Skip to content

IBM Geospatial Studio Workshop

Geospatial Studio Banner

Welcome to the IBM Geospatial Studio Workshop! This hands-on workshop will guide you through deploying and using the Geospatial Exploration and Orchestration Studio - an integrated platform for fine-tuning, inference, and orchestration of geospatial AI models.

🎯 Workshop Overview

This workshop is designed for beginners who have never heard of Geospatial Studio before. By the end of this workshop, you will be able to:

  • ✅ Deploy Geospatial Studio in your environment
  • ✅ Navigate the Studio UI and understand its components
  • ✅ Use the Python SDK to interact with the platform
  • ✅ Run inference with fine-tuned geospatial AI models
  • ✅ Onboard and prepare datasets for model training
  • ✅ Fine-tune models for specific geospatial tasks
  • ✅ Execute end-to-end workflows for real-world applications

⏱️ Workshop Duration

Total Time: Approximately 3-4 hours

  • Pre-work: 1-1.5 hours (deployment and setup)
  • Introduction: 15 minutes (reading)
  • Lab 1 - Getting Started with IBM Geospatial Studio: 10 minutes (Beginner)
  • Lab 2 - Onboarding Pre-computed Examples: 20 minutes (Beginner)
  • Lab 3 - Upload Model Checkpoints and Run Inference: 30 minutes (Intermediate)
  • Lab 4 - Training a Custom Model for Wildfire Burn Scar Detection: 60-90 minutes (Intermediate, includes model training)

🎓 Target Audience

This workshop is ideal for:

  • Data scientists interested in geospatial AI
  • Researchers working with satellite imagery
  • Developers building geospatial applications
  • Anyone curious about applying AI to Earth observation data

Prerequisites: Basic knowledge of Python and familiarity with Jupyter notebooks is helpful but not required.

📦 Getting Workshop Materials

Before starting the labs, clone the workshop repository to access all notebooks and configuration files:

git clone https://github.com/terrastackai/geospatial-studio.git
cd geospatial-studio/workshop/docs/notebooks

This repository includes:

  • ✅ All lab notebooks (Lab 1-4)
  • ✅ JSON configuration files (model templates, datasets, backbones)
  • ✅ Sample data references
  • ✅ Complete workshop documentation

Alternative: You can download individual notebooks from each lab page, but you'll need to manually download associated JSON files for Labs 3 and 4.

🏗️ What is Geospatial Studio?

The Geospatial Exploration and Orchestration Studio is an integrated platform that combines:

  • No-code UI for visual interaction
  • Low-code SDK for programmatic access
  • RESTful APIs for integration

It supports the complete machine learning lifecycle for geospatial data:

  1. Dataset Management - Onboard, validate, and prepare training data
  2. Model Fine-tuning - Customize foundation models for specific tasks
  3. Inference at Scale - Run models on large geospatial datasets
  4. Visualization - View and analyze results interactively

Built on top of:

📋 Workshop Structure

Pre-work (Required)

Complete before the workshop begins. You'll deploy Geospatial Studio in your environment and verify the installation.

Start Pre-work →

Introduction

Learn about Geospatial Studio's architecture, key concepts, and capabilities.

Lab 1: Getting Started with IBM Geospatial Studio

Explore the three ways to interact with Studio: UI, API, and SDK. Navigate the interface, generate API keys, install the Python SDK, and make your first API calls. Perfect for fresh deployments!

Time: 10 minutes | Difficulty: Beginner

Lab 2: Onboarding Pre-computed Examples

Learn how to onboard pre-computed inference examples and geospatial layers. Configure styling for raster and vector data visualization.

Time: 20 minutes | Difficulty: Beginner

Lab 3: Upload Model Checkpoints and Run Inference

Upload fine-tuned model checkpoints and run inference on new geographic areas. Learn to define spatial/temporal domains and visualize results.

Time: 30 minutes | Difficulty: Intermediate

Lab 4: Training a Custom Model for Wildfire Burn Scar Detection

Complete an end-to-end workflow: onboard a training dataset, fine-tune a foundation model for wildfire burn scar detection, and run inference on real wildfire events.

Time: 60-90 minutes (includes model training) | Difficulty: Intermediate

Note: This lab requires GPU access for model training. If GPUs are not available, you can use an existing fine-tuned model or train outside Studio using TerraTorch.

🚀 Getting Started

  1. Complete the Pre-work - Deploy Geospatial Studio in your environment
  2. Follow the Labs - Work through each lab sequentially
  3. Experiment - Try the exercises and explore on your own
  4. Ask Questions - Use the troubleshooting guide and FAQ

📚 Additional Resources

🤝 Contributing

Found an issue or have suggestions? Please open an issue on our GitHub repository.

📄 License

This workshop is licensed under the Apache License 2.0.

Ready to begin? Start with the Pre-work →