Abstract:
In the northern hemisphere (NH), terrestrial ecosystems transition from net sources of CO2 to the atmosphere in winter to net ecosystem carbon sinks during spring. The timing of this transition is determined by the balance between ecosystem respiration (RECO) and primary production. Where there is strong seasonal variation in conditions that support soil decomposition and plant growth, the timing of soil respiration phenology is key to accurately estimating the amplitude of the terrestrial carbon sink. We diagnose an apparent phase bias in the RECO and NEE seasonal cycles estimated by the Terrestrial Carbon Flux (TCF) model framework and investigate its link to soil respiration mechanisms. Satellite observations of vegetation canopy conditions, surface meteorology, and model-enhanced estimates of soil moisture from the NASA SMAP Level 4 Soil Moisture product, are used to model a daily carbon budget for a global network of eddy covariance (EC) flux towers. Modifications to TCF are assessed for their impact on the RECO seasonal cycle: the inhibition of foliar respiration in the light (the Kok effect); a seasonally varying litterfall phenology; an O2 diffusion limitation on heterotrophic respiration (RH); and a vertically resolved soil decomposition model. Each of these model enhancements is also assessed by comparison to in situ chamber measurements. We find that the RECO phase bias can result from a bias in RECO magnitude, and that mechanisms which reduce NH spring RECO, like substrate and O2 diffusion limitations, can mitigate the phase bias. A vertically resolved soil decomposition model mitigates this bias by temporally segmenting and lagging RH throughout the growing season. Applying these model enhancements to estimating RECO at Continuous Soil Respiration (COSORE) sites verifies their improvement of RECO and NEE skill compared to in situ observations. Ultimately, these mechanisms can improve prior estimates of NEE for atmospheric inversion studies.