Authors: Alex Marden*, University of Texas - Austin, Thoralf Meyer, University of Texas - Austin, Amelia Eisenhart, University of Texas - Austin
Topics: Biogeography, Remote Sensing, Land Use
Keywords: savanna, fire ecology, land use, time series analysis, seasonality
Session Type: Poster
Presentation File: No File Uploaded
Fire intensity and vegetation response to fire in savanna systems are heavily influenced by fuel moisture levels, fuel loads, and vegetation structure at time of burn. These fuel characteristics vary seasonally and are affected by human land use. Investigating the seasonal dynamics of fire across diverse land uses improves understanding of fire characteristics and the resulting ecological effects under different anthropogenic conditions. This study investigated the presence and degree of variability in seasonal fire-vegetation dynamics in conservation, pastoral, and farm land uses in Botswana using a time series decomposition method.
MODIS time series were used to construct a fire history for the years 2001-2018 and to identify patterns of savanna fire and vegetation in the study area using Google Earth Engine. Seasonal patterns, multi-year trends, and trend breaks were observed in the different land use types using the Breaks for Additive Seasonal and Trend (BFAST) method. Peak fire season was similar across different land uses and occurred during the late dry season coinciding with buildup of dry fuels. Non-peak season fire patterns varied across land uses—conservation areas had more fires in non-peak seasons than pastoral and ranch areas. Furthermore, there is evidence of a multiyear relationship between variability of EVI and fire occurrence. Time-series methods that separate seasonal and long-term patterns are shown here as a promising tool for monitoring fire regimes in different land uses in semi-arid savanna and grassland systems.