Classifying plant species

Plant species maps provide a baseline for monitoring the world’s ecosystems, which are already responding to climate change. Plant species maps are crucial for many applications, including monitoring invasive species expansion, tracking wildfire disturbance recovery, and detecting vegetation disturbances such as insect infestation. There are many techniques for developing species maps, including ground-based approaches, but remote sensing technology allows for the investigation of ecological processes and systems on larger spatial and temporal scales. Imaging spectroscopy, or hyperspectral remote sensing, makes discrimination of plant species possible because the hundreds of narrow bands can be used to detect subtle spectral shifts between species that are caused by differences in chemistry, physiology, and structure. I use hyperspectral data and machine learning to develop plant species maps that are robust across seasons and years.

Plant species classifications zoomed in on a portion of the Santa Ynez Mountain foothills. Only species and classes present in this subset are included in the legend. For species code refer to Meerdink et al., 2019.

Previous Work

Accurate knowledge of seasonal and inter-annual distributions of plant species is required for many research and management agendas that track ecosystem health. Airborne imaging spectroscopy data have been used successfully to map plant species, but often only in a single season or over a limited spatial extent due to data availability. NASA’s Hyperspectral Infrared Imager (HyspIRI) preparatory airborne campaign flew an imaging spectrometer from 2013 to 2015. This dataset captured a severe drought and thus provided the opportunity to evaluate species discrimination over an extreme range in environmental conditions. We evaluated the portability of image-based training data and accuracy of species discrimination. Our results highlight the need to use reference spectra that adequately represent the phenological and biophysical status of the plant species within an image for accurate mapping. Our research provides relevant insight for advanced species-mapping techniques across broad spatial and temporal scales using imagery from sensors like HyspIRI.

Current Work

Hurricanes create major ecosystem disturbances that can alter the structure and function of native plant communities and increase the presence of non-native invasive plant species. It is projected that hurricane intensity and frequency will increase under future climate conditions, necessitating better understanding and prediction of how hurricanes will affect the dispersal, establishment, and performance of non-native plant invaders. Brazilian peppertree, an invasive plant, is one of the most widespread and problematic invaders in Big Cypress and Everglades National Parks. Approximately $2.5 million per year is spent controlling Brazilian peppertree. Hurricanes may alter the distribution of Brazilian peppertree by dispersing seeds to new habitats, creating novel conditions (e.g., high light in previously close-canopy areas), or both. This project will elucidate the habitats and environmental conditions most susceptible to post-hurricane Brazilian peppertree establishment and determine the post-hurricane distribution. We will use a combination of field surveys, hyperspectral imagery, and greenhouse experiments to understand better how hurricanes may affect the distribution of Brazilian peppertree in coastal habitats. The methodologies developed in this project would not only decrease the cost and amount of workforce necessary for developing vegetation maps but would also allow for annually updated vegetation maps that could provide vital information for ecological forecasting.


Lin, Y., Meerdink, S.K., & Gader, P.D. (2022). Spectral Transformations for Multi-Temporal Hyperspectral Classification. IEEE Geoscience and Remote Sensing Letters, 19, 6005405, 1-5.

Meerdink, S.K., Roberts, D.A., Roth, K.L., King, J.Y., Gader, P.D., & Koltunov, A. (2019). Classifying California plant species temporally using airborne hyperspectral imagery. Remote Sensing of Environment, 232, 111308.