Using satellite data to track changes in earth’s surface


A team from the Earth and Life Institute has generated a database describing land-use change so thematically detailed that it is now used by institutions such as the OECD and the United Nations Food and Agriculture Organization (FAO). The database consists of a series of maps characterising the earth’s surface and its evolution from 1992 to the present, generated by using satellite earth observation data.

At the Earth and Life Institute’s Geomatics Laboratory, they still can’t believe it. In 2010, when the European Space Agency (ESA) entrusted Prof. Pierre Defourny's team with the task of characterising land use, they didn’t think they’d be able to map land use on a global scale consistently across time and space in a way so unique and efficient that the scientific result would be used – as it now is – by major institutions around the world.

The project’s premise dates back to the end of the 2000s. At the time, the United Nations became aware of the need to encourage measures to combat climate change and its repercussions. To address this concern, ESA is implementing the Climate Change Initiative (CCI). The objective is to monitor a series of essential climate variables – ozone, surface temperature, land cover – and measure their evolution by using satellite images. Concretely, it’s a question of revisiting the classification algorithms of images, beginning with those in European satellite archives, in order to track land-use evolution. To do so, the processing chain must allow for reproducing maps regularly over the years, in order to see the evolution of the variables. Once the product is finished, climate modellers, organisations, companies and laboratories are free to use it according to their own needs.

24 maps characterising earth’s surface evolution

We’re all aware that in order to understand and anticipate how the earth’s surface changes as a result of human impact and global warming, we have to monitor in detail certain “vital” parameters’, explains Céline Lamarche, a bioengineer involved in the project. ‘This is why the European Space Agency launched this programme. Participating in this project means having the opportunity to contribute, even minimally, to a solution to the great challenge of climate change. Our contribution is to develop a dataset that characterises global land cover on an annual basis from 1992 to the present day.’ Twenty-four maps – one per year – were created. They make it possible to describe the world using 22 classifications, such as forest areas, agriculture, water and cities, whose monitoring is crucial.

We’re responsible for deriving land-use information from satellite images’, she continues. ‘This is among the input data for European climate models. It’s the result of ten years of work that allowed us to extract information as pure and consistent as possible from noisy satellite data. By improving the quality of “land cover” data, we contribute to improving the representation of the earth's surface in climate models. This is the first time such a product has been created, and we already have a lot of users and not just in the climate realm. In addition to tracking, satellite observations have also allowed us to characterise the seasonality of earth's surface: we can give probabilities of snow, fire and vegetation quantity down to the square kilometre’.

The first maps are medium resolution, the pixels corresponding to areas of 300 by 300 metres. In the future, they should be in high resolution thanks to data provided by the;Sentinel  Satellites of the European Union’s Copernicus programme, which allow for pixels corresponding to 10 meters a side. ‘These open access satellite data revolutionise the way we can observe the earth and process this huge amount of information. This opens the field of possibilities for researchers.

A dialogue between communities

Four to five people work on the ESA CCI project permanently at UCLouvain’, Ms Lamarche says, ‘But our series of land-use maps could not have been created without the collaboration of a dozen other scientific partners, including climatologists. This is the project’s great strength. Working hand in hand with our users has led us to a mutual understanding of their data needs but also to an assurance of their knowledge and good practice. We’ve also developed a tool to facilitate our maps’ use directly by climatologists. If each entity had worked more independently, we wouldn’t have arrived at this result.

In 2017, the Louvain-la-Neuve team disseminated the maps publicly. They have been used by major organisations, such as the Organization for Economic Co-operation and Development (OECD), which uses them to establish ‘Green Growth Indicators’ for countries to determine whether their growth is at the expense of their natural capital. The United Nations Food and Agriculture Organization (FAO) uses them to track the evolution of land-use statistics by country. The UN World Food Programme is another user, as are private industries, which use the maps in flight simulators or for educational materials.

Continuing time series

The project has matured to the point of tangibly serving society’, Ms Lamarche says. ‘It’s fortunate for us that the project’s results have been adopted and we’ve succeeded in building a community of more than 1,500 users with whom we continue to interact.’ The open access map series can be downloaded and viewed interactively.

While Defourny's team will continue to refine and exploit maps covering 1992 to 2015 through new ESA research funding, it will also develop new maps for 2016 to 2019. ‘The product, which is considered sufficiently mature by the community, has been selected to enter an operational phase through the EU’s Copernicus Climate Change Service C3S. When a team of climate modellers decides to adopt your product, it must cover the longest possible period and be updated frequently’.

Anne-Catherine De Bast

A glance at Pierre Defourny's bio and Céline Lamarche's bio

Pierre Defourny is a full professor at UCLouvain and heads the Environmetrics and Geomatics Laboratory (UCLouvain/ELI-Geomatics), which focuses on local and global-scale remote-sensing terrestrial surface-monitoring, as well as geographic information systems land-use modelling to support agriculture, forestry and land-use planning.
He is the scientific leader of the eBELCAM and Sen2-Agri projects, and of the ‘earth surface’ component of the European Space Agency’s 2010 Climate Change Initiative to improve global climate model simulations. His research is leading to the development of operational methods for extracting statistical information from global time series, including crop monitoring and land-use changes in the context of climate change.

Céline Lamarche is a research assistant at the Earth and Life Institute (UCLouvain). After a master's degree in environmental science and technology bio-engineering, she joined the Earth Observation research team of the Environmetrics and Geomatics Laboratory in 2010. She began using satellite images to detect tropical forest changes for the Stereo II and EU-FP7 projects. She then participated in the ESA CCI project (Land Cover Project) for land-cover classification and characterisation of surface dynamics using medium-resolution (300-1000m) time series. She is currently working on this time series and developing high spatial resolution terrestrial mapping with the Sentinel-2 (10m) satellite at the continental scale.

Published on March 12, 2019