To observe forest cover changes, Cambodia’s Forestry Administration has used satellite images at a scale of 1:250,000 (1 cm on the map equals 2.5 kms on the ground), and updating of data on forest cover resources has been done every four years. But as the consequences of deforestation become more and more apparent, especially in an era of warming climate, there is an urgency felt among Cambodian authorities to increase capabilities to be able to spring from observation into action.
Image from Google
What is needed is technology that is able to produce updates on forest cover every so often, that would allow them to home in on what appear to be the deforestation “hotspots,” and help them take action.
“The main advantage of Terra-i is that it produces a monthly report of high-resolution data and imagery refreshed every 16 days,” says CIAT’s Louis Reymondin, co-developer of the forest monitoring system. “It is the only tool that is global in scope and yet can be calibrated according to a country’s specific context.”
In an initiative supported by the Food and Agriculture Organization (FAO), the Cambodian government looked into the feasibility of Terra-i as a monitoring tool in the country, by calibrating it for use in two pilot sites: Mondulkiri province in the country’s southeast, and Phnom Kulen National Park in tourism-famous Siem Reap province. The Ministry of Environment is exploring the possibility of developing eco-tourism in and around Phnom Kulen National Park, and it, together with development partners, is exploring a payment-for-ecosystem services mechanism – where the user of an environmental service makes a direct or indirect payment for the benefit – to support sustainable development of the national park.
Terra-i: an eye on habitat change
Terra-i produces a monthly report of high-resolution data and imagery refreshed every 16 days. Above are images of Cambodia’s Phnom Kulen National Park taken from January 01st, 2007 to August 29th, 2007.
Terra-i’s system is based on the premise that natural vegetation follows a predictable pattern of changes in greenness from one date to the next, brought about by site-specific land and climatic conditions. A so-called computational neural network is “trained” to understand the normal pattern of changes in vegetation greenness (represented by greenness of a pixel in a map) in relation to terrain and rainfall for a site, and then marks areas as changed (called “anomalies”) where the greenness suddenly changes well beyond these normal limits. Running on many computers this analysis is refreshed with new imagery every 16 days and for every 250m square of land.
Calibration had helped determine Cambodia’s natural vegetation pattern for various types of forest, including fluctuations in vegetation that are caused by natural events such as floods or droughts. Any significant deviation from this natural pattern could be safely considered as caused by human intervention.
After calibration, Terra-i recorded these alerts from 2007 – figures that are very similar to the Global Forest Change datasets:
Number of alerts every 16 days within evergreen forests, 2007-2017. The land cover class recording the highest alerts rate in Phnom Kulen was the evergreen forests. Detections of vegetation loss reached a peak in January 2013, decreased in 2014-15, only to sharply increase in early 2016.
Number of alerts every 16 days within evergreen forests, 2007-2017. The land cover class recording the highest alerts rate in Mondulkiri province was the evergreen forests. Most of the alerts occurred between 2012 and 2016, reaching a peak in February 2016.
There is an increasing need to be able to support ground action by providing information to forest rangers and provincial authorities on the state of the forest, as well as on looming encroachment.
Under the UN-REDD Programme – a collaboration between FAO, United Nations Development Programme (UNDP) and UN Environment – FAO has supported Cambodia since 2009. REDD+ stands for Reducing Emissions from Deforestation and forest Degradation. With assistance from various programmes and facilities, the Cambodian government submitted its initial forest reference level; drafted a national REDD+ strategy; and developed a national forest monitoring system, which enables the measurement and reporting of forest-related emission to the UNFCCC (UN Framework Convention on Climate Change) and of the state of the forest to the Global Forest Resource Assessment.
A near real-time monitoring system, such as the open source Terra-i is one of the next steps in the further development of the national forest monitoring system. However, necessary technology, skills, and institutional structures will need to be strengthened. Institutional structures clearly identifying processes and roles and responsibilities if and when an alert for an unusual vegetation change is detected, need to be put in place.