Casablanca-Settat
Fourth region processed: forest cover, species, dieback, biomass and carbon stock. A residual confusion between cork oak and thuya is still being corrected. Biomass and carbon are estimates, with their limits and comparisons detailed below; a version recalibrated on ground truth is available on request.
These layers are sold by region, as high-resolution downloads. Contact us
Why this data
Free global products (Hansen, ESA WorldCover) never distinguish forest species. This species map is built from ground truth, combined with satellite embeddings and a time series of vegetation indices, with training restricted to pure stands and a model-confidence filter to discard unreliable extrapolated areas. Cork oak, eucalyptus, Aleppo pine and thuya are modeled across the whole region; acacia, Canary Island pine and Calabrian pine are too rare in the data to generalize and are reported only at directly surveyed locations. A residual confusion between cork oak and thuya is documented, pending a complementary structural signal (radar) to improve it. Forest dieback is computed on the same footprint (official mask minus Lande). Biomass and carbon layers are published as estimates: their limits and comparisons with other sources are detailed below, and a version recalibrated on ground truth is available on request. Full methodology documented, known limitations stated explicitly.
Biomass and carbon stock: estimate, limits and comparisons
These two layers are published as estimates. The model is a gradient boosting model calibrated on GEDI spaceborne LiDAR measurements (about 37,000 points in the area), using annual satellite embeddings, a harmonic time series of vegetation indices, elevation and rainfall (no raw Sentinel-2 imagery or radar, unavailable for this region). It covers about 93,000 ha of forest.
Regional result: 14.4 Mg/ha of aboveground biomass on average, about 1.35 million tonnes in total, and about 8.5 Mg C/ha of carbon in trees (about 0.79 million tonnes of carbon, 2.9 million tonnes of CO2 equivalent). Belowground biomass is estimated at 25% of aboveground and the carbon fraction at 0.47 (IPCC default values, not species-specific).
Comparison with other sources
| Source | Value | What it indicates |
|---|---|---|
| This model | 14.4 Mg/ha aboveground biomass (regional mean) | Brutia and Canary Island pine highest (46-53, marginal areas), thuya and Aleppo pine about 25, eucalyptus 12, cork oak 9 Mg/ha. |
| ESA CCI Biomass v6.0 (2022), same footprint | 21.9 Mg/ha | The model is about 34% lower here, the opposite of Rabat-Salé-Kénitra where it was higher: the regional comparison depends a lot on the chosen reference. |
| Model accuracy (block validation, unseen at training) | R² 0.49, near-zero mean bias | Better than Rabat-Salé-Kénitra (R² 0.30), but extreme values are still flattened toward the mean: areas observed above 100 Mg/ha are predicted at about 65 Mg/ha, and very open areas are overestimated. |
| Eucalyptus, Sidi Yahya-Gharb plain, north-western Morocco (Boulmane et al. 2017) | Total aboveground biomass 48 Mg/ha at end of rotation, productivity 4.4 Mg/ha/year | Best analog found (same sandy Atlantic-plain context): about 4 times the model's mean for eucalyptus (11.6 Mg/ha), but measured on a stand at the end of its rotation, not on a landscape average. No numeric reference found for thuya or Aleppo pine in this region. |
| GEDI L4A in Mediterranean forest (Forest Ecology and Management, 2023) | Overestimation below 50 Mg/ha, underestimation above | The model is calibrated on GEDI: it inherits this bias, in either direction depending on stand density. |
Limits to be aware of
Per-pixel accuracy remains limited despite its improvement (R² 0.49). High biomass is underestimated and very open areas overestimated. No published biomass study exists for this region, unlike the Maâmora forest for Rabat-Salé-Kénitra. These layers are meant to compare areas with each other, not to build a carbon budget.
A version recalibrated on ground truth (inventory plots and local equations) is available on request. Contact us
Layers
Automated satellite detection (3 VIIRS satellites, about 6 passes a day over Morocco), not verified on the ground. A hotspot can be agricultural burning, an industrial heat source, or a forest fire: the satellite cannot tell them apart. 3-day window.
Click a region to see its available layers. Tanger-Tetouan-Al Hoceima is the first region processed, others follow one by one.
Sources
- Basemap: Esri World Imagery
- Admin. boundaries and regional division: official Morocco sources
- Relief (DEM): SRTM, 30 m resolution (NASA/USGS)
- Watersheds: extracted from the SRTM DEM
- Dams: Geoportail's own data
- Protected areas: UNEP-WCMC and IUCN (2026), Protected Planet, WDPA, September 2026
- Hotspots: data and imagery from NASA FIRMS (Fire Information for Resource Management System), part of LANCE (Land, Atmosphere Near real-time Capability for Earth observations), near real-time
- Cities: geographic landmarks, population from the 2024 census (HCP)
Casablanca-Settat
Available layers
Fourth region processed. Species mapped from ground truth, training restricted to pure stands, using satellite embeddings and a time series of vegetation indices. Cork oak, eucalyptus, Aleppo pine and thuya modeled across the whole region; acacia, Canary Island pine and Calabrian pine are too rare in the data to generalize and are reported only at directly surveyed locations.
Basemap: Esri World Imagery satellite imagery. DEM: SRTM, 30 m resolution — major watersheds are extracted from this DEM. Relief and dams: Geoportail's own data. Administrative boundaries and regional division: official Morocco sources. Cities and population/household data: 2024 census (RGPH), Haut-Commissariat au Plan (HCP).