Brazil: An international team of researchers from Colombia, the UK, USA and Switzerland developed Terra-i. According to researchers, it is the first ever system to monitor deforestation across Latin America in near real-time using satellite data. Preliminary results from the new system revealed that in parts of Colombia, deforestation increased by 340 percent since 2004; and over a million hectares of forest have been lost in the Gran Chaco region of Paraguay.
The Terra-i was launched this week during the Rio+20, UN environment conference. Although Brazil has had a sophisticated near real-time deforestation monitoring system in place since 2008, until now there has been no equivalent for the rest of Latin America.
Terra-i has been developed to monitor changes in land cover every 16 days and for every 250 metres on the ground, in order to help national governments, conservation organisations and those implementing climate-related policy to assess recent trends in deforestation and emerging hotspots of change. The system uses data supplied by NASA’s MODIS satellite sensor and is the result of collaboration between the International Center for Tropical Agriculture (CIAT) in Colombia, The Nature Conservancy (TNC) in the USA and South America, the School of Engineering and Management of Vaud (HEIG-VD) in Switzerland and King’s College London.
Deforestation can lead to widespread loss of biodiversity and also impacts the ‘ecosystem services’ that foster a stable climate and secure freshwater supplies. However, in many parts of the world the scale and pattern of deforestation is infrequently and inconsistently monitored and this makes management of change very difficult.
Huge volumes of data need to be processed to detect land cover change at a 250m spatial resolution every 16 days. Moreover, separating real human-induced changes, such as deforestation, from changes brought about by natural seasonality and by droughts, floods or persistent cloud cover, has made the development of an operational monitoring system a real challenge. The availability of MODIS imagery means that assessment of land cover change can be made in a geographically consistent manner between countries and also updated frequently.
The development of the Terra-i system was led by Louis Reymondin, PhD student in the Department of Geography at King’s College London, supervised by Dr Mark Mulligan, in collaboration with CIAT and HEIG-VD and funded by TNC.
“We developed a computational neural network and ‘trained’ it with data from 2000-2004 to recognise the normal changes in vegetation greenness due to seasonal variation in rainfall in different areas,” said Dr Mulligan.
‘The network now recognises where and when greenness suddenly changes well beyond these normal limits as a result of deforestation. The system runs on data for every 250 square metres of land from Mexico to Argentina shortly after the data comes in from MODIS and highlights every 16 days the pixels that significantly change, writing these results to Google Maps for easy visualisation,’ he said.
Source: King’s College London