Fort Collins, CO
Colorado Agriculture
Down to Earth: Reconstructing Forest Disturbances with Landsat




[Music]


BRIAN WOODWARD>>Throughout history, ecological disturbances have shaped the earth's landscape.Whether forest harvest or wildfire, disturbances can cause pronounced changes to ecosystem dynamics. In response, land managers and scientists around the world are seeking to better understand when, how, and where disturbances are occurring, so that they can make more informed conservation and land management decisions.Since 2002, the Mountain Pine Beetle epidemic spread through the two states, killing over 650,000 acres of native lodgepole and ponderosa pine forests. Sporadic forest fires occasionally sweep through the area, and timber harvests have long been an important part of the region’s economy.  The Colorado Agriculture team partnered with BANR, CSFS, and BDSR to map all high magnitude disturbances between 1975-2011 in Landsat Path 34, Row 32.


BOB STURTEVANT>> The mission of the conservation group here at Ben Delatour is to keep the camp as healthy and sustainable as possible.  I’d like to say we have a  really good process right now but we don’t. We have a basic map of the camp as mainly a recreation map. There’s not a really good, overall map of the camp. And that’s what we’re really looking for working with DEVELOP. 


ERIC ROUNDS>> NASA’s Landsat program has collected continuous earth observation data that documents the changes that have occurred on earth's surface for over 40 years. By employing the full Landsat time series in conjunction with the LandTrendr model, an advanced change detection algorithm, the Colorado Agriculture team was able to analyze, label, and map disturbance history in our study area on a pixel-by-pixel basis.Following the team’s creation of a time series landsat imagery “stack” and the model’s preprocessed Landsat inputs, the LandTrendr algorithm analyzed changes in the spectral trajectory of a pixel throughout the full time series, using selected indices such as tasseled cap wetness and the normalized burn ratio, to delineate periods of stability and change on the landscape. The Colorado Agriculture team took advantage of a second term to expand their study area extent to all managed forested areas in the study area.


BRIAN>> This project utilized imagery from 1975-2011 using Landsat satellites 1-5. For each year, the team downloaded 2-3 scenes atmospherically corrected for surface reflectance for the months of July & August. The selected scenes were preprocessed using LandsatLinkr, a new landsat preprocessing tool created by Justin Braaten at Oregon State University. The tool generates a 6-band LEDAPS, 3-band Tasseled Cap composite images, and a cloudmask for each Landsat scene, which are the inputs required for the LandTrendr algorithm. Concurrently, the team used a modified Landtrendr code that integrates LandsatLinkr cloud free tasseled cap composites as inputs, which would extend the time series to include imagery from Landsat satellites 1, 2, 3 , and 7. The resulting outputs of the LandTrendr algorithm provided the team with a multi-band raster called the “Greatest Disturbance” output. Included in the GD file are bands containing the magnitude of disturbances, the duration of the disturbance event, and, finally, the year that the disturbance was first detected. Using these three primary bands, the team was able to delineate disturbances of interest in a GIS based upon the disturbance magnitude and duration, and identify the year that each disturbance first appeared, known as the year of disturbance onset. 


ERIC>> These points were then analyzed visually using high-resolution NAIP imagery to determine if the evaluation and classification of LandTrendr disturbance outputs within a GIS were accurate. The end-products of the project were used to produce a set of reliable maps, both digital and paper based, that will prove useful to our project partners in a variety of ways- from supporting biomass model estimates to the implementation of sustainable forest harvests. 


[Music by Dr. Welch]