GIS AI for forestry and carbon projects

FloraScope automates the GIS desk work of a carbon project. It reads AI-classified satellite land cover and two decades of forest-change records for your boundary, then checks them against Gold Standard rules.

What it means here

Three layers, and only one of them is a model

“AI GIS” covers a lot of ground. This is what it means in FloraScope, and where each part comes from.

AI-derived data

Land cover classified by a deep-learning model

Dynamic World, from Google and WRI, labels every 10 m Sentinel-2 pixel as trees, crops, water, built area and more. FloraScope reads it for your boundary, year by year.

Automated analysis

Geoprocessing without the clicks

Upload a boundary and Earth Engine computes land cover, tree-cover loss and vegetation trend for that exact shape. No layers to download, clip or reproject.

Rules, not guesses

Eligibility you can trace

Each Gold Standard pathway is checked criterion by criterion against published rules. A flag comes with its reason and its dataset, so it can be argued with.

Division of labour

What the software does, and what stays with people

Screening a site means running the same analysis on every candidate boundary. That part is automated. The parts that decide a project are not.

Automated for your boundary

Minutes per site, on public satellite records.

  • Land cover composition, and how it changed over the years
  • Tree-cover loss by year since 2001
  • Vegetation trend from Sentinel-2 and Landsat
  • Whether the land was forest ten years before the project start
  • Canopy cover, tree structure and biomass
  • Farm plots checked against the project boundary for overlaps and gaps
  • A PDF report that names the source of every figure

Still a person's job

Satellite records cannot settle these. Tenure and additionality appear as open items, not as scores.

  • Field verification of what the satellite shows
  • Land tenure, permits and protected-area status
  • Stakeholder consultation
  • Additionality and regulatory surplus
  • The final call on methodology and project design

Why not a black box

Where we chose rules over models

A carbon project is audited. Whoever validates it will ask where a number came from, and “the model said so” is not an answer. So FloraScope uses machine learning where it is already published and peer reviewed, in the satellite land-cover data, and keeps the decisions in plain rules.

Eligibility is a checklist against Gold Standard requirements. Each criterion is marked as passed, failed or not tested, with the reason. When the satellite data is too coarse for a small plot, the result says so and does not report a number.

Tree-crown detection uses a vegetation index and a threshold you can see and move. It is less clever than a neural network and far easier to defend, because you can show exactly why a pixel counted as canopy.

Every result names its dataset, its resolution and its year. If two land-cover sources disagree about a site, the report shows the disagreement.

Fit

General GIS AI tools, and this one

General-purpose GIS AI covers two things: toolboxes for training and running your own models on imagery, and assistants that build maps from a typed request. Both are for people whose job is GIS, and both start from a blank map.

FloraScope starts from a question: is this land worth developing as a forestry carbon project? It needs a boundary file and nothing else. If you need a map of something else, a general tool is the better choice. If you need to convert or measure a boundary first, the free GIS tools do that without an account.

If you are weighing software against hiring someone, see what a GIS consultant does on a carbon project.

FAQ

GIS AI, in detail

See what the satellites already know

Upload a boundary. The GIS tools are free, and screening is switched on per project.