Team Location: NASA Ames Research Center
Authors: Andrew Nguyen, Alexander Gole, Jarom Randall, Glade Dlott, Sylvia Zhang, Brian Alfaro
Advisors/Science Mentors: Dr. Jay Skiles, Cyndi Schmidt
Abstract: Invasive perennial pepperweed (Lepidium latifolium), commonly referred to as ÛÏtall whitetop,Û has increased in its spatial distribution throughout the western United States in the past 15 years. It is commonly found in alkaline environments such as the managed ponds of the South San Francisco Bay. Pepperweed outcompetes other native species for water and nutrients, further disturbing these already ecologically sensitive environments. The purpose of this study is to map the contemporary distribution of pepperweed throughout the restored South San Francisco Bay Salt Ponds and to create a habitat suitability model to predict the plants’ future spread. Pepperweed reflectance data were collected with the GER 1500 spectroradiometer along with presence and absence data of homogeneous sites in the form of GPS point polygons. The spectra and point data were classified in past and present Hyperion and ASTER images using the Spectral Angle Mapper (SAM) and supervised classifications to map the past and current distribution of this invasive species. The resulting distribution data served as input to a GIS multivariate habitat suitability model that includes variables determined to affect the spread of pepperweed. Variables included surface roughness (terrain), tidal extent, propensity for disturbance, salinity, proximity to levees and water, flow speed and direction, vegetation litter cover, and soil mineral type. The model was then used to predict areas most suitable for pepperweed growth. Another tool that can be used in the creation of a suitability model is the Generalized Additive Model (GAM) available from the Marine Geospatial Ecology Tools toolbox. GAM was used to investigate the statistical probability that each variable contributed to the habitat suitability of pepperweed. These models were applied to past Hyperion and ASTER images, then compared to more recent images for an accuracy assessment.
Video transcript available here.