Google Earth Engine

Render 40 years of satellite imagery, night lights, land cover, climate, and terrain, and compute area statistics with Google Earth Engine.

Tool Reference: 3 tools with full schemas →

Use Google Earth Engine (GEE) to look back in time and across data types: the full Landsat archive back to 1982, Sentinel-2 composites, VIIRS night lights, land cover, rainfall, temperature, terrain, and population, rendered over any area of interest. Beyond imagery, GEE computes numeric statistics for an area, such as monthly rainfall totals, NDVI time series, or how many people live inside a boundary.

GEE complements the EO toolkit rather than replacing it: EO is the place for the freshest Sentinel scenes, while GEE is the place for long archives, composites, night lights, and climate or land-cover context. Like EO, it pairs naturally with the Geo toolkit: use Geo to build a GeoJSON boundary, then pass it to GEE. For whole countries you can skip the polygon entirely and pass a country name.

Prefer video? This 48-second walkthrough shows the whole setup, and its length is the point: connect the toolkit and start asking questions:

Example prompts

Show me how the night lights of Seoul changed between 2013 and 2023, side by side.

Render this valley from Landsat in 1985 and today so I can compare how it developed.

How much rain fell on this watershed each month last year, and how does it compare to the land cover mix?

Setup

GEE needs no account and no credentials: no Google sign-in, no cloud project, no Earth Engine registration. Requests run on Toolforest’s own Earth Engine project.

Open Toolforest and select the MCP server where you want to use GEE.
Find Google Earth Engine in the Planet Earth tab and add it to the MCP server.

That is the whole setup.

ℹ Shared capacity

Because requests run on a shared Earth Engine project, heavy use can hit a rate limit. If a call reports that Earth Engine is rate limiting or out of quota, wait a moment and retry.

Tools

Tool	What it does
list_datasets	List the curated Earth Engine datasets available to render, with native resolution, valid time range, and visualization presets. Call it before choosing a dataset instead of guessing names.
render_image	Render a dataset over an area to a shareable PNG: Landsat and Sentinel composites, night lights, land cover, climate, terrain. Compare mode stitches 2-4 labeled time windows into one image for before/after views.
get_statistics	Compute numeric series and totals for an area: NDVI/NDWI over time, monthly rainfall, temperature means, population counts, and land-cover class fractions.

Notes and limits

GEE exposes a curated dataset registry, not all of Earth Engine. Start with list_datasets to see what is available.
Earth Engine ingestion lags the origin archives by roughly a day. For the freshest Sentinel imagery, use the EO toolkit.
Use a GeoJSON area of interest, a WGS84 bounding box, or a country name. Provide one, not several.
Large areas are sampled at a coarser effective resolution for statistics (reported in the result); population sums always count at native resolution.
Image URLs are temporary. Download or save an image elsewhere if you need it after it expires.

Cookbook

GEE cookbook examples are coming soon. Good first candidates are a night-lights change comparison, a 40-year Landsat before/after of a growing city, and a rainfall-plus-land-cover profile of a watershed.