point dataset can therefore support different-looking heat surfaces depending on the analytical and display scale.

This is not necessarily a defect; it means the map should use a radius that corresponds to the geographic process being discussed. A 500-metre radius might be defensible for walking-access events, whereas a 20-kilometre radius could answer a regional concentration question. The parameter should be chosen from the phenomenon, not merely from what produces the most dramatic visual.

Boundaries and population still matter even when they are not drawn

A heat map can appear to avoid administrative-boundary problems because it does not colour polygons. But the underlying opportunity structure remains. Dense population centres usually produce more observations than sparsely populated places, and reporting systems often have uneven coverage.

If the question concerns risk rather than event concentration, raw point intensity may need to be compared with population, exposure or another denominator. The heat surface alone cannot make that adjustment.