Spatial Compute

Measure relationships between geographic features: proximity, counts, coverage, density, gaps, clusters, and comparisons.

Documentation navigation

How Compute works

Compute takes a feature, scopes it to a Region or location, optionally relates it to another feature, and returns a measurement or a set of geographic results. Use it for questions such as how many restaurants are near parks, which sites are uncovered, or how two Regions compare.

Surface	Entry point
JavaScript SDK	`e2.compute.feature(...)` or `e2.compute.run(...)`
CLI	`e2 compute <primitive> <features...>`
MCP	`e2_compute` and `e2_compare`
Hosted API	Use managed Compute through an authenticated application integration.

Choose an interface

compute.ts

const count = await e2.compute.feature("restaurant").region("Manhattan").near("park", 500).count();

JavaScript SDK

compute.ts

import { EDK } from "@edk/sdk";

const e2 = new EDK();
const count = await e2.compute.feature("restaurant")
  .region("Manhattan")
  .near("park", 500)
  .count();

Build a query with region(), near(), within(), and limit(), then finish with a primitive.

CLI

terminal

e2 compute count restaurant --area Manhattan --near park \
  --distance 500 --mode auto --json

MCP and API

e2_compute.json

{
  "primitive": "COUNT",
  "feature": ["restaurant"],
  "area": "Manhattan",
  "near": "park",
  "distanceMeters": 500,
  "mode": "auto"
}

MCP clients submit this schema to e2_compute. Use e2_compare when comparing multiple features or Regions. Hosted Compute follows the same geographic inputs through an authenticated integration.

Primitives

Primitive	Returns	Example
`nearest()`	Rows sorted by their nearest related row.	`e2.compute.feature("restaurant").area("Manhattan").near("park", 500).nearest()`
`distance()`	Source IDs, nearest IDs, and distances in meters.	`e2.compute.feature("restaurant").area("Manhattan").near("park", 500).distance()`
`within()`	Rows inside the requested relation distance.	`e2.compute.feature("restaurant").area("Manhattan").within("park", 500).within()`
`count()`	The number of matching rows.	`e2.compute.feature("restaurant").area("Manhattan").near("park", 500).count()`
`density()`	Rows per square kilometre of the relation radius.	`e2.compute.feature("restaurant").area("Manhattan").near("park", 500).density()`
`coverage()`	The fraction of target rows served by source rows.	`e2.compute.feature("restaurant").area("Manhattan").within("park", 500).coverage()`
`gaps()`	Target rows without coverage.	`e2.compute.feature("restaurant").area("Manhattan").within("park", 500).gaps()`
`cluster()`	Radius-based clusters with counts and IDs.	`e2.compute.feature("restaurant").area("Manhattan").limit(100).cluster()`

SDK examples

compute.ts

import { EDK } from "@edk/sdk";

const e2 = new EDK();

// NEAREST — rows sorted by their nearest related row
const nearest = await e2.compute.feature("restaurant").area("Manhattan").near("park", 500).nearest();

// DISTANCE — source IDs, nearest IDs, and distances in meters
const distances = await e2.compute.feature("restaurant").area("Manhattan").near("park", 500).distance();

// WITHIN — rows inside the requested relation distance
const inside = await e2.compute.feature("restaurant").area("Manhattan").within("park", 500).within();

// COUNT — the number of matching rows
const count = await e2.compute.feature("restaurant").area("Manhattan").near("park", 500).count();

// DENSITY — rows per square kilometre of the relation radius
const density = await e2.compute.feature("restaurant").area("Manhattan").near("park", 500).density();

// COVERAGE — the fraction of target rows served by source rows
const coverage = await e2.compute.feature("restaurant").area("Manhattan").within("park", 500).coverage();

// GAPS — target rows without coverage
const gaps = await e2.compute.feature("restaurant").area("Manhattan").within("park", 500).gaps();

// CLUSTER — radius-based clusters with counts and IDs
const clusters = await e2.compute.feature("restaurant").area("Manhattan").limit(100).cluster();

CLI examples

terminal

earth compute nearest restaurant --area Manhattan --near park --distance 500 --mode auto --json
earth compute distance restaurant --area Manhattan --near park --distance 500 --mode auto --json
earth compute within restaurant --area Manhattan --within park --distance 500 --mode auto --json
earth compute count restaurant --area Manhattan --near park --distance 500 --mode auto --json
earth compute density restaurant --area Manhattan --near park --distance 500 --mode auto --json
earth compute coverage restaurant --area Manhattan --within park --distance 500 --mode auto --json
earth compute gaps restaurant --area Manhattan --within park --distance 500 --mode auto --json
earth compute cluster restaurant --area Manhattan --limit 100 --mode auto --json

MCP examples

e2_compute.json

{ "primitive": "NEAREST", "feature": ["restaurant"], "area": "Manhattan", "near": "park", "distanceMeters": 500 }
// DISTANCE, COUNT, DENSITY use the same shape with their own primitive:
// { "primitive": "DISTANCE", "feature": ["restaurant"], "area": "Manhattan", "near": "park", "distanceMeters": 500 }
// { "primitive": "WITHIN", "feature": ["restaurant"], "area": "Manhattan", "within": "park", "distanceMeters": 500 }
// { "primitive": "COUNT", "feature": ["restaurant"], "area": "Manhattan", "near": "park", "distanceMeters": 500 }
// { "primitive": "DENSITY", "feature": ["restaurant"], "area": "Manhattan", "near": "park", "distanceMeters": 500 }
// { "primitive": "COVERAGE", "feature": ["restaurant"], "area": "Manhattan", "within": "park", "distanceMeters": 500 }
// { "primitive": "GAPS", "feature": ["restaurant"], "area": "Manhattan", "within": "park", "distanceMeters": 500 }
// { "primitive": "CLUSTER", "feature": ["restaurant"], "area": "Manhattan", "limit": 100 }

All documentation