Predicting Treatment Windows for Invasive Buffelgrass in Southern Arizona using MODIS and Climate Data

Thursday, 18 December 2014
Cynthia Wallace, USGS Western Geographic Science Center, Tucson, AZ, United States, Jake F Weltzin, USA National Phenology Network, Tucson, AZ, United States, Susan M Skirvin, Contract Geostatistician, Tucson, AZ, United States, Caroline Patrick-Birdwell, Southern Arizona Buffelgrass Coordination Center, Tucson, AZ, United States and Helen Raichle, USGS Arizona Water Science Center, Tucson, AZ, United States
The increasing spread and abundance of an invasive perennial grass, buffelgrass (Pennisetum ciliare), represents an important shift in the vegetation composition of the Sonoran Desert in southern Arizona. Buffelgrass out-competes native species and alters fire regimes, and its control and management is a high-priority issue for resource managers who seek to preserve the unique and iconic Sonoran Desert flora. Herbicidal treatment of buffelgrass is most effective when the vegetation is actively growing; however, the erratic timing and length of active buffelgrass growth periods in southern Arizona confound effective management decision-making. The goal of our research is to enable the strategic application of buffelgrass herbicide by using remote sensing data to detect when and where buffelgrass is photosynthetically active. We integrated ground-based observations of buffelgrass phenology (green-up and senescence) in the Tucson, Arizona area with climate information and Moderate-resolution Imaging Spectroradiometer (MODIS) satellite imagery at 250m spatial and both 8-day and 16-day composite temporal resolution to understand dynamics, relationships and resonance between these disparate datasets during 2011 to 2013. Fourier harmonics analysis was used to derive land surface phenology (LSP) metrics from MODIS Enhanced Vegetation Index (EVI) greenness data and to quantify the temporal patterns of the climate and phenophase abundance datasets. Regression analyses and statistical tests were used to identify correlations between temporal patterns of the data sets. Our results reveal strong correlations between the observed greenness of in-situ buffelgrass and satellite LSP metrics, confirming that MODIS-EVI data can be a useful indicator of active buffelgrass growth at multiple scales. The analysis also reveals strong harmonics between precipitation and greenness, but with a lagged response, suggesting that precipitation can be a predictor of the location and intensity of buffelgrass green-up at landscape scales. This information can be used by resource managers to treat buffelgrass during optimal conditions.