Geospatial grids and heatmaps

When you have a lot of points, aggregating them into cells or density often tells the story better than plotting every point.

Geospatial grids

The Geospatial grids layer aggregates your data into grid cells. It supports the common spatial grid systems:

  • H3 — Uber's hexagonal hierarchical grid.
  • S2 — Google's spherical cell grid.
  • A5 — a pentagonal equal-area grid.
  • Geohash — the classic string-based grid.
  • Quadkey — the tile-based grid used by web maps.

The grid layer has three modes:

  • Cells from data - each row already carries a cell-ID column; Icon Map draws those cells.
  • Aggregate points - Icon Map bins your points into cells and reduces them with count, sum, average, min or max.
  • Interpolate points (IDW) - Icon Map estimates a continuous surface from scattered point measurements (see below).

Each cell can be styled with fill, outline, and 3D extrusion — for example, coloring or raising cells by the number of points or the sum of a value inside them.

Grids are ideal for showing patterns across an area while keeping totals exact and the map fast, even with millions of points.

Interpolate points (IDW)

Where aggregation only fills the cells that contain points, interpolation estimates a value for cells between your measurements - turning scattered readings (air quality, prices, soil samples, rainfall) into a continuous surface. Icon Map uses inverse distance weighting (IDW): each cell's value is a weighted average of nearby points, with closer points counting for more.

Choose the Interpolate points (IDW) mode on a grid layer, pick the value field, and tune:

  • Power (p) - how quickly a point's influence falls off with distance. Higher values make the surface more local and "peaky". Default 2.
  • Neighbours (k) - how many nearest points contribute to each cell. Default 12.
  • Maximum distance - a cutoff beyond which points are ignored. Leave it on Auto to let Icon Map choose a sensible cutoff from the data, or set your own.
  • Clip to extent - keep the surface within the area your points actually cover.

Cells beyond the cutoff are left empty on purpose - "honest holes". Rather than inventing values far from any measurement, Icon Map simply leaves those cells blank, so the surface never implies confidence it doesn't have.

To keep the map responsive, the grid auto-coarsens above 20,000 cells, dropping to a coarser resolution and showing a warning rather than silently truncating the surface. The result renders and legends exactly like an aggregated grid, with a gradient over the value range.

Heatmap

The Heatmap layer shows the density of your points as a smooth heat surface. You can control the radius, intensity, and color ramp. For very large datasets, a GPU-accelerated rendering path keeps it responsive.

Use a heatmap when you want to show where activity concentrates rather than read individual values.

Next steps