Meeting Documents

A Guide to Unraveling the Terrestrial Biosphere’s Integrated Sensitivity to Just About Anything (Invited)

Abstract

The world’s land ecosystems are experiencing unprecedented changes at unprecedented rates. In turn, through rapidly changing net land-atmosphere carbon, water and energy fluxes, functional shifts in land ecosystem are modulating the trajectory of the entire Earth System. Satellite observations reveal both episodic and sustained land-atmosphere carbon exchanges and ecosystem state changes which state-of-the-art models struggle to explain. These model errors are innately rooted in process compensation challenges: while ecosystem functional changes are taking place in every corner of the globe in response to environmental change, numerous compensation effects (spatial, temporal and functional) make it insurmountably difficult to resolve the terrestrial biosphere's net sensitivity to global change. Ultimately, resolving the spectrum of dynamical timescales requires precise, observation-informed dynamical knowledge of the biosphere’s responses to human and natural phenomena in order to resolve these integrated sensitivities. Through a generalized Bayesian framework of the biosphere’s states, inherent process controls, exogenous forcing and observation-acquired model skill, we show how observations are central to unraveling mechanistic inter-relationships across the terrestrial biosphere’s time-integrated responses; this framework can then, in principle, be used to probabilistically retrieve the biosphere’s sensitivity to any hypothesized internal state change or exogenous forcing event. We demonstrate this paradigm in action---through the implementation of the CARDAMOM model-data integration framework---to comprehensively resolve the terrestrial biosphere’s response a swath of recent environmental phenomena, including mortality events, fires, inter-annual variability, multi-decadal climate trends and rising atmospheric CO₂. We conclude on how joint state-parameter optimization of ecosystem function is a necessary and timely path forward for accurately leveraging dynamical model knowledge of the terrestrial biosphere within Earth System projections and feedback assessments.