Global Forest Resources Assessments


The FRA 2020 Remote Sensing Survey (FRA 2020 RSS) relies on the photointerpretation of a global sample of high-resolution satellite imagery by a worldwide network of national experts. This network is trained in applying FRA 2020 remote sensing tools and methods. Involving people with local field knowledge of the vegetation and land uses in the assessment contributes to enhancing the quality of the data collected. 

Two task force meetings were held to discuss and define the most appropriate methodology and tools for the FRA 2020 RSS. As a result, around seven pilot countries have been tested to ensure the consistency of the methodology. The FRA RSS process and methodology build on the experience from the Global Forest Survey (GFS) 2015 and the FRA RSS 2010.


The FRA 2020 RSS is based on a sampling design that aims at deriving reliable statistics particularly for the changes in forest area. The sample was drawn from a global grid of equal area hexagons. A stratified random sampling design is used to reduce the uncertainty of the forest areas change estimates. The design is based on eighty strata derived using a combination of the 20 Global Ecological Zones and four categories of tree cover change: “Small change”, “Big Change”, “No change, No Forest” and “No change, Forest”. The map shows in total about 430 000 remote sensing samples selected of hexagons (40 ha) and centroid (1 ha) all over the world.


The analysis of satellite imagery for the FRA 2020 RSS is conducted through participatory and collaborative approaches, engaging national experts in learning-by-doing training sessions on the latest remote sensing methodology and tools, including the Collect Earth Online Platform which FAO has developed in collaboration with NASA and Google.

The assessment of the current status of the sample sites is based on Sentinel-2 mosaics and the land use changes are assessed using Landsat mosaics for the target years 2000, 2010 and 2018, thereby ensuring global consistency of the results. Very high-resolution images, freely available from Google Earth and Bing Maps, are also used as auxiliary data to facilitate the understanding and classification of Landsat and Sentinel mosaics. The data collected by national experts will undergo quality control, done by the FRA team in FAO. To assess the accuracy of the area estimators, a sub-sample (<10% of the sample) will be reinterpreted by external experts. The results will be analyzed and presented at global, regional and Global Ecological Zone levels (GEZ). 


Collect Earth Online (CEO) is the next generation of web-based, crowd-sourcing technology for Earth science analyses. It allows users to collect reference data using high-resolution satellite images and big-data analysis through Google Earth Engine.

Multiple users can simultaneously collect information. Users do not need to worry about software installation and data management. Everything runs online and users can focus on applications.