A stochastic method to evaluate carbon cycle and atmospheric transport models using atmospheric observations
Authors
Ian N. Williams — Lawrence Berkeley National Laboratory
William Riley — Lawrence Berkeley National Laboratory
Margaret S. Torn — Lawrence Berkeley National Laboratory
Sebastien Christophe Biraud — Lawrence Berkeley National Laboratory
Marc L. Fischer — Lawrence Berkeley National Laboratory
Joseph A. Berry — Carnegie Institution for Science
Category
Atmospheric State & Surface
Description
Atmospheric observations provide important constraints on the carbon cycle, but challenges arise when interpreting differences between surface fluxes inferred from observational inverse methods and those predicted by carbon cycle models. This work develops a test of the atmospheric transport models used to infer surface fluxes from concentration measurements by comparing decay rates of observed and simulated fluctuations in boundary-layer concentration gradients. This stochastic method follows an analytical solution to the conservation equation and can successfully diagnose the boundary layer-free troposphere vertical mass exchange in data from the ARM Climate Research Facility’s Southern Great Plains site. The results motivate a stochastic-dynamical model to help identify sources of error in atmospheric inverse methods in terms of joint probability distributions for surface fluxes and vertical mass exchanges. This work is part of the LBNL/ARM Carbon Project measuring and modeling land-atmosphere interactions in the Southern Great Plains and North Slope of Alaska, supported by the DOE ARM, ASR, and Terrestrial Ecosystem Science (TES) programs.
