Live online · 20–22 October 2026
From Satellite Data to Forest Carbon Insights
A Practical GeoAI Bootcamp
Build biomass maps. Assess their reliability. Estimate biomass carbon—and explain what your results can support.
With Courage Kamusoko · English · Intermediate level
Go beyond a finished map
Can you explain the numbers behind your forest biomass map?
A compelling map is only part of the job. Your validation design, sensor choices and carbon-conversion assumptions shape the conclusions you can draw.
This bootcamp connects optical, radar and LiDAR principles with a practical machine learning workflow. Learn to interpret your outputs and prepare clear evidence for research and consulting discussions.
01 / BUILD
Create forest biomass maps
Combine GEDI L4A, Sentinel-2 and ALOS PALSAR-2 in a reusable Random Forest workflow for aboveground biomass density.
02 / EVALUATE
Assess and explain reliability
Interpret spatial validation, model performance, diagnostic uncertainty and explainable ML. See how modelling choices change your conclusions.
03 / INTERPRET
Turn biomass into carbon insights
Estimate biomass-carbon stocks and screen mapped differences between years, with documented assumptions and validation needs.
Included in your registration
A workflow you can return to.
Feedback you can act on.
Technical guidebook + five reusable labs
Receive the Forest Carbon Monitoring Starter Kit, companion notebooks, worked case-study files and a capstone brief.
Instructor-led comparison exercise
Explore how validation design and predictor selection affect reported accuracy and biomass interpretation.
Feedback on your capstone
Get feedback on reproducibility, evaluation, carbon conversions, presentation and the interpretation of your results.
Personal 30-minute consultation
Discuss your study area, troubleshoot a workflow issue or identify next steps in one scheduled consultation.
The three-day programme
Follow the full path from observations to interpretation.
Each day combines accessible technical explanations, demonstrations, lab orientation and questions. Execute and adapt the full notebooks independently, supported by worked outputs.
DAY 01 / OBSERVE
Earth observation for forest biomass
- Optical reflectance, vegetation indices and seasonal effects.
- SAR wavelength, polarization, backscatter and forest response.
- LiDAR structure, GEDI footprints and quality filtering.
- Labs 1a & 1b: prepare 2019 and 2024 training data.
DAY 02 / MODEL
Machine learning and model reliability
- Random Forest regression, tuning and evaluation.
- Random versus spatially blocked validation; R², RMSE and bias.
- Ensemble spread, input sensitivity, SHAP and partial dependence.
- Labs 2a & 2b: model annual AGBD and interpret diagnostics.
DAY 03 / INTERPRET
From biomass maps to biomass carbon
- Aboveground and belowground biomass-carbon estimates.
- Density, area totals and corresponding COâ‚‚ mass.
- Why two good biomass maps can produce a bad change map.
- Lab 3: stock tables, change screening and capstone preparation.
All sessions: 4–7 pm JST / 7–10 am GMT, with a 15-minute break each day. Other sensors—including Landsat, Sentinel-1, ICESat-2 and ESA BIOMASS—provide conceptual context; the five kit labs use GEDI L4A, Sentinel-2 and ALOS PALSAR-2.
Apply the workflow to your own question
Your capstone: an interpretable forest-carbon analysis
Use your own area of interest to prepare outputs that another researcher or consultant can review.
- AGBD map
- Model-validation summary
- Biomass-carbon stock table
- Change-screening figure
- One-page interpretation
Completed independently after the live sessions. Review covers reproducibility, sound evaluation, correct units and conversions, presentation and defensible interpretation. Submission and consultation arrangements will be shared separately.
Who this is for
Bring your GIS experience.
Build your forest-carbon workflow.
For forest geospatial consultants, researchers, graduate students and GIS or remote sensing professionals seeking a practical introduction to biomass modelling and biomass-carbon interpretation.
Before you begin
- Basic GIS and remote sensing knowledge, including rasters and coordinate reference systems.
- Familiarity with introductory statistics; basic Python notebook skills are an advantage.
- A computer, reliable internet, modern browser, Google Drive and Colab access.
- An Earth Engine-enabled Cloud project with suitable permissions and an eligible commercial or noncommercial configuration.
Complete account setup and a sample notebook check before Day 1. Read Chapters 1 and 2 of the guidebook in advance. QGIS is useful for viewing outputs.
Course fees & registration
Choose your registration category.
The same core course and learning materials are provided across all registration categories.
Available when registration and payment are completed by 14 October 2026.
Register at the early bird rate →For professionals, researchers and graduate students registering at the standard rate.
Register for the bootcamp →Included for participants with an active AI.Geolabs Premium membership.
Access member registration →Optional paid cloud computing or platform charges are separate from the course fee.
Practical questions
Know what to expect.
Will I run every notebook during the live sessions?
The live sessions focus on concepts, demonstrations, lab orientation and questions. You execute and adapt the full notebooks independently. Prepared outputs support discussion when exports, model searches or longer calculations cannot finish during a session.
Can I use my own study area?
Yes. The capstone uses your personal area of interest. Your consultation can help identify data availability, configuration issues and next steps for adapting the workflow.
What does the personal consultation include?
One scheduled 30-minute consultation to discuss your study area, troubleshoot a workflow issue or plan the next steps for your project. Booking arrangements will be shared separately.
Do the uncertainty layers provide confidence intervals?
No. Ensemble spread and Monte Carlo input sensitivity are diagnostic indicators. They are not calibrated confidence intervals or a complete uncertainty budget. You will learn how to interpret and communicate that distinction.
Does this training produce verified carbon credits?
No. The workflow supports learning, research, method development and feasibility assessment. It estimates aboveground and belowground biomass carbon rather than all ecosystem carbon pools. Mapped change needs independent corroboration; completion does not constitute carbon-credit certification or a verified MRV evidence package.
20–22 October 2026 · Live online
Make your next biomass map more useful.
Connect satellite observations, model evaluation and biomass-carbon interpretation in one practical learning experience.
Choose your registration →Scope: Training and feasibility analysis. Aboveground and belowground biomass carbon are covered; a complete ecosystem carbon inventory is outside the programme. Diagnostic uncertainty and mapped differences must be interpreted alongside assumptions, limitations and independent validation.