Geospatial decision-support · Google Earth Engine

Above-ground biomass & carbon, county by county.

Carbon-Eye estimates above-ground biomass and carbon stock across Kenya’s 47 counties - fusing Sentinel radar and optical data, terrain, climate and soils with a machine-learning ensemble in Google Earth Engine.

Sentinel-1 / Sentinel-2 · Random Forest · Gradient Boosting · SVM

Biomass<10 → 100+ t/ha

Mean above-ground biomass per country, ESA CCI 2022 (t/ha). Measured per country from ESA CCI and Hansen data; Kenya's 47 counties come from Carbon-Eye's own models - see the county atlas.

Drag to rotate · pinch or the +/− buttons to zoom · click a country to open it in the explorer.

The county carbon atlas

Every Kenyan county, mapped from one national analysis

376.8M t C
above-ground carbon in Kenya's vegetation
summed over all 47 counties
6.8 t C/ha
national average
per hectare of mapped land
24.5 t C/ha
densest county: Nyeri
montane forest on Mt Kenya and the Aberdares
1.4B t CO₂e
the same stock as CO₂ - not emissions
about 15× Kenya's 2015 greenhouse-gas emissions

Satellite estimates from run 52a2006ae7c0 (2026-11-18), checked against the ESA CCI reference map: Kenya's total is within 4% of it. Screening estimates, not certified carbon accounting.

Background photo: Evening over the Chyulu Hills - grassland, scattered trees and volcanic cones. Photo: Sho Hatakeyama, CC0, via Wikimedia Commons (resized)

How Carbon-Eye works

Inputs → predictor stack → models → validated maps

Five steps, all inside Google Earth Engine, from a county boundary to carbon maps, statistics and downloads.

  1. 01

    Select area & year

    Choose one or more Kenyan counties, or draw a custom boundary, and an ESA CCI AGB reference year (or a range, averaged across every snapshot in it).

  2. 02

    Build predictor stack

    Earth Engine assembles a cloud-filtered, multi-source stack of optical, radar, terrain, climate and soil layers.

  3. 03

    Sample & split

    Reference carbon points are sampled from the stack and split into training and testing sets.

  4. 04

    Train models

    Random Forest, Gradient Tree Boosting and SVM each learn the link between predictors and carbon.

  5. 05

    Map & validate

    Estimates are mapped in carbon or biomass units, checked against held-out data (RMSE, MAE, R²), and summarised by county alongside tree-cover loss.

Cached runs and on-demand diagnostics keep it responsive for non-specialist users.

See it working

One real analysis, start to finish

Kabarak University's campus in Nakuru County, drawn on the map and analysed live in under a minute. These are screenshots of the actual results, not mock-ups.

Carbon-Eye's carbon map of Kabarak University campus drawn in red to yellow over satellite imagery of the surrounding farms
The carbon map over real satellite imagery - Kabarak University campus, 94 ha: 6.7 t C/ha on average, 626 t C in all.
Precision card giving a 90% range of 6.0 to 15.6 t C/ha for the area average
How precise it is - 90% ranges measured on land the model never trained on.
Carbon pools bar splitting 6.7 t C/ha above ground and 1.9 t C/ha below ground
Above and below ground - roots add 1.9 t C/ha to the satellite estimate.
Risk screen card rating deforestation, fire and protected-area risk as low
Would the carbon stay? Deforestation, fire and protected-area checks, with their sources.
Land and soil context card showing cropland 22%, land-cover bars and topsoil organic carbon
What the land actually is - cropland, buildings and trees from three independent maps, with a warning when a boundary doesn't look like the farm it's meant to be.
Line chart of Nyeri County's mean carbon from 2010 to 2022 with a shaded 90% range
Every county's carbon over time in the public atlas (Nyeri shown), with a plain verdict on whether the change is real.

Screenshots from carbon-eye.org, 2026-11-15. Every county's figures are public in the county atlas.

Validation

Checked against data the models never saw

Every run holds out spatial blocks of land the models never trained on and reports its error against ESA CCI AGB v6 (2020). Measured on 2026-10-15 with the previous model settings. The tuned settings in use since 2026-10-16 had lower error at the same test areas (Random Forest 8.4%, boosting 4.4% on average), so these figures are, if anything, conservative.

Share of a median guess's error the ensemble removes

Held-out spatial blocks per test area; RMSE in t C/ha.

  • Kitui smallholder farm (277 ha)34%

    RMSE 7.9 vs 12.0 for the median guess

  • 0.25 ha plot, Kitui42%

    RMSE 7.5 vs 13.1 for the median guess

  • Mau farmland (~970 ha)30%

    RMSE 6.6 vs 9.4 for the median guess

  • Kakamega rainforest (~2,500 ha)74%

    RMSE 14.5 vs 55.9 for the median guess

  • Kirinyaga county71%

    RMSE 8.0 vs 28.1 for the median guess

  • Makueni county68%

    RMSE 4.7 vs 14.8 for the median guess

Error removed
30%-74%
vs. a median guess, at 6 Kenyan test areas
Ensemble RMSE
4.7-14.5 t C/ha
on held-out spatial blocks
90% ranges held
90%
of held-out points, against a 90% target

Agreement is with the ESA reference map, not field plots, so every run also reports an independent GEDI lidar check. Every site, errors by carbon density and a model benchmark are in the validation report.

Kenya on the ground

What the numbers stand for

Montane forest holds many times the carbon of savanna; a young plantation or a mangrove seedling is carbon still to come. Carbon-Eye's maps tell these landscapes apart from orbit - these are the same kinds of places, at eye level.

Moss-hung trees and thick undergrowth in montane forest
Montane forest on Mt Kenya, along the Naro Moru River trailPhoto: Daniel Case, CC BY-SA 4.0, via Wikimedia Commons (resized)
Tall forest trees with a vine-covered trunk and hills of forest behind
Inside the Mau Forest, Kenya's largest closed-canopy forest and a water tower for the Rift ValleyPhoto: Kaa.rie, CC BY-SA 4.0, via Wikimedia Commons (resized)
A single young mangrove tree on a tidal mudflat with mangrove forest behind
A young mangrove on the mudflats of Manda Island, Lamu - where mangrove restoration happensPhoto: Faizan2023, CC BY-SA 4.0, via Wikimedia Commons (resized)
Rolling green tea fields divided by rows of trees with forest on the hills
Tea estates, tree lines and forest around KerichoPhoto: Ahero dala, CC BY-SA 4.0, via Wikimedia Commons (resized)
Rows of young slender trees planted on red soil
A young tree plantation at Dzikunze, KilifiPhoto: Chris Obiero, CC BY-SA 4.0, via Wikimedia Commons (resized)
A lone flat-topped acacia tree on open grassland under a wide sky
An acacia on the Maasai Mara - savanna holds little carbon per hectare over a vast areaPhoto: Daniel Case, CC BY-SA 4.0, via Wikimedia Commons (resized)

Who it's for

Built for non-specialists and technical users alike

A steep hillside covered in bamboo and dense montane forest
County environment & forestry teamsPhoto: Shyamal L., CC BY-SA 4.0, via Wikimedia Commons (resized)
A single young mangrove tree on a tidal mudflat with mangrove forest behind
Conservation & restoration organisationsPhoto: Faizan2023, CC BY-SA 4.0, via Wikimedia Commons (resized)
Two small owls, one grey and one rufous, perched together on a branch
Researchers & studentsPhoto: John Mwacharo, CC BY-SA 4.0, via Wikimedia Commons (resized)
Dense green montane forest with tall trees and thick undergrowth
Carbon-project teams (early screening)Photo: Josep M. Gracia, CC BY-SA 4.0, via Wikimedia Commons (resized)
A large field of cabbages stretching to a few trees on the horizon
Land-use planners & NGOsPhoto: Desmond Parsimei, CC BY-SA 4.0, via Wikimedia Commons (resized)

Designed with Kenya Forest Service · KEFRI · NGOs · Students & academic researchers in mind.

In context

How Carbon-Eye compares

FeatureCarbon-EyeField inventorySingle-index map
Wall-to-wall coverageYesNoYes
Time to resultsMinutesMonthsMinutes
Multi-source predictorsYesn/aNo
Model ensemble + spreadYesNoNo
Held-out validation (R², RMSE)YesYesNo
County statistics & CSVYesManualNo
Certification-grade resultNoYesNo

Pricing

Free for county screening

  • County screening

    County officials, researchers, students, NGOs

    Free

  • Project developer

    Carbon project developers and investors at pre-feasibility

    Pricing on request

  • API and partners

    Platforms, agencies and research partners integrating Carbon-Eye

    Pricing on request

Compare plans

Questions

Frequently asked

Try Carbon-Eye for your county

Open the live Earth Engine app, or leave your email for a walkthrough, methodology note and updates.

Background photo: Gura Falls in the Aberdare forest, Nyeri - forests that feed Kenya's rivers. Photo: Matteo Frigo, CC BY-SA 4.0, via Wikimedia Commons (resized)