Mapping the Galápagos Islands for a Better Understanding of Human Impacts on the Environment

Last summer, a team of researchers from the University of Chicago embarked on an ongoing project to map the Galápagos Islands using Mapillary. Annually mapping the islands with street-level imagery contributes to a better understanding of how people affect the natural environment and our very own Chris Beddow had the opportunity to assist the team this summer with their important work.

Billboards, banners, and advertisements in the Quito airport reference the Galápagos Islands as a natural paradise, with the words stamped over images of sea turtles and iguanas on pristine beaches flanked by blue ocean and green forest. My expectations en route were defined by these messages as well as the islands’ global reputation as a model of ecological purity and unique biodiversity. I didn’t know quite how to reconcile these ideas with the reason I was traveling to the Galápagos Islands: to tag along with a project studying urbanization.

The Galápagos Urbanization & Sustainable Development study started in summer 2018. Led jointly by the University of Chicago’s Mansueto Institute for Urban Innovation and the Program on the Global Environment (PGE), my main point of contact was Mansueto Postdoctoral Fellow Daniel Zünd. In the previous year, Daniel spearheaded this program with a student researcher by mapping the island of San Cristobal. He and others at the Mansueto Institute formed a partnership between University of Chicago, Universidad San Francisco de Quito (USFQ), and the Melbourne School of Design (MSD) centered around urban development research in the Galápagos, particular Isla Santa Cruz and Isla San Cristobal.

Isla Santa Cruz and Isla San Cristobal Isla Santa Cruz and Isla San Cristobal

It was early July, and I was hopping my way closer to one of these urban Galápagos hubs called Puerto Ayora ,where I was to meet Daniel’s team of students from the University of Chicago. Leaving the Quito airport, I was charged a fee and issued a document certifying inspection of my luggage, ensuring I didn’t carry any exotic plant or animal species. As the airplane descended toward the airport on Baltra Island—former site of a US Navy facility during the second World War—I was handed a form with a customs questionnaire, asking to specify if I was carrying any agricultural products, or even camping gear and hiking boots. By the time I landed, signs warned against carrying non-recyclable plastic bottles beyond the customs boundary, and I was convinced that I must be about to enter the epitome of an immaculate, natural environment.

I was delighted to meet the four talented students contributing to this year’s bicycle mapping efforts. Each of the students was researching a specific question related to the urban growth and change on the islands. The student projects highlighted both the problems faced by the urbanizing Galápagos, and explored possible solutions.

Students from Mansueto From left to right: Daniel, Sam, Ryan, Amy, and Jein. Photo courtesy of Mansueto Institute for Urban Innovation.

Sam’s project looked at the infrastructure at risk from sea level rise; Amy estimated the necessary surface area of solar panels needed to supply the energy consumed; Ryan identified urban footprints and compared them against other island regions in the world; Jein examined the relationship between urban infrastructure and economic activity. All of these topics reveal a part of the island that tourists, with an appetite for ecological wonder, are both overlooking and instigating—the pressure to expand the urban areas to their limits by calling for more hotels, more restaurants, more tourism employees, more housing for those employees, and all the material and economic accessories to this.

Daniel and the four students had arrived a week prior to me on Isla San Cristobal, and had completed mapping there before arriving to Isla Santa Cruz. They had mostly mapped on bicycles, but also attached a camera to a taxi to catch a few stretches of rural road. Because this area had been mapped in 2018, the new imagery offers a chance to enhance the data extractions by providing more dense coverage, and to compare changes over time.

Ryan attaches a camera to a taxi truck. (photo by Amy Tian) Ryan attaches a camera to a taxi truck. (photo by Amy Tian)

During my first day on the island, I would witness many scenes typical of islands and coasts ranging from Panama to Zanzibar, Sicily to Sri Lanka: plastics, paper, building materials, and other products of human habitation were ubiquitous along the roads, while cattle grazed in clear cut pastures surrounded by barbed wire, chickens waddled through city streets dodging scooters and pickup trucks, and debris made of painted wood or styrofoam congregated around docks and piers, in between boats and buoys. On my last day, I also tramped my way to the set of antenna towers on top of Cerro Crocker, the high point of Isla Santa Cruz. Emerging through the fog after a 2 hour walk to the peak, the largest antenna juts up into the sky like a flagpole planted on the summit, clearly declaring that this is a human island.

A view over the top of Cerro Crocker and Isla Santa Cruz A view over the top of Cerro Crocker and Isla Santa Cruz

Data illustrates this situation more objectively, but the data is also incomplete. Previous research illustrated some of the major factors at play: new building construction, scarcity of fresh water, increased energy consumption (mostly burning diesel fuel), population growth, growing volumes of tourist traffic, and complications around how to regulate migration from mainland Ecuador. Daniel’s 2018 project made a snapshot of the urban change with 360 degree image capture using a GoPro Fusion and a bicycle on Isla San Cristobal, which this year was refreshed on San Cristobal and replicated on Isla Santa Cruz.

The imagery alone gives a clear view into the amount of concrete, iron, wiring, and other developments that come along with urbanization. Building footprints visible from satellite show the spatial extent of human settlement, but the new street-level imagery reveals that while the development reaches outward limits, it also is growing upward. While Mapillary computer vision detects and produces point data for such urban infrastructure as crosswalks, sidewalks, power lines, and trash bins, the images also document something more subtle: rebars, rods made of reinforcing steel, were indicators of ongoing upward construction that tells a deeper story about the present and future state of the islands.

Bicycle mapping is a precarious operation on its own, but a familiar situation to Mapillary users. The camera needs to be well positioned—in this case attached to a bicycle helmet, the dual lenses facing forward and backward at 0 and 180 degrees. The battery needs to be charged, and the rider then can use the handful of hours that the battery lasts in order to ride through the streets of the small but growing Galápagos towns. Because the team had two cameras and four mappers, rather than one of each like in 2018, the streets to be captured could be divided into smaller tasks. Navigation can be a bit difficult while mapping, so it helps to ride in pairs with one person focusing on the camera and the other focusing on a mobile navigation application as well as the upcoming turns in the road.

Jein Park and Chris Beddow mapping in Bellavista. (photo by Amy Tian) Jein Park and Chris Beddow mapping in Bellavista. (photo by Amy Tian)

On my first morning on Isla Santa Cruz, the team had already finished mapping the larger town of Puerto Ayora. It was now time to move inland, just a few minutes up the main highway to a small town called Bellavista. We gathered at a main square, adjacent to the church and the elementary school, we we established as our meeting point for the day. Jein and Sam took responsibility for the two bicycle helmets, including the GoPro Fusion mounted on top of each.

Sam had volunteered to continue up the road to an even smaller town, Santa Rosa, before riding downhill on the long stretch of protected bicycle lane connecting it with Bellavista. Jein would map everything to the south and west of the L-shaped main highway, and I volunteered to ride beside her with my mobile phone, tracking out progress and planning a route on the fly with an offline OpenStreetMap basemap for reference.

Imagery from Bellavista

The planning was mostly simple due to the small size of the towns and their neighborhoods, but even in big city mapping, all plans start with small areas such as this. The main idea was to covered the assigned quadrant in as few turns as possible, avoiding covering the same road twice and planning for efficiency. Often this meant a figure 8 shaped route, but it also meant being cautious. At some points, we had to emerge onto the main highway, and look both ways before riding down it and looping back to another street. In other cases, the residential roads were either unpaved, formerly paved and fading, or otherwise eroded with potholes. Finally, the local dogs were not entirely friendly, and liked to bark and give chase as we biked through.

Reviewing the image capture session with Python scripts. (Photo by Amy Tian) Reviewing the image capture session with Python scripts. (Photo by Amy Tian)

With the first section of Bellavista finished, we gathered in the shade to check the data. This task is something typically solved in other large scale projects using the Mapillary for Drivers app, which helps someone mapping in the field to track progress, while a supervisor can split mapping into grid-shaped tasks and assign it to field mappers. In Bellavista, we had a few issues that made Daniel’s Python scripts fit the task better: lack of internet and mobile connectivity, and need to check not just our own GPS tracks but making sure the images had valid GPS.

Daniel’s script would move the images from the GoPro Fusion camera to a folder on his laptop, then grab the GPS location of each and plot them against an OpenStreetMap background. Often we found that the GPS or images had encountered some errors, sometimes due to a memory card problem, other times maybe due to signal interference. We’d find a couple gaps that needed filling, and do another 15 minutes of riding with the camera to clean things up, then check again. This way all mapping was finished on site, with no surprises later. Considering the maximum upload speed of something like 200kbps on the Galápagos, upload of the images would have to wait.

Mapping Santa Rosa meant just a few loops and one long ride back to Bellavista.

Meanwhile, Sam was on quite the adventure in Santa Rosa. He encountered a few areas where the map had a road but nothing existed on the map, and otherwise got to explore a more rural farming community of the island, a prime example of human impact. After finishing up, he had to cover a long distance over about 30 minutes to get back to Bellavista. While the map shows just a highway here, there’s also a separate bike path that glides down and over a few rolling hills back to the main square of Bellavista.

Amy, Ryan, Jein, and Sam hard at work in Puerto Ayora. (photo by Daniel Zünd) Amy, Ryan, Jein, and Sam hard at work in Puerto Ayora. (photo by Daniel Zünd)

Once we had regrouped, everybody met at our favorite deli in Puerto Ayora, which had slightly faster internet than usually and a great menu—perfect for doing some remote mapping. Armed with a hard drive full of data and fresh installs of QGIS, we all set to work analyzing some of the existing data and thinking about what to do with the new data we’d anticipating processing and downloading the following week.

The students used data from multiple sources, including the local and national government, previous research teams, OpenStreetMap, and last year’s extractions from Mapillary. We found that it was useful to use Mapillary imagery for expanding the datasets in the future, and I soon was adding new businesses and points of interest to the map based on 2018 imagery, the type of data that Jein was using to analyze the spatial distribution of economic activity. Ryan also generalized the Mapillary segmentation values—things like roads, vegetation, wire groups, and water—into categories of urban, natural, and other. This would help produce a heatmap of the urban presence, which can be compared across years. This also was produced for Torshavn in the Faroe Islands and Avalon on Santa Catalina Island, as a comparative study of island towns.

Ryan’s analysis of urbanization levels in Puerto Ayora Ryan’s analysis of urbanization levels in Puerto Ayora

On my last morning on Isla Santa Cruz, Daniel loaned me one of the GoPro Fusion cameras. I was headed back to the airport at Baltra, then to Quito, while the rest of the team would take a boat to Isla San Cristobal before also returning to Quito the following day. I had two tasks ahead of me: map the highway to the airport in 360 degrees, and get to Quito to start uploading the complete set of imagery that Daniel provided to me on an external hard drive.

Mapping the highway was swift and easy, as I replicated Ryan’s previous method of using zip ties to stabilize the selfie stick against the truck’s roll bar, and then mounted the camera on top. Normally, I would attach an extra extender to get the camera above the truck’s roof for better perspective, but working with limited resources things turned out well.

A stretch of highway alongside agricultural areas and powerlines

Another boat, a bus, and a flight later I arrived in Quito, and started uploading. The process would take quite some time, despite excellent internet connection—the images needed to be transferred first to the cloud, then I would have them stitched using a script which can take 5 seconds per image, then the images would be uploaded to Mapillary and processed. The quickest part is the uploading and transfer, but the stitching is a special aspect of the GoPro Fusion that needs to be considered in all plans. We learned immediately that we would have been better off stitching the imagery offline while still on the islands, so it was ready to upload directly to Mapillary upon arrival in Quito. Within a few days, everything started to populate online, including the imagery for viewing and the extracted data for download.

Shop signs, advertisements, and poles from this image were extracted as point data Shop signs, advertisements, and poles from this image were extracted as point data

The entire project is still in its early stages, as this data will come to be more and more revealing as it ages and is compared to new data each year. During my short week assisting with this project, I found it to be an eye-opening experience of the Galápagos Islands. Studying the human aspect of these islands is crucial, and gave me a perspective that was very parallel yet strongly contrasting with the tourist experience that is focused on exploring the natural aspect of the islands. In the 21st Century, the Galápagos Islands are very much part of a larger pattern of globalization, a microcosm of the worldwide pattern of humans reshaping the environment and carelessly interfering in fragile ecosystems. The struggle between urbanization and wilderness, economic development and resource management, environmental policies and social policies—these are the reality on the Galápagos and merit continued studies.

Next year, the Mansueto Institute will expand its mission, using 360 degree cameras and Mapillary to visualize the differing levels of development on other islands, as well as further investigating the change in urban areas that were mapped already in 2018 and 2019. Daniel and the staff from University of Chicago and the Mansueto Institute are planning to grow their partnerships in the region, while continuing to have the university financially sponsor the students who participate in the ongoing research. You can follow more of the Mansueto Institute’s work on Twitter, and don’t miss Amy Tian and Jein Park’s diaries about their own experience during this project. If you’re visiting the Galápagos Islands, you can also help contribute by mapping with your mobile phone or an action camera, while you can even get involved from home by editing OpenStreetMap using the existing imagery or by helping to trace buildings and roads.

/Chris, Solutions Engineer

Continue the conversation