11-04, 13:30–14:00 (America/New_York), Lake Audubon
California’s beach plants are visually similar to the sand they grow on. I will classify plant pixels using aerial imagery and elevation data using R’s spatial packages.
Plants that grow on California’s beaches are surprisingly hard to detect with traditional remote sensing methods. Plants are small and patchy (< 1 meter wide). A beach typically has a dozen or so different plant species. They are usually silvery in color (highly reflective), not green, and if they are green, they’re sticky so they are covered in sand. It’s difficult to distinguish small, sand colored plant pixels from actual sand. To make matters worse, the sand they live on shifts daily so the physical structure of their habitat is not stable or the same from day to day. This means your imagery needs to be from the same day as your elevation data or nothing will match up topographically. What’s a dune researcher to do? In this talk, I’ll explore some avenues for working with this difficult ecosystem. I will try to not only tell the plants from the sand, but tell the plants apart from each other using aerial imagery and elevation data using R’s spatial packages.
Michele Tobias holds a PhD in geography from University of California Davis, as well as an MS from University of Michigan, and a BA from UCLA. She currently works at UC Davis DataLab as a geospatial data scientist helping researchers with their data needs. Her personal research interests include using open source tools to understand California’s sandy beach vegetation and geomorphology. She also has an interest in using her geospatial skills to help underserved research communities. Michele has served on the board of directors for OSGeo (international) and Technocation (OSGeo US), as well as arts non-profit organizations, and is a founding member and coordinator of #maptimeDavis.