Causes of uncertainty in observed and projected heterotrophic respiration from Earth System Models

Heterotrophic respiration (RH) is a large component of the terrestrial carbon cycle, but one poorly simulated by Earth system models (ESMs), which diverge significantly in their historical and future RH projections. There is little understanding, however, of the causes of this variability and its consequences for future model development and scenario evaluation, and examining the relationships between RH and key climate variables may help to understand where and why models are divergent. We quantified the statistical relationships between RH and other terrestrial/climate variables across a suite of 25 ESMs from the Coupled Model Intercomparison Project phase 5 (CMIP5) for the 20th and 21st centuries, comparing the models both to each other and to an observation-driven global RH dataset. Compared to observations, ESMs consistency overestimate both the magnitude and climate sensitivity of global RH. The relationship between RH and surface air temperature (TAS) is strong, especially at high latitudes, and largely consistent across models. The observed RH and precipitation (PR) relationship is strong and positive ( r  ≥ 0.5, P  < 0.005), but few models consistently show this sensitivity of RH to PR. The RH-TAS relationship explored here, and more pattern scaling methods more generally, can be used to efficiently explore uncertainty and projected changes in RH under a wide range of future emission scenarios, and understand how models' structural and parametric choices produce divergent results. Because uncertainty in RH has large effects on ESM projections of future climate, this may help direct attention to relationships in the carbon cycle that contribute to this uncertainty.