Assessing ecosystem function of a Piñon-Juniper woodland using a time series of high resolution satellite imagery and eddy covariance measurements

Combining recent advancements in satellite remote sensing with current eddy covariance measurement networks is a powerful way to improve our understanding of ecosystem processes. Remote sensing of semi-arid ecosystems requires temporal coverage sufficient to capture discrete responses in productivity as a result of stochastic patterns of precipitation, and adequate spatial resolution to monitor the patchwork of ecosystem heterogeneity. Eddy-covariance towers continuously measure ecosystem-atmosphere carbon and water exchange. However, even with ancillary data regarding phenologic patterns of the region, tower measurements are unable to inform us about differential response from the collection of plant functional types present within the measured tower footprint. We therefore tested the integration of eddy covariance data with a time series of high resolution (5 meter) RapidEye satellite images collected from late 2009 through mid 2011 over a 49 x 49 km area of piñon-juniper (PJ) woodland south of Mountainair, NM that includes two eddy covariance towers. One tower is in intact PJ woodland and the second tower is in a 200 m x 200 m section of PJ woodland in which all piñon >7 cm dbh (~1600 trees) were girdled to simulate the widespread piñon mortality that occurred throughout the SW in 2002. Due to the high spatial and temporal variability in soil moisture and sparse canopy cover at these sites (maximum LAI is ~ 2.1 and 1.8 in the control and girdled sites, respectively), we used site-specific lab based soil moisture reflectance curves to correct for moisture driven variability in soil reflectance. We used three vegetation indices to compare the phenological patterns of specific plant functional types at both tower sites: the traditional vegetation indices NDVI and MSAVI2, as well as a red-edge (690-730 nm) index NDRE which has demonstrated ability to remotely sense plant stress. We combine these remotely-sensed phenological patterns with the flux tower measurements using a site specific assimilation scheme that integrates temporal variability in reflectance from RapidEye with WorldView-2 derived classification maps to examine how widespread piñon mortality alters ecosystem processes in PJ woodlands. Our results suggest that the growing light gaps within the girdled canopy have facilitated an increase in cover by C4 grasses and forbs. The use of high spatial and temporal resolution satellite imagery in this study allowed the quantification of this progression in response to seasonal precipitation patterns and provides a mechanistic approach with which to explain changes in the eddy covariance data. These structural changes, in part, explain why widespread coniferous mortality in PJ woodlands alters ecosystem function to more closely resemble a juniper savanna.