Issue 6, October 3, 2018
We live on a planet that sustains an enormous multitude of life forms. It is a fragile planet, where even small changes in our aggregate behaviour can have an outsize impact on the critical flows and processes that allow us to survive.
Despite appearing to be the most intelligent life forms that we know about in the visible portion of our universe, we know very little about how our behaviour contributes to changes, and the extent to which these changes tend to revert to the mean over time, or spiral out of control.
There are some things that we do know. We know that deforestation destroys habitats, can eliminate a wide variety of species, and can take decades to even partially restore. We know that access to fresh water and power generation are likely to be significant drivers of wealth and political tensions in the years ahead. We also know that human land use choices, such as irrigating deserts, creating large expanses of concrete land cover, or clearing mature forest for grazing cattle can significantly impact carbon sinks, long-run land productivity, and regional micro-climates. We have an inkling for the big things, but we lack the data, measurement capacity, and modelling to track impact through time, and to inform convincing evidence-based policy decisions.
That’s not to say that there are not many amazing organisations trying to do better. In fact, the AI for Good Foundation is proud to partner with some of these incredible teams to identify how AI technologies can support our joint goals of a happier healthier planet.
Explore our site to find out more about what we are doing in this space. Join our mailing lists, and donate to help us provide more technology support to solve these problems on the ground.
Yours, from the field,
Monthly Interview: Kathleen Buckingham
Kathleen is a Research Manager for the Global Restoration Initiative in the Forests Program at the World Resources Institute. Her research focuses on developing diagnostic tools to assist stakeholders to plan and implement successful forest and landscape restoration.
Could you tell us a little bit about your background and interest in the use of data analytics in deforestation, and the future of sustainable forest restoration more broadly?
I work as a Research Manger on the Global Restoration Initiative at the World Resources Institute based in Washington DC. WRI is a global research organization that spans more than 50 countries, with offices in the United States, China, India, Brazil, Indonesia and more. Our challenge is to turn big ideas into action to sustain our natural resources—the foundation of economic opportunity and human well-being. Our work focuses on six critical issues at the intersection of environment and development: climate, energy, food, forests, water, and cities and transport.
WRI focuses on scaling change. Our Forest team, notably Global Forest Watch, have used pioneering approaches to monitoring forest change. GFW is an open-source web application to monitor global forests in near real-time. GFW is an initiative of the World Resources Institute, with partners including Google, USAID, the University of Maryland, Esri, and many other academic, non-profit, public, and private organizations. Forest and Landscape Restoration, on the other hand, has traditionally not been engaged in such a technological way. Our team has been focused on supporting governments through the Restoration Opportunities Assessment Methodology (ROAM) or other tools to help assess, inspire, enable and mobilize action to restore vitality to degraded landscapes and forests around the globe. WRI is partnering with governments, businesses, and communities around the world to restore 500 million hectares of deforested and degraded land, an area half the size of China.
As a Research Manager, my work focuses on searching for solutions that can be used to scale. I was inspired by work we conducted on social network analysis regarding ways of accessing and using data. On the whole, the environmental field tends to be behind the curve regarding embracing technology. The research is constrained by understanding the spectrum of what is possible.
What do you think are the biggest challenges that lie ahead?
We have many challenges. A key one that springs to mind is we need to build upon success. To do this, we need to know what has succeeded and what has failed for restoration and why. This involves aggregating data and case studies. Aggregating data on restoration is difficult – restoration is a process not an end goal. Restoration therefore takes many different forms and uses different interventions to increase ecological integrity. There is so much progress being made. We need to be able to identify land-use change and build upon successes, and failures at scale. Currently the fragmentation of knowledge is a boundary to progress. We could benefit from new technology and techniques to scrape and mine data. We don’t want to create a one-size-fits all framework but need to enable change at scale. To do that, we need to aggregate data differently.
How do you think that technology, and Artificial Intelligence, in particular, might facilitate solutions to these challenges?
Forest and Landscape Restoration can help mitigate climate change, increase food security and have positive impact on water resources and energy. In short ecosystem services and livelihood benefits. By 2050 we are facing a global population of more than 9.8 billion. Forest and landscape restoration presents a pragmatic approach to natural resource management through recognizing a matrix of different land-use options through the landscape approach. Restoration is not about fortress conservation, conservation is part of the picture, but we need to restore ecological functionality – be that to agricultural land, plantation forestry, woodlots or biodiverse rich forests.
I see technology and Artificial intelligence playing a role in two key ways – using computer vision and natural language processing. Restoration is gaining traction globally. The UN have set a target to have 350 million hectares of land under restoration by 2030. There are many mechanisms through which countries and organizations have pledged to be part of the movement, through the Bonn Challenge, Afr100 and 20×20 Initiative. We need to go beyond pledges to think about action on the ground, through aggregated data and monitoring change. The challenge is huge. To see the change, we need we need to use different tools. We need to change business as usual, if we are truly going to get to scale.
Currently in forestry, we use satellite imagery to identify land use change. This can be quite straight forward when considering tree canopy cover and loss. However, restoration is about regrowth and trees outside the forest. Change at a land-scape scale can be difficult to monitor. There are tools to monitor trees outside the forest, such as Collect Earth. How could we utilize new technologies, machine learning and computer vision to automate the process to understand land use change at a finer scale than canopy cover and loss? Environmentalism is often a negative discipline, but it doesn’t have to be. What if we could capture positive change, using new technologies?
Furthermore, natural language processing potentially has an important role to play. Restoration encompasses so many different ministries from agriculture to forestry to mining. For those involved in restoring landscapes, it can be difficult to understand laws and regulations, to understand whether there are policy incentives or conflicting policies. Natural language processing has a huge role to play in breaking down boundaries with qualitative data, to increase access to knowledge. Hopefully when regulation becomes more transparent, this could lead to more incentives and less conflicting policy.
What are the key organizations that will enable achieving these goals?
Fundamentally we need to partner differently. One sector cannot address these challenges at scale. Environmentalists need to collaborate with data scientists. However, we need to recognize the importance of knowledge at different scales. At WRI, we rely on in-country teams to communicate the challenges and questions that need addressing at the country and landscape level. I read an interesting book recently called AIQ: How People and Machines are Smarter Together by James Scott and Nick Polson. The key take-away is that we need to find a balance between AI and IQ. With restoration, as long as we keep our challenges and questions rooted in the issues generated at the country and landscape level, technology has a huge role to play in scaling solutions.
I was inspired by the session on “Fragile Earth” at KDD in London – It seems that there are many data scientists looking to answer real world problems. The solution is to connect data scientists to organizations such as WRI. We are aware of the challenges, have access to ground truthing and data, but currently lack the capacity to go beyond pilots. In the future, we could build networks and create solutions together through a range of different skills sets and capacities. WRI recently was engaged in a Hackathon, such events and partnerships could lead to systematic change.