Sensors and Journalism

The People Within the Pixels

By Lela Prashad Only recently has remote sensing become a viable tool for journalists looking to obtain images of events as they happen on the ground in near real time. Sensors that capture images from satellites, airplanes, and drones are now commonly used to visualize the effects of people’s behavior on the ground, rendering conflicts, urbanization, deforestation, and pollution in a new light. Most remote sensing used in journalism today is visible imagery, relying on sensors that see the Earth’s surface in the same colors that our eyes do but from a higher altitude and wider field of view. A few pioneering journalists, however, are beginning to use even more sophisticated remote sensing data that measures more than what the eye can see. As journalists begin to integrate remote sensing into their reporting, they are encountering many of the same ethical considerations government and academic researchers have in the past, especially regarding security, privacy, and access to data. The United Nations Platform for Space-based Information for Disaster Management and Emergency Response (UN-SPIDER) is charged with ensuring that imagery is shared promptly after a natural disaster for humanitarian assistance through the International Charter on Space and Major Disasters; the United States Agency for International Development (USAID) Famine Early Warning System has worked tirelessly to reduce complex remote sensing into simpler tools and put imagery into the hands of decision-makers around the world. As noted above, access to data is key. In the past, privacy and security were less scrutinized since very high-resolution imagery was not easily available. However, as technology improves, we must revisit some remote sensing experts’ theories that imaging people’s behavior on the ground, even at relatively course scales, has ethical concerns. While remote sensing provides valuable data, often unobtainable from other sources, imagery on its own displays a significantly incomplete picture, especially when used to understand people and their activities on the ground. Though an image is a single snapshot taken at one point in time, it contains discernable features and objects and can appear to tell a complete story. In order to fulfill a narrative, though, we must integrate imagery with what we’ll call ground-truthing—supplementing the image with other data sources and information to link its pixels to environmental phenomena and human activity. Remotely sensed imagery documenting the environment has been collected since the 1820s when Joseph Nicéphore Niépce captured the first permanent images of nature. Today, sensors mounted on airborne vehicles can offer images that replicate the way we perceive the world, but with additional energy and color beyond what our eyes can see. Outside of our familiar redto- violet spectrum of visible light, sensor instruments render near-infrared light reflected strongly from chlorophyll in plants on the ground. Nearinfrared imagery can be used to monitor forest and crop health in ecological studies, while instruments that capture thermal infrared can also map the composition of materials on surfaces and the patterns of heat they create— including the building materials of our cities. Still other instruments use active sources, such as radar, to broadcast signals to Earth that return imagery through the clouds, trees, or ground to reveal hidden fault traces under dense canopy and archeological ruins below the Earth’s soil. Until recently, these sensors have been prohibitively expensive for people outside of government and private companies to design and launch on their own. While NASA, the National Oceanic and Atmospheric Administration (NOAA), and the U.S. Geological Survey provide most of their sensors’ data for free, this information does not usually come in formats that are easy to

understand or use without advanced technical knowledge. Other countries outside the United States do not always freely provide their data, and private companies rarely do so either. Today, low-cost, open source initiatives are aiming to put remote sensing technology within reach for anyone interested in obtaining sophisticated imagery from balloons and kites, UAVs, and small satellites. Advocacy groups, like Public Lab; do-it-yourself (DIY) science and technology hobbyists; and a few intrepid journalists have started to utilize these sensors to mine for public knowledge. While these sensors are still far from the quality and coverage of flagship science satellites, such as Landsat, ASTER, or MODIS, their potential is quickly growing. A distinct advantage of satellite imagery is that is captures a synoptic view of the Earth’s surface—without regard for political boundaries—and outputs measurements that are equivalent and comparable no matter where on Earth they are acquired. Ground-based observations are often restricted to borders and boundaries where local governments or organizations have obtained funding and permission to install monitoring equipment. Comparing observations and measurements between two locations on the ground can be difficult, even within a single region, if different methods are used. There are currently many places on Earth with very limited or no ground-based measurements. Even imagery obtained from airplanes has the limitations of ground-based sensors, as planes must fly repeatedly over the same area to build up an archive over time to track changes and do not offer global coverage since they are not in orbit around our planet. No study that utilizes imagery is complete without ground-truthed local data, and this is especially true for studies focused on people and communities. One of the most common kinds of remote sensing products that attempts to image human activity on the Earth is land-use classification. Land-use classifications utilize imagery from satellite and airborne sensors to categorize imaged pixels into different categories. A simple land-use classification might distinguish urban, rural, and water-covered areas by identifying unique signatures of each type. Classifications have many advantages— they can be automated and run across archives of imagery acquired

over decades; they usually employ the same algorithm in any image so that a classified pixel of urban scape in New Delhi is calculated in the same manner as an urban pixel in New Mexico. Much more detailed classifications can be created to study urban growth and change over time, or loss of biodiversity and urban green space. While land-use classifications are ubiquitous and widely used across physical and social science studies, there is not a single gold standard classification for all purposes. The concepts of rural and urban are defined differently by individuals and disciplines and are always in flux; remote sensing cannot address questions of people and cultural institutions. However, these classifications are not problematic when their limitations are understood. In fact, they are indispensable when comparing phenomena across many regions that fall within a wide range. Efforts to map slums with remote sensing imagery highlight the need to integrate information about the people being mapped within the pixels. Governments, the United Nations, and NGOs use this type of imagery to detect slums by automatic measure and map their extents. Slums are typically undefined, informal settlements, and mapping boundaries and population density can help researchers understand the extent of assistance people living within them may need. There are significant ethical questions about mapping people from above, however, especially when it’s without their input—even when the maps are used to advocate on their behalf. Not all of these mappers share the residents’ interests. Many slums around the world encroach on valuable property and local authorities may be interested in using these maps to identify them and remove their residents. Developing automated classifications of slums does not lead to understanding how people live within these pixels, how they make a meaningful life with severely limited resources, or who they are as individuals. Imagery, however, can provide important tools for residents of slums and their advocates to understand environmental hazards plaguing the areas, such as air and water pollution, flood and landslide risks. Mapping communities for empowerment and advocacy has been an effective way for residents of marginalized com munities to affect change. Slum/Shack Dwellers International and Map Kibera are strong example organizations that seek to put remote sensing and mapping tools into the hands of those who are often mapped remotely by others, allowing them ownership of the information they collect. It can be easy to see the ethical implications of imagery that attempts to map and classify people and their behavior. It is difficult to discern the indirect impacts other kinds of environmental remote sensing have on people’s lives. Ecologically, remote sensing has been an extremely valuable tool in measuring and tracking deforestation around the world, especially in places where people are cutting and burning forests. This imagery is effectively used to estimate the loss of biodiversity, loss of biomass that can absorb greenhouse gases, and the generation of these emissions from burning forests. It’s now easy to track where local community members living in or near forests in environmentally sensitive areas are effectively destroying trees and landscape—though only the impacts of their activities can be imaged, not the reasons. The economic, cultural, and social pressures that are contributing to deforestation are as important to understand as the extent and environmental impact of deforestation itself. It’s clearly not possible to image these human factors directly; they require a very different kind of ground-truthed information. The necessity for investigating the people behind imagery is becoming clear to those seeking to affect change. The TREE Foundation project to protect the church forests of Ethiopia, founded by ecologists Dr. Meg Lowman and Dr. Alemayehu Wassie Eshete, seeks to address these issues. Satellite remote sensing data shows that nearly the entirety of northern Ethiopia has been deforested by local people planting crops for subsistence farming. Small islands of forests have been preserved by ancient Coptic Christian churches who hold the forests around the churches to be sacred. The TREE Foundation and partner scientists have conducted studies with satellite remote sensing and measurements made in the forests in partnership with community leaders in the churches. The church communities are participating in

the research and organizing efforts to build walls to preserve the renaming forests, which serve as islands of biodiversity for plant, insect, animal, and bird species. In the United States, a crowdsourced storytelling effort called iSeeChange is bringing together citizens and scientists to document the trends people see in weather patterns and how climate change impacts their daily lives. In the fall of 2013, a NASA Databridge workshop was held in Pasadena, Calif., to discuss how to integrate NASA remote sensing data and sensor design more closely with people and their activities. The workshop was facilitated by iSeeChange and representatives of NASA’s new Orbiting Carbon Observatory satellite. This communication between scientists, engineers, and the public represents the future potential of making imagery more understandable and accessible to everyone. The barrier to entry is rapidly lowering for those who want to design and launch sensors on airplanes, drones, and satellites and those who seek sophisticated types of imagery that extend beyond what the human eye sees. Journalists have a powerful new potential to integrate remote sensing imagery into their reporting and connect to a near-real feed of data for environmental issues, conflicts, and people’s activity on the ground. While imagery may appear to represent a full story, an image, no matter how sophisticated, is only one data point taken at a snapshot in time. Remote sensing imagery must be integrated with other ground-truthed data, local knowledge, and a firm understanding of people and communities to pull together a full story. Journalists, researchers, and others seeking to derive knowledge and understanding from sensors should see imagery as both a powerful tool and limited facet of a human story taking place within the pixels.