Chat2Geo provides a ChatGPT-like web application for performing remote-sensing-based geospatial analysis. It targets users needing to analyze satellite imagery and integrate custom data, offering a user-friendly interface powered by Google Earth Engine and AI for tasks like land cover mapping and change detection.

How It Works

The application utilizes a Next.js frontend with a backend powered by Google Cloud Platform services. It integrates Google Earth Engine for real-time access to satellite data and large-scale geospatial computations. Retrieval-Augmented Generation (RAG) is employed to combine geospatial insights with user-provided textual data, enhancing analysis capabilities. Supabase handles database and authentication, while Vertex AI can host custom AI models for specific tasks.

Quick Start & Requirements

• Install dependencies: npm install
• Requires a Google Earth Engine account and project.
• Environment variables must be configured for GCP (API keys, service account), LLM API (e.g., OpenAI), and Supabase.
• Optional keys for Esri integration and Mailgun for feedback submission.
• Run development server: npm run dev
• Demo: https://chat2geo.georetina.ai

Highlighted Details

• Chat-style interface with natural language interaction for geospatial functions.
• Integrates Google Earth Engine for real-time satellite imagery analysis.
• Supports uploading and integrating user's own vector data.
• Includes analyses for Urban Heat Island (UHI), Land Cover mapping/change detection, and preliminary air pollution analysis.
• Features RAG for knowledge base integration with custom documents.

Maintenance & Community

• Project is actively maintained and updated in parallel with the GRAI 2.0 enterprise version.
• Contributions are welcomed; contact via email (shahabj.github@gmail.com) or GitHub issues.
• Contributing Guidelines are available.

Licensing & Compatibility

• Open-source version is free to use under its license.
• Google Earth Engine has restrictions on commercial usage.
• Compatibility for commercial use with GEE data is limited.

Limitations & Caveats

The application is noted as not yet production-ready, with known and potential unknown bugs. Some functionalities are not fully implemented, and setup instructions may have omissions. Geospatial analyses are GEE-based and may use data that is not up-to-date, requiring careful interpretation of results. Certain components may require refactoring or redesign.