Detection of extreme climate events in semi-arid biomes using a combination of near-field and satellite based remote sensing across the New Mexico Elevation Gradient network of flux towers

Abstract:

Semi-arid biomes in the Southwestern U.S. over the past decade have experienced high inter- and intra-annual variability in precipitation and vapor-pressure deficit (VPD), and from recent observations, are particularly vulnerable to both VPD and drought. Given the large land area occupied by semi-arid biomes in the U.S., the ability to quantify how climate extremes alter ecosystem function, in addition to being able to use satellites to remotely detect when these climate extremes occur, is crucial to scale the impact of these events on regional carbon dynamics. In an effort to understand how well commonly employed remote sensing platforms capture the impact of extreme events on semi-arid biomes, we coupled a 9-year record of eddy-covariance measurements made across an elevation/aridity gradient in NM with remote sensing data sets from tower-based phenocams, MODIS and Landsat 7 ETM+. We compared anomalies in air temperature, vapor pressure deficit, and precipitation, to the degree in variability of remote sensing vegetation indices (e.g, NDVI, EVI, 2G-Rbi, LST, etc.), and tower-derived gross primary productivity (GPP), across a range of temporal lags to quantify : 1) how sensitive vegetation indices from various platforms, LST, and carbon uptake are to climate disturbances, and the extremity of the disturbance; 2) how well correlated vegetation indices and tower fluxes are on monthly, seasonal and annual time scales, and if the degree to which they are correlated is related to the extent of climate anomalies during that period; and 3) the lags in the response of both GPP and vegetation indices to climate-anomalies and how well correlated these were on various time scales. Our initial results show differential sensitivities across a range of semi-arid ecosystems to drought and vapor pressure deficit. We see the strongest sensitivity of vegetation indices, and correlations between vegetation indices and tower GPP in the low and high elevation biomes that have a more distinct seasonality in ecosystem function than the middle elevation biomes. We discuss the implications of these results for the future of using empirical-based models that use vegetation indices to estimate terrestrial uptake in semi-arid biomes.