The United Nations Global Pulse, an innovation initiative on big data and data science, and Western Digital recently announced the winners of the Data for Climate Action Challenge (D4CA) at the Data Innovation: Generating Climate Solutions event during the United Nations climate change conference (COP23) in Bonn, Germany.
An unprecedented open innovation challenge to harness data science and big data from the private sector to fight climate change, D4CA was launched in early 2017 and called on innovators, scientists, and climate experts to use data to accelerate climate solutions. Access to large amounts of data – anonymized and aggregated to protect privacy – accelerates the ability to spot connections, gain insight and develop predictive algorithms that can provide more precise direction and decisions. The Data for Climate Action Challenge demonstrates what’s possible when public and private sector organizations partner for social good.
Bouncing Back from Fires with Data
WIRED recently sat down with the D4CA Thematic winning team from the University of Cape Town and the University of Buffalo. The team’s winning project, titled “EMMA: Ecosystem Monitoring for Management Application”, used satellite data to develop a platform for near real-time prediction and monitoring of the impact of forest fires in South Africa’s Cape Floristic Region.
Interviewees: Jasper Slingsby (South African Environmental Observation Network and SEEC, Biological Sciences, University of Cape Town) and Adam Wilson (University of Buffalo)
WIRED Brand Lab: Can you give us an overview of your submission?
Jasper: We were looking to come up with a change-detection algorithm that can help track changes in shrub-based ecosystems, in particular the Cape Floristic Region (CFR), a Global Biodiversity Hotspot and UNESCO World Heritage Site in South Africa.
At present there are very good change detection systems for forest ecosystems. But for ecosystems that show strong seasonal variation and temporal oddities like fire and post-fire recovery, there really isn’t anything out there that can do that.
WBL: Why is it so difficult to track changes in these particular environments?
Adam: When looking at satellite imagery of forests over time, for example in mid-summer, you wouldn’t see much variability, and changes are a lot slower. But for many of the world’s ecosystems that are not forest, the change can be much faster. That makes it more challenging to use satellite data to monitor and identify unexpected changes. It’s even more of a challenge to understand whether change is within the normal behaviour of an ecosystem, or if it’s above or below average.
In this particular ecosystem in South Africa that we studied, there’s a big difference between vegetation cover a year after a fire and five years after a fire. Inside that, there’s variation in whether the area is recovering as expected or not.
WBL: Are the fires in this region natural or man-made?
Jasper: It’s typically a mix of the two, but fundamentally, fire is a natural part of this system. If you don’t burn off vegetation, over time things will start to die off. It needs fire to keep it healthy.
WBL: Why is this region in particular so important?
Adam: The total area of this region is around 90,000 square kilometres. That’s one percent of Africa’s land area, but it contains 20 percent of Africa’s plants – or three percent of the world’s plants. Two-thirds of the 9,000 species we’ve found there are endemic to that region and not found anywhere else in the world.
WBL: Are there other parts of the world that have similar, shrub based ecosystems?
Jasper: Essentially, Australia, California, in the US, and the Mediterranean region would all have similar ecosystems to this. The model we’ve built can be applied to different ecosystems across the world.
WBL: How is monitoring currently conducted in the region?
Jasper: In the nature reserves and the parks, it’s people walking around on foot. In the privately held parts of the region, it just isn’t monitored at all. In South Africa, we have really strong legislation around environmental protection, but our monitoring is really bad, so enforcement is bad. This could help change that.
WBL: What datasets did you use?
Adam: We used Modus, which is a NASA satellite imagery product. We then combined that with climate data to establish which areas were behaving strangely. Once we identified those, we used high-resolution imagery from Planet Labs to help diagnose those areas.
WBL: What did you see happening to the region?
Jasper: In one instance, we saw a site that was burnt in 2015 at the beginning of a drought we’re still in – which has been the worst drought we’ve ever seen on record. This particular region just hasn’t recovered like we thought it should, based on previous records.
So we’re already seeing slower post-fire recovery than the records would tell us. We’re expecting the climate to get drier and hotter, so we’ll expect recovery to be slower and slower.
Looking at the implications of that, things like carbon storage capability and the hydrology of an area will all be changed, as well as the effects and the recurrence of fire. In the short term, hotter and drier means more fire, but in the long term it means less fire, because the biomass accumulation, or the fuel for the fire, won’t be accruing as fast.
WBL: What are the long-term impacts of that change?
Jasper: It’s more likely we’ll shift to a system favouring invasive species that are from hotter and drier climates, like those found in Australia. The fires we have here are puny in comparison to the fires in California and Australia. If we continue to have a lot of species invasion from Australian plant species, we’re going to get fires on the same scale we see in these places. People here aren’t geared to deal with that.
WBL: How successful was the algorithm you built at showing whether regions were under or over performing in terms of re-growth?
Adam: It works really well. The model predicts the vegetation dynamics as a function of climate and other environmental data. So it’s not simply about fitting a curve through observed points. It’s capable of predicting what the vegetation would look like under certain conditions.
When you’re looking at a dynamic ecosystem, it’s hard to know if what we’re seeing is normal or expected. What the model does, is it gives an expectation for any given location, then you can compare expectation with what the ecosystem would be doing, and flag it if something strange is going on.
WBL: How much of the area you studied was performing irregularly?
Jasper: Five to ten percent of sites were highlighted showing either drought, clearing or invasive species arriving. This is really important for land managers to have access to this kind of information.
WBL: What happens next?
Jasper: The next major step is a smartphone app, which allows people to log things using a camera phone while out in the field. We want to make it open to citizen scientists so they can go out and explore areas that have changed a lot. Beyond that we want to run the model across the whole Cape region to give authorities a better way of monitoring changes.
This content is produced by WIRED Brand Lab in collaboration with Western Digital Corporation
Data Makes Possible will be following the winners as they work to implement their solutions and bring real change to our world, and we’ll be publishing interviews with the thematic and data visualization winners throughout January and February.