Observations of regional net ecosystem exchange (NEE) of CO2 for 1997–2007 were analyzed for climatic controls on interannual variability (IAV). Quantifying IAV of regional (104–106 km2) NEE over long time periods is key to understanding potential feedbacks between climate and the carbon cycle. Four independent techniques estimated monthly regional NEE for 104 km2 in a spatially heterogeneous temperate-boreal transition region of the north central United States, centered on the Park Falls, Wisconsin, United States, National Oceanic and Atmospheric Administration tall tower site. These techniques included two bottom-up methods, based on flux tower upscaling and forest inventory based demographic modeling, respectively, and two top-down methods, based on tall tower equilibrium boundary layer budgets and tracer-transport inversion, respectively. While all four methods revealed a moderate carbon sink, they diverged significantly in magnitude. Coherence of relative magnitude and variability of NEE anomalies was strong across the methods. The strongest coherence was a trend of declining carbon sink since 2002. Most climatic controls were not strongly correlated with IAV. Significant controls on IAV were those related to hydrology, such as water table depth, and atmospheric CO2. Weaker relationships were found with phenological controls such as autumn soil temperature. Hydrologic relationships were strongest with a 1 year lag, potentially highlighting a previously unrecognized predictor of IAV in this region. These results highlight a need for continued development of techniques to estimate regional IAV and incorporation of hydrologic cycling into couple carbon-climate models.
