Geography

What is Land Cover? Definitions and the Role of Remote Sensing

Land cover refers to the physical and biological material that occupies the Earth's surface — the actual substance of what covers the ground, whether forest canopy, open water, bare rock, snow, or asphalt. It is fundamentally different from land use, which describes the human purpose assigned to that surface (e.g., a forested area may be a timber production zone or a conservation reserve — the land cover is forest in both cases, but the use differs).

Geographers map land cover at global scales using remotely sensed imagery — data collected by satellite-borne and airborne sensors that detect electromagnetic radiation reflected or emitted from the Earth's surface. The two principal sensing methods are:

• Passive remote sensing: instruments detect reflected solar radiation or emitted thermal infrared energy. The Landsat series (NASA/USGS), Sentinel-2 (ESA), and MODIS (NASA) are key passive sensors. Landsat has the longest continuous archive (since 1972), enabling change-detection studies over decades.
• Active remote sensing: the sensor emits its own energy pulse (radar or lidar) and measures the return. Synthetic Aperture Radar (SAR), as on ESA's Sentinel-1, can penetrate cloud cover and detect surface structure, making it invaluable for monitoring tropical forests and wetlands where persistent cloud obscures passive sensors.

Satellite imagery is processed into land cover classification maps using spectral signatures — characteristic patterns of reflectance across wavelengths unique to each surface type. Vegetation reflects strongly in the near-infrared and absorbs red light; built surfaces reflect broadly across the visible spectrum; water absorbs near-infrared almost completely. The Normalised Difference Vegetation Index (NDVI) — calculated as (NIR − Red) / (NIR + Red) — quantifies vegetative vigour and is widely used to delineate vegetated from non-vegetated surfaces.

Key global land cover datasets produced from remote sensing include the ESA Climate Change Initiative Land Cover (CCI-LC) dataset at 300 m resolution, the Global Forest Watch (University of Maryland/Google), and the MODIS MCD12Q1 product. These datasets typically employ the United Nations Land Cover Classification System (LCCS) or the International Geosphere–Biosphere Programme (IGBP) legend, both of which recognise 17 major land cover classes.

Once classified, data are frequently displayed as choropleth maps — thematic maps that shade geographic units (grid cells, countries, or biomes) according to quantitative values such as percentage forest cover or cropland area. Choropleth maps allow rapid identification of spatial patterns but require careful choice of class boundaries (e.g., equal interval vs. quantile classification) to avoid misleading visual impressions.

Applied example — Australian context: Geoscience Australia and the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) produce the Australian Land Use and Management (ALUM) Classification, updated using Landsat time-series. The 2021 ALUM national dataset revealed that production from dryland agriculture (including livestock grazing on native vegetation) dominates the Australian land cover footprint, occupying approximately 54% of the continent, while conservation and natural environments account for roughly 38%.

Spatial Distribution of Forests and Deforestation Pressures

Forests are the most biologically complex and carbon-dense terrestrial land cover type. Globally, forests cover approximately 4.06 billion hectares (FAO Global Forest Resources Assessment 2020) — roughly 31% of Earth's land area. Their distribution is governed principally by temperature and moisture availability, producing three broad latitudinal forest belts:

• Tropical and subtropical forests (equatorial and sub-equatorial latitudes, roughly 23.5°N–23.5°S): The largest contiguous blocks exist in the Amazon Basin (Brazil), the Congo Basin (Democratic Republic of Congo), and the Indo-Pacific Arc (Indonesia, Papua New Guinea, Malaysia). Year-round warmth and high precipitation sustain multi-storeyed closed-canopy rainforest with exceptional biodiversity. The Amazon alone stores approximately 150–200 billion tonnes of carbon.
• Temperate broadleaf and mixed forests (mid-latitudes, roughly 25°–50° N and S): Concentrated in eastern North America, Europe, eastern China, and south-eastern Australia. Seasonally deciduous or mixed evergreen–deciduous structure. Much of Europe's original temperate forest was cleared millennia ago; eastern Australia's eucalypt forests represent the largest remaining temperate woodland complex in the Southern Hemisphere.
• Boreal forest (taiga) (high latitudes, roughly 50°–70°N): Stretching across Russia (Siberia), Canada, Alaska, and Scandinavia. Dominated by coniferous species (spruce, fir, pine, larch). Contains approximately 30% of the world's terrestrial carbon pool, much of it locked in permafrost soils rather than biomass.

Remote sensing has enabled precise quantification of forest loss. Global Forest Watch data (Hansen et al., 2013, updated annually) revealed that between 2000 and 2023 the world lost approximately 3.5 million km² of tree cover — an area larger than India. Tropical primary forest loss (the most ecologically damaging category) averaged approximately 4.1 million hectares per year between 2015 and 2020.

The spatial pattern of deforestation is not random. It clusters along:

• Agricultural frontier zones — the arc of deforestation in the Brazilian Cerrado and Amazon; palm oil expansion frontiers in Borneo and Sumatra.
• Road networks — forest loss propagates outward from newly constructed roads ('fishbone' deforestation pattern, clearly visible on Landsat imagery of Rondônia, Brazil).
• Political boundaries — contrasts are visible across borders where governance and enforcement differ; the Colombia–Brazil Amazon border and the DRC–Republic of Congo boundary are classic remote-sensing examples.

Australian applied example: Queensland is globally recognised as a deforestation hotspot. Between 1988 and 2006, Queensland cleared more than 7 million hectares of native vegetation, primarily for cattle grazing. Reintroduction of the Vegetation Management Act 2004 (Qld) reduced clearing rates, but amendments in 2016 allowed re-clearing of regrowth, and satellite-based monitoring by the Queensland Department of Environment and Science (using Landsat/Sentinel-2 composites) recorded clearing rising again after 2016. This Queensland case study illustrates how remote sensing provides the evidence base for regulatory policy: annual statewide clearing reports use Statewide Landcover and Trees Study (SLATS) data derived from satellite imagery.

In choropleth mapping of forest cover, a critical consideration is the definition threshold applied: the FAO defines forest as land with >10% canopy cover and trees >5 m tall over an area >0.5 ha — a definition broad enough to include sparse woodland and plantation. Band A students should critique map data by interrogating the underlying classification rules.