IrelandPublic sector2026 – ongoingCentral Statistics Office
Random-forest peatland mapping for ecosystem accounts
Ireland's CSO reported 0.887 precision for its peatland map updated using random forests
Problem
Ireland's existing land-cover map was updated too infrequently for ecosystem reporting. European mapping products lacked a class corresponding to Irish peatlands.
Approach
The CSO classified Sentinel-2 imagery using a random-forest model trained mainly on existing national mapping. It cleaned the predictions and checked the resulting map against satellite imagery manually labelled by staff. The map fed into ecosystem extent accounts.
Results
Takeaway
Published accuracy described the final processed map. Validation used satellite imagery rather than field visits.
Categories
- Industry
- Public sector
- Business function
- Internal work and knowledge management
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
The CSO explained its random-forest mapping pipeline and human validation. It reported precision, recall and F1 results for the processed map.
Full details are in the original sources.
