Documenting mangrove forests in the Galápagos Islands, with help from Google Earth
Editor’s note: Today’s post is by Nicolas Moity, a marine sciences and GIS researcher for the Charles Darwin Foundation, the nonprofit dedicated to research in the Galápagos Islands. Nicolas and his Darwin Foundation research colleagues recently completed a research project mapping the distribution of mangrove forests on the islands using Google Earth.
The coastal mangrove trees and shrubs in tropical environments are vital habitats for sea lions, birds and fish. The mangroves also act as carbon sinks, storing important quantities of carbon, which helps fight climate change, and they protect coastlines from erosion. But in many parts of the world, mangroves are disappearing at alarming rates, due to forest clearing, pollution and tourism. The Galápagos Islands, where I work, are an excellent place to study mangroves because the coastal ecosystems are virtually free of human impacts.
Using Google Earth, we were not only able to map the mangrove forests along the Galápagos’ various coastlines, but also compare their growth between 2004 and 2014. Happily, in the absence of threats like deforestation and pollution, the mangroves can thrive on their own: We found that the mangrove forests increased by 24 percent over the 10-year period.
Alternatives to mapping mangroves by air or on the ground
As a biologist, working in the Galápagos for the Charles Darwin Foundation is a dream come true. In a small place rich with biodiversity and isolated from the effects of humans, we can quickly see the results of our research. As part of recent protections for the Galápagos Marine Reserve around the archipelago, my research colleagues and I were asked to provide precise and updated data about mangrove distribution across all of the islands — something that had never been done before with such precision.
The question was how to do this research accurately and cost-effectively. The Galápagos archipelago has 15 main islands and a long coastline to study. We could have used airplanes to study the mangroves by air, but the Charles Darwin Foundation is a nonprofit and we want to be careful stewards of its money. Drones would have been less expensive, but their battery life is only about 20 minutes at a time, and they’d take far too long to cover the Galápagos coastline. We considered using Landsat and Sentinel satellite imagery, but their low resolution would make it difficult to spot the mangroves, which are often less than a quarter of a hectare in the Galápagos.
Then we thought, why not use Google Earth’s high-resolution images? Since the Galápagos islands are mostly bare lava, it was relatively easy to see the coastal mangroves on our computer screens. We just needed a way to map and measure them.
To give ourselves a head start, we collected data from prior mangrove studies using aerial photographs and Landsat images, as well as “ground truth” data that we collected in the field recording GPS points for mangroves along coastlines. We converted this data into map layers using Google Earth’s native format, KML (Keyhole Markup Language). This way, we already knew where some of the mangroves were located.
Calculating mangrove distribution with polygons
By zooming into 300 meters eye altitude on the Google Earth images, the greens and coarse textures of the mangroves stand out fairly easily from the smooth turquoise of the water and the grays and browns of bare lava. We carefully drew polygons along the edges of the mangroves and the coastline; if we weren’t sure we were looking at mangroves, we zoomed in closer at the 150-meter level to confirm our analysis.
It took about two months to create polygons for all of the Galápagos coastline mangroves. By converting the polygons into mapping tiles and adding them to ArcGIS, our mapping and analysis software, we calculated that 35 percent of the Galápagos coastline is covered by mangroves, or about 3,657 hectares in total.
As scientists, we like to be sure that our research methods are accurate and reproducible — meaning that other researchers could adopt our approach to mangrove research in other parts of the world. Using the same Google Earth images we used for our research along with ground-truth data about the Galápagos mangroves, we applied the maximum likelihood classification (MLC) algorithm, which is a semi-automatic technique. With a few sample polygons of known mangroves, lava, and the sea, the algorithm classifies each section of an image.
We also compared our method with an object-based image analysis approach. Image objects are groups of pixels that are similar to one another based on their spectral properties, but also their size, shape, and texture, as well as context from a neighborhood surrounding the pixels. This type of classification attempts to mimic the type of analysis done by humans during visual interpretation.
In both cases, our method — which relied on humans to decide whether part of an image that looks like a mangrove really is a mangrove — was far more accurate than the other methods. Automatic classification does save time, but we demonstrated that with the combination of Google Earth images plus the manual work of drawing polygons, we get significantly better results.
The next question: Are mangroves growing or shrinking?
When scientists finish a research project, the results often inspire us to keep asking questions. Once we had measurements for Galápagos mangroves, we realized we had access to earlier Google Earth images that could help us understand whether the mangroves were growing or shrinking.
To do so, we had to redraw all the polygons using satellite images from 2004. Thanks to Google Earth’s historical imagery feature, we could do that for some areas of the Archipelago. We didn’t redraw all of the areas of the Archipelago due to time constraints, but we used a significant and representative coastline sample to infer the 10-year change.
We chose a young island, a middle-aged island, and an old island to analyze if there were differences in mangrove change depending on the islands’ age. Once we drew polygons for both 2004 and 2014, we could apply all sort of analysis to explore how the change had occurred, such as, by expansion of old patches or by creation of new ones. In some cases, the newer Google Earth images had polygons that did not appear on the older images — signs of new mangrove clusters. That’s how we determined that from 2004 to 2014, the mangrove forests increased in size by 24 percent.
This is a positive sign for people who are working to save mangroves around the world. In the Galápagos, the growth of the mangroves is 100 percent natural — without interference from humans, the trees are thriving. In places like Myanmar where mangroves are under stress, reforestation programs can counter the effects of forest clearing and development. We’re eager to share our methods of using Google Earth imagery for studying mangroves, especially for countries with arid or semi-arid environments that have limited funding for research.
Read more about the Charles Darwin Foundation’s mangrove research in the PLOS ONE article.
