Earth Day: Agile Science in the Anthropocene

On the first Earth Day in 1970—with boot prints on the moon fresh in our memory—humanity was buzzing with a desire to make a conscious, global effort to be better stewards of our planet, and endowed with the confidence that we can do great things. It was not an accident that, just two years later, the first Landsat satellite launched, and the era of global remote sensing began. Today, on Earth Day’s 50th anniversary, Landsat, Sentinel, Planet and scores of other government- or commercially-operated satellites collect a staggering amount of data about the surface of our planet.

The story it tells, is grave. Earth has entered the “Anthropocene,” a new geological epoch defined by the enormous reach of humanity over the biosphere. Humans now directly occupy or control the fundamental ecological processes of roughly 50 percent of the Earth’s land surface, and have a pervasive and poorly quantified influence on the oceans. Humanity has altered nearly all of our planet’s biogeochemical cycles, even in the most remote ecosystems. Most prominently, atmospheric carbon dioxide concentrations likely exceed levels experienced throughout the Pleistocene, beginning 2.5 million years ago. And now, we are all grappling with the unique global COVID-19 crisis.

The value of data-driven decision making has never been higher. Data must be analyzed, interpreted and synthesized to foster decision making. Where does this process begin? If you travel upstream from a database or a decision support tool—to the point where the pixels are first touched by a human mind—there is often a scientist, asking questions. Being curious. Designing a new sensor or writing a new piece of software.

To celebrate Earth Day, and the third anniversary of Planet’s  Education and Research Program, I wanted to share a few of the requests our team gets, each day, from university students and researchers across the world.

Rosa Aguilar, MSc. and Professor Monika Kuffer, from the University of Twente in the Netherlands, wrote, in 2018: “We have several ongoing research projects where we would like to use Planet data: mapping of urban structure types and their relation to urban heat islands, mapping of urban growth, understanding of spatial factors driving this development, mapping urban poverty of African cities, comparing machine learning and deep learning methods, and generating in class exercises based on outcomes of these research projects with international Masters students.”

This April, Aguilar and Kuffer published a portion of this work in the journal Remote Sensing, leveraging Planet data to better measure progress toward the U.N.’s Sustainable Development Goal 11: Sustainable Cities and Communities. Aguilar is now working toward a Ph.D.

From the Education and Research Program to Planet’s partnerships with NASA and the European Space Agency, the scientific community is developing a manifold understanding of the coupled human-Earth system. In her master’s thesis at the Naval Postgraduate School, Katherine Rice used transfer learning—a neural network approach—to first detect, and then classify different types of maritime vessels. Tongshu Zheng, a Ph.D. student at Duke University, recently developed a novel machine learning algorithm to analyze particulate aerosols—a type of air pollution—over cities in China. Suzanne Palmer, a Lecturer at the University of the West Indies, recently asked: “I’d like to assess the effects of the installation of artificial structures on shallow coral reef habitats—what is their impact on the surrounding environment over time?”

Dave Knapp, lead scientific software engineer at the Center for Global Diversity and Conservation Science at Arizona State University, leads a student workshop for analysis of Planet imagery. Photo by author.

Questions drive scientific research. Ultimately, our common effort to synthesize findings and integrate techniques and results, leads to smarter decisions and improved human welfare and ecosystem management. As the Education and Research Program grows, we’re focused on enabling all students and faculty with the option to apply for access, and partnering with universities across the world with site licenses that incorporate Planet’s data and tools into their own learning environments and research programs. We’ve been amazed at the diversity of questions, and awed by the power of agile science to meet the moment of the Anthropocene.

These advances also help Planet improve our own systems. As the program was just getting started, Professor Matt McCabe, of King Abdullah University of Science and Technology in Saudi Arabia, wrote: “I am interested in exploring a number of applications of Planet data to Earth observation problems, specifically focusing on agricultural issues of both local (Saudi Arabia) and global interest. One of these will look at the capacity to monitor changes in cropping patterns…. Ultimately, the high spatial and temporal resolution of the Planet data will have a major impact on our research applications in agricultural water use and crop health estimation.”

McCabe’s group, including Rasmus Houborg and Bruno Aragon, ultimately developed novel sensor fusion techniques to generate a high-resolution understanding of crop cycles, biomass production and water-use. Houborg has since joined the Planet imaging team, working to incorporate some of these techniques into our own systems.

When humans first traveled to space in the 1960s, the view of our blue planet, hanging in the blackness, reminded us that we are part of a common global ecosystem.

Since the Education and Research program began, we have had approximately 8,000 users, have been cited in over 400 publications, and have developed over 30 partnerships with schools. In the past month, the program has responded to a number of requests from Earth science researchers that are unable to access their field sites, or are investigating the effects of COVID-19 on the Earth system.

Our team looks forward to learning where these questions lead and working together to become better stewards of Earth.