Authors: Gjertrud (Sierra) Aney*, New Mexico State University, Michaela Buenemann, New Mexico State University
Topics: Natural Resources, Landscape, Remote Sensing
Keywords: Remote sensing, Landsat, drought, tree mortality, New Mexico
Session Type: Illustrated Paper
Start / End Time: 10:00 AM / 11:40 AM
Room: Canal St. Corridor, Sheraton, 3rd Floor
Presentation File: No File Uploaded
Tree die-off driven by extreme atmospheric conditions and insect infestations presents a major problem for land managers across the U.S. Southwest. The 2002/2003 drought and bark beetle outbreak, for example, triggered severe tree mortality in the pinyon-juniper woodlands of New Mexico, which resulted in increased fire risk due to more standing dry fire-fuel; habitat loss for woodland organisms; and altered rates and patterns of land surface energy, water, and nutrient exchanges. To enhance the scientific understanding of woodland dynamics and support the sustainable management of woodlands, our objectives were to: 1) map tree cover changes before, during, and after the drought (1997, 2000, 2001, 2002, 2003, 2004, and 2005) in a 30,000 km2 area in north-central New Mexico; 2) predict probability of tree mortality across the area; and 3) explain tree mortality given a range of potential human and environmental predictor variables. To meet these objectives, we used multiple endmember spectral mixture analysis of Landsat 5 Thematic Mapper imagery, image differencing, and two spatial models (logistic regression and Maxent). Our results suggest that mortality varied across space (e.g., greatest in mid-elevations) and between time periods (e.g., greatest in the 2001-2002 period). Moreover, environmental variables were more predictive of tree mortality than human variables, whereby the relative importance of environmental variables differed between time periods (e.g., variables related to soil moisture were the strongest predictors during the 2001-2002 period).