Deep Learning Could Predict Deforestation Before It Happens

Each passing Earth Day brings a stronger sense of urgency to protect our planet as we race toward exhausting its finite resources, like land, food, and water. However, each year also brings new promise for solving this challenge as government, civil society, and industry make greater commitments to a sustainable future. That’s why today, Orbital Insight and the World Resources Institute are pleased to announce a new partnership to apply cutting-edge and experimental algorithms to predict, and help prevent, deforestation. The work will be part of the Global Forest Watch (GFW) initiative, a force of over 65 partners collaborating to monitor forests in near real-time.

About one-third of the world’s forests have been lost, and another 20 percent have been degraded. The loss of forests has major implications for human well-being, as these ecosystems are key to regulating the climate and providing food and water to billions around the world. Even more troubling, forests continue to be under pressure. The world is losing an area of forest equivalent to 50 soccer fields every minute of every day, adding to the already devastating toll. Our plan is to reverse this trend.

Until now, the ability to track global tree cover loss has been primarily reactive, identifying areas of global tree cover loss only after the fact. Orbital Insight will work with GFW to develop, test and deploy deep learning algorithms that analyze high resolution satellite images to detect patterns that may indicate impending deforestation. Our goal is to begin detecting deforestation on a global scale before it happens. This is a global challenge that we can only address if we apply a wide lens to make meaningful connections about our world.

Similar to the way our brains learn to recognize faces, Orbital Insight algorithms will learn to recognize patterns, such as road-building in previously undisturbed forest areas. Global Forest Watch researchers will use the findings to forecast deforestation while there is still opportunity for intervention. The potential for such a capability is tremendous. For example, GFW currently provides monthly alerts of forest change used by forest rangers and other stakeholders. However, by the time they get to the location of the alert, the trees have already been lost. This partnership could build a predictive tool to change that — reducing response times and even deploying resources to at-risk areas to prevent loss.

Our partnership also benefits companies and consumers seeking to eliminate deforestation associated with production of major agricultural commodities. Agriculture drives approximately 80 percent of deforestation worldwide, and a handful of key global commodities — palm oil, beef, soy, pulp and paper — accounts for over 70 percent of tropical deforestation. Orbital Insight’s algorithms will increase transparency around the impact of global commodity supply chains on forest loss by detecting production patterns that lead to forest loss or degradation. Our objective is to make it easier to determine whether commodities are sourced sustainably and hold companies accountable to their zero-deforestation pledges.

Thanks to unprecedented access to data and technologies, stakeholders like Global Forest Watch can make great advancements in furthering their mission, from moving toward a preventative model to creating accountability. Greater transparency around the impact of human activity on the world’s limited natural resources will enhance efforts to protect resources that are the foundation of economic opportunity and human well-being.