|Lawrence, Rick - MONTANA STATE UNIVERSITY|
|Wood, Shana - MONTANA STATE UNIVERSITY|
Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: October 29, 2005
Publication Date: November 1, 2005
Citation: Lawrence, R.L., Wood, S.D., Sheley, R.L. 2005. Mapping invasive plants using hyperspectral imagery and Breiman Cutler classifications (RandomForest). Remote Sensing of Environment. 100(2006):356-362. Interpretive Summary: Accurate weed maps are critical for planning and implementing effective management. Traditional hand mapping is very time-consuming and expensive. We tested the potential to use remotely sensed imagery to map leafy spurge and spotted knapweed. The sensing unit with 128-bands provided maps that were 74% accurate. Map accuracy was increased up to 86% when the images were refined using special techniques increasing their clarification. These accuracies are within the highest hand mapping accuracy levels.
Technical Abstract: Invasive nonindigenous plants are threatening the biological integrity of North American rangelands, as well as the economies that are supported by those ecosystems. Spatial information is critical to fulfilling invasive plant management strategies. Traditional invasive plant mapping has utilized ground-based hand or GPS mapping. The shortfalls of ground-based methods include the limited spatial extent covered and the associated time and cost. Mapping vegetation with remote sensing covers large spatial areas and maps can be updated at an interval determined by rangeland needs. The objective of the study was to map leafy spurge (Euphorbia esula) and spotted knapweed (Centaurea maculosa) using 128-band hyperspectral (5 m and 3 m resolution) imagery and assess the accuracy of these resulting maps. Random forest, a classification method using an ensemble of classification trees, was used to classify the imagery. Overall accuracy was 84% for the spotted knapweed classification, with class accuracies ranging from 60% to 93%; overall accuracy was 86% for the leafy spurge classification, with class accuracies ranging from 66% to 93%. Our results indicate that (1) the random forest algorithm can achieve substantial improvements in accuracy over single classification trees with this data and (2) it might be unneccessary to have separate accuracy assessment data when using random forest, as the algorithm provided a reliable internal estimate of accuracy.