H3

H3 visualization is a way to aggregate point data into a grid of H3 cells.

H3 visualizations are defined using "type": "h3" . They generally share the same properties and behaviors as color range visualizations for polygons. Properties of specific relevance to H3 are:

Field name	Description
aggregation	The aggregation method that will be used on points within each cell. Supported values are count (default), sum, min, max, and mean .
binMode	fixed (default), low, medium, high, or auto. fixed keeps the cell resolution constant at all zooms; the other modes pick a resolution from the current zoom, where low produces larger hexagons and high smaller ones (auto behaves like medium).
baseBinLevel	Required. If binMode is fixed, this is the H3 cell resolution that the map will use. H3 cells vary from resolution 0 (largest) to resolution 15 (smallest). In the zoom-adaptive modes, this is the reference resolution used for calculating class breaks — you will get best results choosing a resolution that matches the fixed resolution you would choose for the most commonly-viewed zoom of your map.
numericAttribute	The numeric column to aggregate. Required unless aggregation is count , in which case the column choice is irrelevant.

This is an example of an H3 visualization

defined with the following style

{
  "config": {
    "steps": {"type": "quantiles", "count": 5},
    "aggregation": "sum",
    "binMode": "fixed",
    "baseBinLevel": 3,
    "numericAttribute": "capacity_mw"
  },
  "paint": {"color": "@riverine"},
  "type": "h3",
  "version": "2.3.1",
  "label": {},
  "legend": {"displayName": "auto"}
}

aggregation: "count" doesn't need a numericAttribute — it counts the points in each cell:

{
  "version": "2.3.1",
  "type": "h3",
  "config": {
    "aggregation": "count",
    "baseBinLevel": 4,
    "binMode": "medium",
    "steps": {"type": "continuous"}
  },
  "paint": {"color": "@galaxy", "opacity": 0.85, "strokeColor": "auto", "strokeWidth": 1},
  "legend": {"displayName": "auto"}
}

Aggregation popups

H3 cells display aggregated values, not raw feature attributes. Their popups reference special attribute names — "felt:cluster_size" for the feature count, and "felt:sum:column", "felt:mean:column", "felt:min:column", "felt:max:column" for aggregates. If no popup is provided, Felt auto-generates sensible stats from the aggregation type. See the popup block.

{
  "version": "2.3.1",
  "type": "h3",
  "config": {
    "numericAttribute": "revenue",
    "aggregation": "sum",
    "baseBinLevel": 5,
    "binMode": "medium",
    "steps": {"type": "jenks", "count": 5}
  },
  "paint": {"color": "@copper", "opacity": 0.85, "strokeColor": "auto", "strokeWidth": 1, "isClickable": true},
  "popup": {
    "titleAttribute": "felt:sum:revenue",
    "keyAttributes": ["felt:cluster_size", "felt:mean:revenue", "felt:min:revenue", "felt:max:revenue"]
  },
  "attributes": {"revenue": {"displayName": "Revenue ($)", "format": {"thousandSeparated": true, "mantissa": 0}}},
  "legend": {"displayName": "auto"}
}

H3 cells are rendered as polygons, so size has no effect on them — control cell size with baseBinLevel (H3 resolution) and binMode (how that resolution adapts to zoom). The attributes block renames the source column, so "revenue" makes a popup entry read "Sum of Revenue ($)".