Earth Engine MCP Server

A Model Context Protocol (MCP) server for Google Earth Engine that enables satellite imagery analysis through Claude Desktop or any MCP-compatible client.

🚀 Quick Start

Prerequisites

Google Earth Engine Account: Sign up at earthengine.google.com
Service Account Key: Create a service account in Google Cloud Console with Earth Engine API access
Node.js: Version 18 or higher
Google Cloud Storage Permissions (for exports): See GCS Permissions Guide

Installation

# Clone the repository
git clone https://github.com/Dhenenjay/earth-engine-mcp.git
cd earth-engine-mcp

# Install dependencies
npm install
# or
pnpm install

Configuration

Download your Earth Engine service account JSON key from Google Cloud Console
Save it to a secure location (e.g., C:\Users\YourName\ee-key.json)

🔧 Usage

Option 1: Claude Desktop (Recommended)

Edit Claude Desktop configuration:Open %APPDATA%\Claude\claude_desktop_config.json and add:
{ "mcpServers": { "earth-engine": { "command": "node", "args": ["C:\\path\\to\\earth-engine-mcp\\mcp-earth-engine.js"], "env": { "EARTH_ENGINE_PRIVATE_KEY": "C:\\path\\to\\your\\ee-key.json" } } } }
Restart Claude Desktop
Use Earth Engine tools in Claude:
- "Search for Sentinel-2 datasets"
- "Calculate NDVI for San Francisco"
- "Get Landsat imagery for January 2024"

Option 2: Direct MCP Client

// Connect to the MCP server
const server = spawn('node', ['mcp-earth-engine.js'], {
  env: {
    EARTH_ENGINE_PRIVATE_KEY: '/path/to/ee-key.json'
  }
});

Option 3: SSE Endpoint (For Web Clients)

# Start the SSE server
EARTH_ENGINE_PRIVATE_KEY=/path/to/ee-key.json npm run dev

# Connect at: http://localhost:3000/api/mcp/sse

📚 Available Tools

1. search_catalog

Search the Earth Engine data catalog for datasets.

{
  "query": "sentinel-2"
}

2. get_band_names

Get available bands for a dataset.

{
  "datasetId": "COPERNICUS/S2_SR_HARMONIZED"
}

3. filter_collection

Filter satellite imagery by date and location.

{
  "datasetId": "COPERNICUS/S2_SR_HARMONIZED",
  "startDate": "2024-01-01",
  "endDate": "2024-01-31",
  "region": {
    "type": "Point",
    "coordinates": [-122.4194, 37.7749]
  }
}

4. calculate_ndvi

Calculate Normalized Difference Vegetation Index.

{
  "imageId": "COPERNICUS/S2_SR_HARMONIZED",
  "redBand": "B4",
  "nirBand": "B8"
}

5. get_map_url

Generate map visualization URLs.

{
  "imageId": "COPERNICUS/S2_SR_HARMONIZED",
  "visParams": {
    "bands": ["B4", "B3", "B2"],
    "min": 0,
    "max": 3000
  }
}

6. calculate_statistics

Calculate image statistics for a region.

{
  "imageId": "COPERNICUS/S2_SR_HARMONIZED",
  "region": {
    "type": "Polygon",
    "coordinates": [[[-122.5, 37.7], [-122.4, 37.7], [-122.4, 37.8], [-122.5, 37.8], [-122.5, 37.7]]]
  },
  "scale": 30
}

7. export_image_to_cloud_storage 🆕

Export satellite imagery to Google Cloud Storage (works with service accounts).

{
  "collection": "COPERNICUS/S2_SR_HARMONIZED",
  "start_date": "2024-01-01",
  "end_date": "2024-01-31",
  "region": "Los Angeles",
  "scale": 10,
  "bucket": "earth-engine-exports-PROJECT_ID"
}

Note: Service accounts cannot export to Google Drive. Use GCS exports instead. See GCS Permissions Guide for setup instructions.

🧪 Testing

# Run the test suite
node test-earth-engine.js

# Test individual tools
node mcp-earth-engine.js
# Then send JSON-RPC messages via stdin

🌍 Example Workflows

NDVI Analysis for Agriculture

Search for Sentinel-2 imagery
Filter by your farm's location and growing season
Calculate NDVI to assess crop health
Get statistics to track changes over time

Deforestation Monitoring

Filter Landsat imagery for two time periods
Calculate vegetation indices for both
Compare statistics to detect forest loss

Urban Heat Island Analysis

Search for temperature datasets (MODIS)
Filter for summer months in urban areas
Calculate statistics to identify heat patterns

📊 Supported Datasets

Sentinel-2: High-resolution optical imagery (10m)
Landsat 8/9: Multispectral imagery (30m)
MODIS: Daily global coverage
CHIRPS: Precipitation data
SRTM: Digital elevation models
And many more in the Earth Engine catalog

🔒 Security

Never commit your service account key to version control
Store keys in secure locations with restricted access
Use environment variables for production deployments
Rotate service account keys regularly

🤝 Contributing

Contributions are welcome! Please:

Fork the repository
Create a feature branch
Add tests for new features
Submit a pull request

📝 License

MIT License - See LICENSE file for details

🆘 Troubleshooting

"Earth Engine not initialized"

Check your service account key path
Verify the key has Earth Engine API access
Ensure the project has Earth Engine API enabled

"Dataset not found"

Use the exact dataset ID from Earth Engine catalog
Check if the dataset requires special access

Connection issues

Verify Node.js version (18+)
Check firewall settings
Ensure all dependencies are installed

📧 Support

GitHub Issues: github.com/Dhenenjay/earth-engine-mcp/issues
Earth Engine Documentation: developers.google.com/earth-engine

🎉 Quick Test

After setup, try this in Claude Desktop:

"Search for Sentinel-2 satellite imagery and calculate NDVI for San Francisco in January 2024"

The server will:

Search the Earth Engine catalog
Filter Sentinel-2 data for the specified time and location
Calculate NDVI vegetation index
Return statistics and visualization URLs

Happy Earth Observing! 🛰️

Earth Engine MCP Server