Abstract Title: Spatiotemporal analysis of remotely-sensed imagery to explore the relationship between vegetation change and fire regime on military training lands.
Author(s):
Anne Jacquin - University of Toulouse, Purpan School of Engineers (France)
Shawn Hutchinson* - Kansas State University
Michel Goulard - National Institute of Agronomy Research (France)
Stacy L. Hutchinson - Kansas State University
Abstract:
Fire is a significant disturbance in grasslands and, combined with other environmental and anthropogenic factors, can exert strong control over many spatiotemporal patterns observed in vegetation. Depending on the fire regime and local conditions, amounts of aboveground herbaceous biomass can change significantly. On military installations, maintaining healthy vegetated landscapes and continuous vegetation cover is critical to providing realistic, and sustainable, soldier training experiences. The U.S. Army Integrated Training Area Management (ITAM) program is responsible for ensuring training lands are available to meet Army operational needs now, and into the future, while simultaneously minimizing landscape degradation. At Fort Riley, Kansas, grassland burning is a common management practice to prevent shrub encroachment, improve habitat for select wildlife, and to maintain accessible training areas for soldier use. One effort currently underway to assist the Fort Riley ITAM program is to help illustrate the impact of fire in meeting these sometimes conflicting grassland management goals. Remotely-sensed image datasets were used to create indicators related to (1) vegetation change and (2) characterizing the fire regime in terms of frequency and seasonality. Both indicators were measured between 2000 and 2010 using MODIS and Landsat 5 TM time series images. Image analysis was performed using a spatial generalized linear model (GLM) where the structure of residuals were used to stratify the study area according to spatial variations in the relationship between vegetation change and fire regime. The use of this spatial statistical tool produced a parsimonious model which was found to be consistent with expert knowledge.
