Fynbos-monitoring tool beats a trail for fire-prone areas
A South African-led team has won a UN conservation award for its near-real time monitoring of the Cape’s unique fynbos.
The Cape Floristic Region is the smallest of the globe’s six floral biomes, with very high biodiversity.
According to the UN Educational, Scientific and Cultural Organisation, a fifth of Africa’s flora are contained in this landscape, which takes up less than 1% of the continent’s land area. But this region, parts of which form a UN Educational, Scientific and Cultural Organisation world heritage site, is under threat from drought, increased fires and invasive species.
"By detecting potential threats to the ecosystem in near-real time, our tool can inform the responses of conservation authorities, citizen scientists and policy makers while simultaneously collecting data for long-term ecological research," says Jasper Slingsby, a biodiversity scientist at the South African Environmental Observation Network who headed the project.
The tool — called Emma (ecosystem monitoring for management application) — uses satellite images to produce near-real time reports on the state of and changes in vegetation. The pilot won the UN’s Data for Climate Action Challenge’s climate mitigation category.
"Big data and analytics is a game changer for policy making," says Robert Orr, a special adviser to the UN secretary-general on climate change.
"There is just one problem: much of the data is privately owned and we have to find ways to share it."
While these tools exist for forests, such as the Global Forest Watch, which monitors forest health and deforestation in real time, most other ecosystems, such as savannas, do not have the same monitoring tools. This is the first time that the fynbos can be monitored in near-real time using large satellite-derived data sets.
Emma is a collaboration between Slingsby; Glenn Moncrieff, a data scientist at local company Ixio Analytics; and University of Buffalo bioecologist Adam Wilson in the US.
The ability to access long-term data, in near-real time, on an area’s biodiversity as well as environment is "critical for ecosystem management and stewardship", Slingsby says.
A major difficulty in modelling the fynbos region, which is frequently affected by fires, is that it is difficult to tell what is natural variation compared with an aberrant problem.
"Wildfire is an important part of the ecosystem, so there is a natural cycle of vegetation loss and recovery," says Wilson.
"To monitor this ecosystem, you need to be able to identify changes that fall outside of these natural variations, and our system does this quite well."
Slingsby explains on his blog what the tool, which is still in pilot phase, seeks to do: "Detecting abnormal change in these ecosystems is highly challenging. The state of vegetation varies dramatically due to natural disturbances, long-term trends or cyclical functions, such as those relating to fire, post-fire recovery or seasonality," he writes.
Systems like Emma have been very useful for monitoring forest ecosystems, but we don’t have any way of doing this in non-forest ecosystems such as shrublands and natural grasslands, as far as I am awareGlenn Moncrieff
"We don’t want to compare the observed vegetation signal [how green it is] against the long-term average at that site; we want to compare it to the expected signal at that site for the age of the vegetation since the last fire and that time of year," Slingsby writes.
The pilot therefore uses data from Nasa’s Terra and Aqua satellites and compares that with models of what the landscape should look like, making allowances for the season and the time since the last fire — among other things.
The team is also developing a smartphone app, so that a citizen scientist, field ranger or landowner can map their landscape, take notes and mark specific locations.
But while the fynbos region is unique because of its high biodiversity, it is not the only region in the world whose vegetation is affected by periodic wildfires.
"Systems like Emma have been very useful for monitoring forest ecosystems, but we don’t have any way of doing this in non-forest ecosystems such as shrublands and natural grasslands, as far as I am aware," Ixio Analytics’s Moncrieff says.
"The tool we’ve created could have a lot of utility for other fire-prone ecosystems, such as the Mediterranean Basin, California or southwestern Australia."