We used data assimilation to estimate the
contributions of GPP, heterotrophic (Rh) and autotrophic (Ra) respiration to
Net Ecosystem Exchange at the Niwot Ridge long-term eddy covariance site using
5 years of data. The assimilation problem is solved by optimizing state and
parameter values in a version of the PnET ecosystem model by minimizing the
misfit between modeled and observed NEE, subject to Bayesian prior estimates of
the model parameters and initial state. Seventeen free parameters, about half
of the total, are estimated, with the remaining parameters defined from other
studies. The model computes GPP, Rh and Ra fluxes for each day and night, and
thus produces an estimate of the separation of NEE into its components. We
checked the model’s partitioning of the NEE into GPP and total respiration by
comparing the modeled and observed diurnal NEE cycle, and evaluated the Rh-Ra
partitioning by comparing modeled and observed Net Primary Productivity, which
constrains this partitioning since GPP- Ra=NPP. While some discrepancies exist,
overall the assimilation model had considerable skill on diurnal to interannual
timescales.
Author: W. Sacks, D. Schimel, R. Monson, G. Churkina (sacks at ucar dot edu)
Filesize: 12.68 Kb