Profiles from the Peninsula is a series dedicated to spotlighting the partners who make up the Baja Working Group, and their projects. This week’s profile is on Astrid Hsu, research associate at the Aburto Lab at Scripps Institution of Oceanography, and her work leveraging drone technology to improve ecosystem monitoring efforts.
Puedes leer este blog en español aquí.
Profiles from the Peninsula is a series dedicated to spotlighting the partners who make up the Baja Working Group, and their projects. Each week, we will bring you a new profile in the form of a blog like this one. More information about the working group can be found here.
Astrid Hsu is a research associate with the Aburto Lab at Scripps Institution of Oceanography. Her role in the lab includes everything from designing and leading field expeditions across coastal Mexico to hosting undergraduate seminars for student volunteers.
Fieldwork from a mangrove site in La Paz, Baja Cailfornia Sur. (Photo credit: Astrid Hsu)
Over the past two years, Astrid has focused on developing the use of remote sensing to study mangrove forests across the Baja region, and other parts of Mexico. With the goal of facilitating the monitoring of these wetland habitats, Astrid—in collaboration with Engineers for Exploration—has been working to use drones to capture imagery that can contribute to regionally-specific machine-learning algorithms.
Photo 1: Results of our first version of the vector machine learning for detecting mangrove habitat. (Photo credit: Dillon Hicks)
Photo 2: Results of improved vector machine learning for detecting mangrove habitat. (Photo credit: Dillon Hicks)
Photo 3: Example of a hand labeled site for training algorithms. In this case, using QGIS and drawing polygons around habitat identified as mangrove. (Photo credit: Astrid Hsu)
Mangrove ecosystems serve as critical habitats for both terrestrial and aquatic species, while also providing various ecosystem services to humans including storm protection, erosion control, and carbon sequestration. However, the on-the-ground monitoring of mangroves is time and labor intensive. Mangrove forests are dense, and in Baja California in particular, they are also relatively low to the ground. For those who want to gather data on a particular forest, this means a lot of slow crawling through thickets of forest, all while racing the rising tide. With the use of drones, a series of aerial images can be taken and stitched together to form mosaic-like maps. This approach can provide a holistic view of a forest and serve as data inputs for a machine-learning algorithm. So far, Astrid and her team have been able to develop a machine-learning algorithm for the mangroves of La Paz region of Baja California Sur, providing important information on the extent and biomass of the forests there.
Fieldwork from a mangrove site in La Paz, Baja California Sur. (Photo credit: Astrid Hsu)
Their work has also created opportunities for conservation managers in Mexico, who have received training on the use of drones to monitor mangroves and other habitats, as well as on ways to apply these tools in their own ecosystem monitoring efforts.
According to Astrid, “as the climate warms, mangroves will increase their distribution northward. This is an opportunity for the border region to learn from its southern partners about the value of this habitat that may be seen as ‘invasive’, and to reconsider management.
Photo 1: Capacity-building with Mexico's National Comission of Natural Protected Areas, in a workshop on how the use of drones machine learning can be leveraged for management. (Photo credit: Diego Gamero)
Photo 2: Beni Guerrero and Eric Lo flying a drone over the mangroves of Baja California Sur. Photo credit: Astrid Hsu)
The Baja Working Group is a collaboration between the Climate Science Alliance and the International Community Foundation. Learn more here.