Location: Rangeland Resources & Systems ResearchTitle: Big data reveals drivers of vegetation phenology and cattle productivity in semi-arid rangelands
Submitted to: Annals of the American Association of Geographers
Publication Type: Abstract Only
Publication Acceptance Date: 11/13/2019
Publication Date: N/A
Technical Abstract: Vegetation dynamics on semi-arid rangelands are characterized by strong spatial and temporal variability. The contributions of this spatial and temporal variability to cattle productivity (body weight gains) have been little studied in combination. Understanding relationships between cattle weight gains and vegetation dynamics during the growing season, as well as the underlying drivers of spatio-temporal vegetation variability, are integral to informing rangeland management decision-making. We analyzed relationships between cattle weight gains and vegetation dynamics from 2013 to 2019 in the shortgrass steppe using vegetation production and phenology metrics extracted from freely available satellite imagery for the Central Plains Experimental Station in Colorado, USA. We derived a pseudo-daily vegetation index from a fusion of two satellite sensors, allowing for higher spatial and temporal resolution than from either sensor alone. Total vegetation production and phenology (a surrogate for timing of production) varied substantially within and across 130-ha pastures, highlighting the importance of the satellite fusion data product. Vegetation metrics (production and phenology) strongly influenced cattle weight gains within and across years. Drivers of observed phenology patterns were daily precipitation amounts, edaphic variables (e.g., topography, soils) and plant community composition classes derived from very high resolution airborne hyperspectral data acquired by the National Ecological Observatory Network (NEON) at the site. This study provides new insights into using Big Data analytics to ascertain highly dynamic relationships among landscape heterogeneity, vegetation metrics and cattle production in semi-arid rangelands. Products of these analytics can be integrated with near-real-time data streams into predictive models and decision support tools.