Element 1.7
Analysis tools, catalogues and earth-observation AI
- Sheet
- 1.7
- Chapter
- 1 Place and planning
- State
- Drafted
- Release
- 16 November 2026
- Editor
- Contributed by Praecise as open research
- Reviewers
- To be named at review
- Safety
- S0, as the chapter
- Licence
- CC BY-SA 4.0
- Resources
- 11
The element
- Requirement
The group analyses imagery and map data with tools whose terms it understands, using open standards so that its data and methods outlive any one platform, and treats AI model outputs as estimates to be checked on the ground.
- How to tell
- The group can state, for each platform it uses, who may use it free and for what. Google Earth Engine, for example, is free to nonprofits, academic users, news media, trainees and individuals for noncommercial work, but free to government agencies only from Least Developed Countries, nationally recognised Indigenous governments, or for scholarly research; operational use is paid (Google Earth Engine: noncommercial use).
- Data is held in open formats and described so others can find it: STAC Items, Catalogs and Collections for imagery (SpatioTemporal Asset Catalog (STAC) specification), and OGC API - Features for vector data, whose Part 1 is also ISO 19168-1:2020 (OGC API - Features).
- Every AI-derived layer carries the model name, version and a check against field points.
- Options
- Desktop analysis in free software, offline, on the group's own machine (QGIS).
- Cloud processing through an open API. openEO, now an OGC Community Standard, connects to back-ends such as the Copernicus Data Space Ecosystem with clients in Python, R, JavaScript and Julia and a QGIS plugin (openEO, Copernicus Data Space Ecosystem). Moving between providers is easier because the API is shared.
- A proprietary cloud platform under its free tier, accepting its terms (Google Earth Engine: noncommercial use).
- Foundation models for classification and change detection with few labels. Open models include Prithvi-EO-2.0, trained on 4.2 million Harmonized Landsat and Sentinel-2 samples (Prithvi-EO-2.0: A Versatile Multi-Temporal Foundation Model for Earth Observation Applications); Clay, under Apache for code and weights (Clay Foundation Model); and TerraMind from IBM, ESA and Forschungszentrum Jülich, which can generate one data type from another (TerraMind 1.0). SatMAE (2022) showed the masked-autoencoder approach on satellite time series (SatMAE: pre-training transformers for temporal and multi-spectral satellite imagery). Ready-made yearly embeddings are a lighter route (AlphaEarth Foundations helps map our planet in unprecedented detail).
- Settled, open, contested
Settled: open standards and open-source tools keep data portable (SpatioTemporal Asset Catalog (STAC) specification, openEO, QGIS). Open: how well foundation models hold on small rural sites; their papers report benchmark gains (SatMAE: pre-training transformers for temporal and multi-spectral satellite imagery, Prithvi-EO-2.0: A Versatile Multi-Temporal Foundation Model for Earth Observation Applications), not community-scale field tests. Maturity of the models is Prototype by this library's reading. Contested: free proprietary platforms against open ones. The first give scale at no cost for qualifying users; the second avoid terms that can change and that exclude some government and commercial uses (Google Earth Engine: noncommercial use).
The four stages
Learn: coordinate systems and layers (QGIS), and the terms of any platform (Google Earth Engine: noncommercial use). Design: choose desktop, open API or platform; describe data with open standards (SpatioTemporal Asset Catalog (STAC) specification, OGC API - Features). Build: set up workflows (openEO, Copernicus Data Space Ecosystem) and, where labels are few, fine-tune a model (Prithvi-EO-2.0: A Versatile Multi-Temporal Foundation Model for Earth Observation Applications, Clay Foundation Model, TerraMind 1.0). Deploy: rerun yearly, check against field points, record versions.
Resources at each stage
1 Learn
Understand
- AlphaEarth Foundations helps map our planet in unprecedented detail Case study, S0
- Clay Foundation Model Repository, S0
- Prithvi-EO-2.0: A Versatile Multi-Temporal Foundation Model for Earth Observation Applications Paper, S0
- SatMAE: pre-training transformers for temporal and multi-spectral satellite imagery Paper, S0
- TerraMind 1.0 Repository, S0
2 Design
Size and choose
- Clay Foundation Model Repository, S0
- Copernicus Data Space Ecosystem Dataset, S0
- Google Earth Engine: noncommercial use Tool, S0
- OGC API - Features Standard, S0
- openEO Standard, S0
- QGIS Tool, S0
- SpatioTemporal Asset Catalog (STAC) specification Standard, S0
- TerraMind 1.0 Repository, S0
4 Deploy
Install, run and measure
- Copernicus Data Space Ecosystem Dataset, S0
- Google Earth Engine: noncommercial use Tool, S0
- openEO Standard, S0
What changes between 30 adults and 300 people
At 30 adults, desktop software and a free archive are enough, and AI models are optional. At 300 with farming, forestry or hazard duties, repeatable cloud workflows and model-assisted maps save time, and the choice of platform becomes a governance question because its terms bind the group's operations.
Reference designs
Open for contributions Designs with drawings, specifications and a bill of materials. Add to it
Parts list
Every resource that supports this element. Those marked External carry the label “External resource. Not made or verified by Public Protocol”.
| Resource | Type | Origin | Maturity | Safety | Licence, as stated at its source |
|---|---|---|---|---|---|
| AlphaEarth Foundations helps map our planet in unprecedented detail | Case study | External | Not assessed | S0 | Not stated on the page opened |
| Clay Foundation Model | Repository | External | Prototype | S0 | Apache (code and model weights); documentation CC-BY-4.0 |
| Copernicus Data Space Ecosystem | Dataset | External | Established | S0 | Not stated on the page opened |
| Google Earth Engine: noncommercial use | Tool | External | Established | S0 | Not stated on the page opened |
| OGC API - Features | Standard | External | Established | S0 | Part 1 available free on the OGC website; other terms not stated on the page opened |
| openEO | Standard | External | Established | S0 | Apache License, Version 2.0 (documentation) |
| Prithvi-EO-2.0: A Versatile Multi-Temporal Foundation Model for Earth Observation Applications | Paper | External | Not assessed | S0 | CC BY 4.0 |
| QGIS | Tool | External | Established | S0 | GNU GPLv2+ |
| SatMAE: pre-training transformers for temporal and multi-spectral satellite imagery | Paper | External | Prototype | S0 | CC BY 4.0 |
| SpatioTemporal Asset Catalog (STAC) specification | Standard | External | Established | S0 | Creative Commons Attribution 4.0 International (site footer) |
| TerraMind 1.0 | Repository | External | Prototype | S0 | Apache-2.0 |
The standing questions
Who can afford this: software is free; computing for models is not. Who does the work: one or two members with programming skill. Who decides: the group, on which platform's terms it accepts.
Environments
No source found that ties these tools to particular environments. Model training data covers the world unevenly; no source opened in this work quantifies this by setting.
Safety
State and evidence
Drafted
An editor has written the element's text from the resources listed under it, citing them. It is checked when it is reviewed.
Evidence from a real place or system is recorded here, with what did not work. What the states mean
History
- Drafted in the release of 16 November 2026.