Using landscape context to map invasive species with medium‐resolution satellite imagery

Abstract

The spread of invasive species is a global problem of major ecological and economic concern. Landscape level assessment of invasive spread is critical, but remote sensing (RS) analyses are often complicated by the spectral similarity of species and the need to balance spatial resolution with data storage and analysis complexity. One example is the ridge and slough landscape (RSL) of the Florida Everglades, where inflowing nutrients have facilitated large‐scale cattail invasions. Hand delineation of aerial imagery has been successful in mapping cattail spread, but this technique requires considerable time and effort. Computerized classification of medium‐resolution imagery would increase the ability of scientists to provide up‐to‐date data for water management decisions. Advances in RS technologies have created opportunities that were not previously available in landscapes such as the RSL—to automatically classify sawgrass and cattail communities with medium‐resolution satellite imagery using knowledge of the invasion ecology of cattail and landscape context. We developed a computer‐classification technique that provided measure of cattail expansion that matched ground‐truthed data and show an increase in cattail area (similar to previous estimates), but a reduction in the rate of expansion over time. Although this technique can miss small patches of plants that might indicated new invasions, its rapid mapping can improve tracking of invasion fronts in the Everglades and other landscapes.