hide
Free keywords:
carbon cycle data assimilation; inverse modeling; adjoint;
terrestrial biosphere; uncertainty analysis; atmospheric carbon
dioxide
Primary productivity npp; comparing global-models; carbon-
dioxide emissions; inverse model; CO2; transport; exchange;
vegetation; uncertainties; constraints
Abstract:
This paper demonstrates a new method of assimilating atmospheric concentration data into terrestrial biosphere models. Using a combination of adjoint and tangent linear models of both the underlying biosphere model and the atmospheric transport model, we directly infer optimal model parameters and their uncertainties. We also compute biospheric fluxes and their uncertainties arising from these parameters. We demonstrate the method using the Simple Diagnostic Biosphere Model (SDBM) and data on the seasonal cycle of CO2 from 41 observing sites. In the model, the light-use efficiency for several biomes is well-constrained by concentration observations. Optimal values generally increase with latitude as required to match the seasonal cycle. Modeled Q(10) values are poorly constrained unless local flux measurements are also used. Values also increase with latitude but are less than the commonly assumed value of 2.
