Land Cover & Land Use Classification – United Arab Emirates (10m Satellite Imagery)
Location: United Arab Emirates
Imagery type: 10‑metre resolution multispectral satellite imagery (e.g., Sentinel‑2)
Output format: Polygon (vector)
Project Overview
Nazru’s AI platform was deployed to produce a detailed land cover and land use map of the United Arab Emirates using 10‑metre resolution satellite imagery. The project classified the entire country into 14 distinct classes: water, wetland, pasture, residential, wild_grass, urban_built, transportation, shrubland, sand_desert, open_space_developed, industrial, forest, cultivated_crops, and bare_soil. The output supports urban planning, environmental monitoring, agricultural assessment, and sustainable development across the UAE’s diverse landscapes – from the Rub’ al Khali (Empty Quarter) desert to the coastal cities and mountain regions.
The Challenge
Classifying land cover across the UAE from 10m satellite data presents specific challenges:
Hyper‑arid desert landscape – vast expanses of sand desert (Rub’ al Khali) dominate, with limited natural vegetation.
Rapid urban expansion – extraordinary growth in Dubai, Abu Dhabi, Sharjah, and other cities creates complex built‑up patterns.
Coastal and marine features – extensive coastline along the Arabian Gulf and Gulf of Oman, with artificial islands (e.g., Palm Jumeirah, Yas Island), coastal wetlands, and sabkhas.
Agricultural areas – limited but concentrated in oases (e.g., Al Ain, Liwa) and modern irrigated farms.
Industrial zones – oil and gas facilities, ports, and free zones (e.g., Jebel Ali, Ruwais).
Separation of similar classes –
sand_desertvs.bare_soil;residentialvs.urban_built;wild_grassvs.shrubland;wetlandvs.water.
The Nazru Solution
Nazru’s platform integrated multi‑spectral satellite bands with a deep learning semantic segmentation model trained on Gulf region landscapes. The pipeline:
Pre‑processing – atmospheric correction, cloud masking, and resampling to consistent 10m resolution.
Classification – pixel‑wise assignment to one of 14 classes.
Post‑processing – raster‑to‑polygon conversion with noise removal and minimum mapping unit (MMU) filtering.
Accuracy assessment – validated against high‑resolution reference data.
Class definitions (contextualised for the UAE):
| Class | Description |
|---|---|
water | Arabian Gulf, Gulf of Oman, artificial canals, lakes, and coastal waters |
wetland | Coastal sabkhas, tidal flats, mangroves (e.g., Abu Dhabi mangroves), and seasonally inundated areas |
pasture | Grazing lands (limited, mostly in mountainous areas like Ras Al‑Khaimah, Fujairah) |
residential | Dense residential areas – villas, apartment blocks, traditional housing |
wild_grass | Sparse natural grasses and seasonal herbs (mainly after winter rains) |
urban_built | Mixed‑use built‑up areas – commercial, institutional, and high‑density urban centres (Dubai, Abu Dhabi, Sharjah) |
transportation | Roads, highways, airports, and transport corridors |
shrubland | Areas dominated by low, woody shrubs (natural vegetation in wadis and foothills) |
sand_desert | Active sand dunes, sand sheets, and desert surfaces (Rub’ al Khali, Al Dhafra) |
open_space_developed | Parks, golf courses, sports fields, and vacant developed lots |
industrial | Oil and gas facilities (e.g., Ruwais, Habshan), ports (Jebel Ali), factories, and free zones |
forest | Mangrove forests (Abu Dhabi), date palm plantations, and urban tree cover |
cultivated_crops | Irrigated farmlands, date palm groves, and oases (Al Ain, Liwa, Dhaid) |
bare_soil | Unvegetated soil, rocky surfaces, and gravel plains (not sand dunes) |
Key Results & Benefits
Detailed 14‑class land cover map of the entire UAE at 10m resolution.
Clear separation of desert, urban, industrial, agricultural, coastal, and natural vegetation classes.
Supports:
Urban planning and infrastructure development (monitoring Dubai and Abu Dhabi expansion).
Environmental conservation (mangrove protection, desertification monitoring, biodiversity assessment).
Agricultural monitoring (oasis and date palm assessment).
Industrial and transportation network mapping.
Scalable – can be applied to time‑series data for change detection (e.g., tracking artificial island construction, urban sprawl).
GIS‑ready – polygon outputs with class attributes and area measurements.
Key Technologies Used
Semantic segmentation (U‑Net / DeepLabV3+) for pixel‑wise classification.
Multi‑spectral indices – NDVI, NDWI, SAVI, and custom indices for arid land discrimination.
Polygon vectorisation – clean, topologically correct polygons.
Atmospheric correction – using standard pre‑processing pipelines (e.g., Sen2Cor).
