My research group studies mangrove forests, which are vital coastal ecosystems that buffer infrastructure during extreme weather and support local fisheries. When I first began my internship at NASA in 2016, I had never heard of mangroves or learned about the scope of global forest losses, but I began reading news articles about a series of widespread mangrove losses occurring in the Gulf of Carpentaria in Australia. Thousands of hectares of forests died that year, and scientists didn’t gain a complete understanding of what caused the devastation until much later. I decided to build a program that could use satellite imagery to monitor the location and drivers of mangrove loss, potentially helping to prevent another large-scale dieback in the future.
Google Earth Engine provided me with the scope of datasets and computing power necessary to analyze forest change on a global scale. I began my project at NASA with no knowledge of satellites or image processing, but guidance from my mentors, Dr. David Lagomasino and Dr. Lola Fatoyinbo, and my intensive studying of the Earth Engine developer resources helped me move from endless notes and plans to actual working code.
In mapping past global mangrove losses and drivers, we used long-term Landsat satellite imagery to identify regions of disturbance. Machine learning algorithms helped to identify where mangroves were converted to urban regions, agriculture, aquaculture or mudflats. Using the Earth Engine Apps interface, we’re working towards making our data both openly accessible and widely understandable for users of any background. Communicating our results at a comprehensible level is arguably as important as the science itself, as the ultimate goal of the project is to deliver our data to mangrove-reliant communities on the ground.