Skip to main content
ARS Home » Research » Publications at this Location » Publication #242071

Title: Testing an Invasive Weed Prediction Model for Leafy Spurge using Hyperspectral Remote Sensing

item Hunt, Earle - Ray

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 7/27/2009
Publication Date: 9/10/2009
Citation: Hunt Jr, E.R. 2009. Testing an invasive weed prediction model for leafy spurge using hyperspectral remote sensing [abstract]. 2009 HyspIRI Science Workshop. 2009 CDROM.

Interpretive Summary:

Technical Abstract: Leafy spurge (Euphorbia esula L.) is a noxious invasive weed that infests over 1.2 million hectares of land in North America. One of the fundamental needs in leafy spurge management is cost-effective, large-scale, and long-term documentation and monitoring of plant populations. Leafy spurge is a good candidate for detection via remote sensing because the distinctive yellow-green color of its bracts is spectrally unique when compared to co-occurring green vegetation. During 1999, Airborne Visible-Infrared Imaging Spectrometer (AVIRIS) imagery were acquired in northeastern Wyoming and ground vegetation data were collected nearby Devils Tower National Monument in Crook County, Wyoming. Hyperspectral analyses were used to classify leafy spurge presence/absence; overall accuracy using the spectral angle mapper was 76%. The classification data were used to test the Weed Invasion Susceptibility Prediction (WISP) model, which uses available geospatial data layers to predict the potential distribution of various invasive weeds. We tested the WISP model at two new locations, Fishlake National Forest in Utah and the South Unit of Theodore Roosevelt National Park in North Dakota. Both sites had model predictions significantly better than chance using kappa analyses. Future applications of the WISP model may be incorporation into decision support systems.