Monitoring cover change Why land cover?

Measuring land cover change helps monitor pressures on ecosystems and biodiversity.

Loss of biodiversity and Clearance of dry in the Argentinian pressures on ecosystem Gran Chaco region services are among the most pressing global challenges. 19921992 2015 Global biodiversity loss is so intense that it has recently been described as ‘biological annihilation’.

Detrimental changes in land cover and are the leading contributors to terrestrial biodiversity loss. These changes generally occur slowly, but they are associated with declines in species diversity

and populations, and can have a 10 30 50 61 100 120 150 180 210 11 40 60 62 110 130 153 190 0 50 100 200 Kilometers major impact on ecosystems. 2015 1992 2015 Monitoring the various elements of biodiversity at global scales is difficult. However, it is increasingly possible to measure changes in the extent and spatial structure of natural habitats. Land cover change is the best measure currently available to monitor pressures on ecosystems and biodiversity globally.

10 30 50 61 100 120 150 180 Note:210 Clearance of broad-leaved forest (green) for 0 50 100 200 Kilometers 11 40 60 62 110 130 153 190 cropland (yellow) and cattle grazing ( in brown) has occurred on a massive scale. Source: ESA/UCL Geomatics (2017) 2 Tianjin, China. Combination of 3 radar scans, Copernicus 2015. Why now?

Advances in Earth observation and data processing improve measurement of land cover changes at the global scale.

Land cover datasets are typically The OECD has taken this based on observations from satellite­ opportunity to strengthen the mounted sensors. These sensors information base for its policy record the intensity of reflectance analysis. The suite of OECD’s land of different wavelengths of the cover indicators draws on several electromagnetic spectrum from the Earth observation-based datasets. surface of the Earth. The raw data The initial batch of indicators are a are then processed into analysis- first step to improve information on: ready intermediate datasets. zz loss and gain of natural and Land cover detection is complex. semi-natural vegetated land At global scales it requires massive zz conversions between land cover volumes of data and computing classes, including conversions capacity. For algorithms to to and from cropland and automatically classify land cover conversion to artificial surfaces types they must be ‘trained’ or zz built-up area growth programmed to recognise land cover using prepared datasets where the zz surface change. land cover type is already known for a large number of locations. In recent years, new datasets have been made available that allow for Indicators on land cover change are high‑level some analysis of changes in land NOTE proxies for pressures on cover consistently at the global scale. terrestrial ecosystems These datasets are results of decades and biodiversity. These of Earth observation missions by indicators are state-of-the- different national and supranational art and will likely be further refined. There space organisations. are limitations inherent to almost all Earth observation-derived information, such as sensitivity to the resolution and classification scheme used, and also limits on what can be remotely observed and automatically classified in practice using the tools and techniques available. 3 3 Tianjin, China. Combination of 3 radar scans, Copernicus 2015. An OECD Green Growth headline indicator

Changes in the biophysical characteristics of natural habitats are the best proxy available to monitor pressures on ecosystems and biodiversity.

The OECD headline indicator seeks land. This broad defi nition has the to measure changes from more advantage of being fl exible enough natural to more anthropogenic land to apply usefully across the Earth’s cover types. This is because at the biomes in all their extraordinary global scale, natural vegetated areas diversity. are critical for the conservation of In its current edition, the headline biodiversity and the provision of indicator distinguishes nine broad ecosystem services. land cover classes: tree cover, The headline indicator measures , wetland, shrubland, sparse changes in natural and semi­natural – classifi ed as (semi­) vegetated land which is used to natural land – as well as cropland, identify the less anthropogenic bare land, inland water and artifi cial and less intensively used vegetated surfaces.

Monitoring progress towards Green Growth Our ability to sustain economic and social progress in the long run will depend on our capacity to reduce dependence on natural capital as a source of growth, abate pollution, enhance the quality of physical and human capital and reinforce our institutions. An indicator on changes in land use and land cover has been included in the set of Green Growth headline indicators. They are to be used in the country reviews carried out by the OECD, in policy analyses at national and sub-national level, and in public communication by the OECD. Read the 2017 report and browse data on http://oe.cd/ggi.

4 Urbanisation in the Alpine valleys

Land cover change in Haute-Savoie (left), Lake Geneva and Valais (upper left, centre and right) and Aosta (lower-right) regions of France, Switzerland and Italy. Urbanisation (red) of cropland and grassland (yellow and orange) is widespread.

19921992 20152015

Source:10 ESA/UCL-Geomatics12 40 61 (2017)80 100 130 180 200 210 11 30 60 70 90 110 150 190 201 220 0 15 30 60 Kilometers

This indicator measures quantity (the area converted), not quality or value. It cannot distinguish between the loss of habitats NOTE with high biodiversity value (e.g. primary tropical ) and with lower value (e.g. some commercial forestry). Furthermore, it does not capture degradation of the land if the class remains the same. Future work will attempt to better identify sensitive ecosystems where supported by available data at the global scale.

5 Loss and gain of natural and semi-natural areas

OECD and G20 countries continue to convert land from its more natural state to more anthropogenic systems, with potentially harmful impacts on biodiversity and ecosystems.

Worldwide, 2.7% of (semi­)natural vegetated land has been lost to other land cover types since 1992. This represents an area twice the size of Spain.

z OECD and G20 countries account for over half of this loss, which occurs primarily in Brazil, the People’s Republic of China, the Russian Federation, the United States and Indonesia.

z The gap between (semi­)natural land gain and loss for OECD and G20 countries is approximately the area of France: 636 000 km2

z Among OECD countries, from 1992 to 2015, the most intense losses of (semi­)natural land have occurred in Korea and Israel.

Most natural and semi-natural land is converted to cropland Loss of natural and semi-natural vegetated land by cover type,1992-2015

To cropland 66% To arti cial To To bare OECD surfaces water land 16% 9% 8%

World 81% 5% 5% 9%

The headline indicator is based on data from the Climate Change Initiative Land Cover project, lead by the European Space Agency. NOTE These datasets provide information on change between land cover types. They aim to provide a harmonised, stable and consistent account of long- term land cover status and changes that is ‘immune’ to short-term seasonal and temporary changes (e.g. snow cover in winter, crop rotation, forest burn scars). This data product has been developed to support climate modelling and contribute to the Global Climate Observing System. 6 Figure 3. Loss and gain of natural and semi-natural Lossvegetated and gainland, of1992-2015 natural and semi-natural vegetated land, 1992-2015 Loss Gain 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 (%) Korea Israel Indonesia Estonia Portugal Latvia Saudi Arabia Brazil Slovenia Denmark Italy Spain Turkey Netherlands Japan India Lithuania Belgium Czech Republic Germany China Luxembourg EU28 United Kingdom Costa Rica Argentina Poland Austria France Switzerland Hungary World Sweden Slovak Republic Finland Gains of (semi-)natural G20 land are unlikely to Greece NOTE compensate for losses: Mexico the net changes should be Norway Colombia interpreted carefully. For South Africa example, the costs of the United States loss of old-growth forest are not comparable OECD to the benefits provided by equally sized new Chile forest plantation. The aggregated category of Ireland “natural and semi-natural vegetated land” can Australia Russian Federation also conceal important land cover conversions New Zealand between tree-covered areas, , Canada wetland, shrubland and sparse vegetation, Iceland such as forest clearing for grazing. 7 Agricultural expansion

Agricultural expansion is the Conversions of (semi-)natural main driver of natural and vegetated areas to cropland, 1992-2015 (% 1992) semi-natural land loss. (%) 0246810 12 14 Korea Indonesia Losses in (semi­)natural Estonia vegetated land are most often Latvia Portugal due to conversions to cropland. Israel Slovenia Of OECD countries, Korea, Brazil Estonia, Latvia and Portugal saw Spain Italy the most intense conversions to Denmark cropland. Turkey Japan Indonesia In most countries, new Lithuania cropland is converted primarily Czech Republic Luxembourg from tree­covered areas but China conversions from grassland Costa Rica EU28 and shrubland are important Argentina in some countries. In some Poland Germany cases the change may be due Belgium to re­cultivation of previously France Hungary abandoned . World Slovak Republic Austria Sweden G20 United Kingdom Greece Mexico Finland Switzerland Colombia South Africa Netherlands OECD United States Norway Australia Russian Federation Ireland Chile Saudi Arabia New Zealand Canada Iceland 8 Patterns of land cover change 1992-2015, thousand km2

OECD To other: 340 China From treecover: 90 From tree cover: 467 To cropland: 155

To wetland: 15

To cropland:244 From grassland: 117 To other:50 From grassland: 227 To wetland: 3

To grassland: 241 To grassland: 106 From other: 135 From other: 426

To treecover: 501 From wetland: 1 To treecover:74 From wetland: 123

From cropland: 250 From cropland: 133 To artificialsurfaces: 88 To artificialsurfaces: 152

Indonesia Brazil

To cropland:364 From tree cover: 440 From treecover: 103 To cropland:100

To grassland: 31 To wetland: 19 From other:4 To other:5 From other: 102 To other: 103

To treecover: 50 From grassland: 5 From cropland: 54 From wetland: 7 To tree cover: 111 From cropland: 86 To artificialsurfaces: 7 To arti cial surfaces:12

Some conversions from “natural” areas to agricultural land may be beneficial. For example, some types of traditional farmland provide essential NOTE habitats for birds and may have higher biodiversity than some tree-covered areas. The observed conversions from wetlands do not necessarily represent the loss of, for example, Ramsar-type wetland like marshes, peatlands, floodplains, saltmarshes, mangroves, intertidal mudflats, etc. 9 Urbanisation

Conversions from cropland Urban expansion is another to artificial surfaces,1992-2015 major driver of land cover (% of 1992 cropland)

change. Construction of (%) 0 2 4 6 8 10 12 buildings and other artificial Japan Switzerland surfaces contributes to the Belgium loss of sensitive ecosystems Netherlands Luxembourg and fragmentation of natural Germany Austria habitats. Norway Czech Republic Slovak Republic Most new artificial surfaces Israel are built on cropland, with the Slovenia United Kingdom exception of a handful of countries Hungary where development mostly takes Portugal France place on tree‑covered areas, China EU28 grasslands or shrubland. Ireland Denmark In addition to the pressures on Italy biodiversity caused by the loss Korea Poland of farmland and the associated OECD natural habitats, conversions G20 United States to artificial surfaces involve soil Sweden sealing, which irreversibly degrades Spain Saudi Arabia soil and increases flood risk. Air, World noise and light pollution and Indonesia Mexico decreasing access to extra-urban Costa Rica Estonia green space affect quality of life. Lithuania India Among OECD countries, Japan, Greece Switzerland, Belgium, the New Zealand Russia Netherlands, and Luxembourg saw Latvia relatively large conversions from Canada Brazil cropland to artificial surfaces. Turkey Argentina Colombia Note: Cropland does not include pasture land, Chile so the total ‘lost’ agricultural land may be larger. South Africa Australia Finland Iceland none observed 10 Globally, an area the size of the United Kingdom (244 000 km2) has been converted to built-up areas since 1990. Built-up area in thousand km2 in 2014 and new constructions since 1990.

OECD USA Germany France 27 Rest of 23 5 150 6 since since 1990 OECD 1990 Japan 123 38 Italy 31 16 27 since since 4 4 since since 1990 1990 1990 1990

Rest of Russia the world South China Africa 24 16 7 4 since since 193 1990 105 1990 India Brazil 32 Indonesia 69 21 16 21 46 5 9 since since since since since 1990 1990 1990 1990 1990

The Global Human Settlement Layer developed by the European Commission’s Joint Research Centre complements the headline indicator NOTE with more detailed information on built-up areas – their extent and change over time. This dataset allows for a more accurate mapping of human populations. Built-up areas mean presence of buildings, where a building is defi ned as any roofed built- up structure. This defi nition excludes other parts of urban environments such as paved surfaces (roads, parking lots), commercial and industrial sites (ports, landfi lls, quarries, runways) and urban green spaces (parks, gardens). The measured area is smaller than “urban area” defi ned in broader land use terms. 11 Surface water change

Globally, since 1984, around 180 000 km2 of Permanent surface water land was inundated, mainly through dam- change, 1984-2015 (%) building, with significant consequences on 25 20 15 10 5 0 0510 15 20 25 30 35 40 SAU (200%, n.a.) freshwater ecosystems. During the same ESP (48%, 6%) 2 IND period, 90 000 km (the area of Portugal) of ZAF surface water was lost through drought and TUR MEX unsustainable abstraction for irrigation. CHN CRI AUS Damming is known to be one of the most IDN COL important anthropogenic impacts on DEU FRA freshwater ecosystems. Dams fragment BRA river systems and potentially block PRT GRC migration routes, leading to the loss of KOR megafauna. They change the downstream ITA ARG flooding patterns and sediment deposition World RUS leading to the loss of floodplains, riparian G20 zones and wetlands. Global dam building is JPN POL booming with thousands of major projects EU28 USA planned, including a large number in AUT biodiverse locations like the Amazon basin. OECD CHL Dam-building seemed notably intense and NZL extensive in India, China and Brazil. CAN LVA Globally, surface water loss has been most HUN SWE extensive and most intense in Central IRL CHE Asia where it was (and continues to be) LTU driven by droughts, river diversion and FIN DNK abstraction. Elsewhere, losses are often EST caused by long-term droughts. Loss SVK Gain Permanent to not water GBR Not water to permanent It is important to note that both surface Permanent to seasonal ISL Seasonal to permanent NOR water losses and gains can be detrimental NLD LUX to biodiversity and ecosystems. CZE SVN ISR BEL 12 Note: 1984-2015 change as share of 1984 permanent water Surface water change of a section of the Paraná river in Argentina, 1984-2015

25 20 15 10 5 0 0510 15 20 25 30 35 40 SAU (200%, n.a.) ESP (48%, 6%) IND ZAF TUR MEX CHN CRI AUS IDN COL DEU FRA BRA PRT GRC KOR ITA ARG World RUS G20 JPN POL EU28 USA AUT Source: Pekel et al. (2016) OECD CHL NZL CAN LVA HUN SWE IRL The Global Surface Water Change, a project CHE LTU of the European Commission’s Joint Research FIN NOTE Centre, documents different facets of surface DNK water state at a 30m resolution globally. This EST unprecedently rich dataset provides information Loss SVK Gain on water occurrence, recurrence, seasonality, Permanent to not water GBR Not water to permanent Permanent to seasonal ISL Seasonal to permanent water change intensity and transitions. NOR NLD Permanent surface water is defined as areas that were water LUX for every month of the reference year. Seasonal surface water CZE is defined as areas that were water for 1 to 11 months of the SVN reference year. These data refer only to water surface area, they ISR BEL do not estimate the volume of water gained or lost. Note: 1984-2015 change as share of 1984 permanent water 13 Marajó island, Pará, Brazil. Copernicus Sentinel-2 images Earth in 13 spectral Next steps channels, ESA, 2017.

z The suite of OECD land cover indicators presented here show initial results from several global land cover mapping efforts. The indicators allow, for the fi rst time, some comparison of changes across a range of land cover types at the global scale in a consistent manner. This is a major advance compared to what was previously possible.

z The indicators will be complemented, or superseded, as new datasets become available in the future, potentially, those developed for the Global Biodiversity Observing System to measure a set of Essential Biodiversity Variables (EBV).

z Insights into the extent and changes in land fragmentation could be developed. Methodologies have been established, but it is not clear whether available data could support a global indicator. Development of a habitat fragmentation indicator at the global scale would be a valuable extension.

Monitoring land cover change globally can support countries in their efforts to measure progress towards the United Nations’ Sustainable Development Goals, the Aichi Biodiversity Targets and other international targets

Aichi target 5: By 2020, the rate of loss of all natural habitats, including forests, is at least halved and where feasible brought close to zero, and degradation and fragmentation is signifi cantly reduced. See also target 15.

Sustainable Development Goal 15: Protect, restore, and promote sustainable use of terrestrial ecosystems, sustainably manage forests, combat desertifi cation, and halt and reverse land degradation and halt biodiversity loss. See also Goals 6 and 11.

14 Marajó island, Pará, Brazil. Copernicus Sentinel-2 images Earth in 13 spectral channels, ESA, 2017.

15 Loss of biodiversity and pressures on ecosystem services are global challenges. Monitoring Land cover change is the best measure available to monitor pressures on terrestrial ecosystems and biodiversity. The OECD is developing new indicators on Land cover land cover changes and conversions. change Work on land cover change indicators is in progress. Read the full working paper and browse the data: http://oe.cd/land-cover

CONTACT Environmental Performance and Information team Head of Division Nathalie Girouard [email protected] Senior Economist Ivan Haščič [email protected] GIS Analyst Alex Mackie [email protected]

This document and any map included herein are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area.

The statistical data for Israel are supplied by and under the responsibility of the relevant Israeli authorities. The use of such data by the OECD is without prejudice to the status of the Golan Heights, East Jerusalem and Israeli settlements in the West Bank under the terms of international law.

April 2018

Image credit: Reindeer Island, Manitoba, Canada. Copernicus Sentinel data (2017) processed by ESA