Satellite observations reveal seasonal redistribution of northern ecosystem productivity in response to interannual climate variability

Abstract

Interannual variability (IAV) in ecosystem productivity may reveal vulnerabilities of vegetation to climate stressors. We analyzed IAV of northern hemisphere ecosystems using several remote sensing datasets, including longstanding observations of the normalized difference vegetation index (NDVI) and more novel metrics for productivity including solar-induced chlorophyll fluorescence (SIF) and the near-infrared reflectance of vegetation (NIRv). Although previous studies have suggested SIF better tracks variations in ecosystem productivity at seasonal timescales, we found that satellite datasets (including SIF) and eddy covariance flux tower observations were subject to significant uncertainty when assessing IAV at fine spatial scales. Even when observations were aggregated regionally, IAV in productivity estimated by the various satellite products were not always well correlated. In response to these inconsistencies, we applied a statistical method on regionally aggregated productivity data in four selected North American ecoregions and identified two dominant modes of IAV—seasonal redistribution and amplification—that were consistent across satellite datasets. The seasonal redistribution mode, which played a stronger role at lower latitudes, associated high (low) spring productivity with warm (cold) spring and summer temperatures, but also with lower (higher) productivity and water availability in summer and fall, indicating that enhanced growth in spring may contribute to an earlier depletion of water resources. The amplification mode associated an increase (decrease) in productivity across the growing season with above-average (below-average) summer moisture conditions. Even though our statistical analysis at large spatial scales revealed meaningful links between terrestrial productivity and climate drivers, our analysis does suggest that IAV and long-term trends in presently available novel and more established satellite observations must be interpreted cautiously, with careful attention to the spatial scales at which a robust signal emerges.