Out of the Woods: Delineating Disturbances in the Southern Rockies

EarthzineDetecting Land Cover Change & Disturbances, DEVELOP 2016 Summer VPS, DEVELOP Virtual Poster Session

Category: Land Cover Change & Disturbances

Project Team: Rocky Mountain Agriculture

Team Location: U.S. Geological Service at Colorado State University ‰ÛÒ Fort Collins, Colorado

Imagery from Landsat 4, 5, and 8 were analyzed with LandTrendr software to create a map of disturbance events and assist in future management decisions. Image Credit: Rocky Mountains Agriculture Team

Imagery from Landsat 4, 5, and 8 were analyzed with LandTrendr software to create a map of disturbance events and assist in future management decisions. Image Credit: Rocky Mountains Agriculture Team

Authors:

Peder Engelstad

Christopher Beddow

Stephanie Krail

Amandeep Vashisht

Mentors/Advisors:

Dr. Paul Evangelista (Colorado State University, Natural Resources Ecology Lab)

Tony Vorster (Bioenergy Alliance Network of the Rockies Feedstock Supply Team)

Past/Other Contributors:

Brian Woodward (Center Lead)

Abstract:

In recent decades, the Rocky Mountains of northern Colorado and southern Wyoming have experienced extremely high levels of forest disturbance. Methodologies for mapping and labeling disturbance and classifying historical harvest and thinning events on the landscape level have not been readily available in the past. However, recent literature has paved the way for refined approaches, such as change detection software and predictive classification models. This project provided a more complete dataset for the National Park Service Rocky Mountain National Park (RMNP), the Colorado Forest Restoration Institute (CFRI), and the Bioenergy Alliance Network of the Rockies (BANR) Feedstock Supply Team. Landsat 4 and 5 Thematic Mapper (TM) and Landsat 8 Operational Land Imager (OLI) imagery were integrated into the LandTrendr algorithm to detect magnitude, duration, and extent of past forest disturbances. A suite of classification algorithms, including the Boosted Regression Trees (BRT) and the Random Forests (RF) classification models, were used to conduct analyses at the landscape level across a temporal scale of 30 years. A labeled disturbance history, including harvest and thinning events, were provided to project partners, which filled gaps in their past records can lead to enhanced decision-making in the future.

Previous story | Main Page | Next story