Multi-scale model inter-comparisons of CO2 and H2O ...

Object Details


XVI International Conference on Computational Methods in Water Resources (CMWR-XVI) Ingeniørhuset

Multi-scale model inter-comparisons of CO2 and H2O exchange rates in inhomogeneous canopies
Author:Gabriel Katul <> (Duke University)
Mario Siqueira <> (Duke University)
Presenter:Gabriel Katul <> (Duke University)
Date: 2006-06-18     Track: Special Sessions     Session: Ecohydrology: From Detailed Descriptions To General Synthesis?

Models for the exchange of CO2 and H2O between the atmosphere and terrestrial ecosystems are needed for assessing the effects of anthropogenic CO2 emissions on atmospheric concentration of CO2. To date, no single model captures the entire spectrum of variability of the processes affecting CO2 and H2O transfer and storage within terrestrial ecosystems; rather, a modular approach is adopted in which the forcing and response variables are coupled over an inherent or assumed time scale that is then integrated to longer time scales. The effect of such modular parameterization of the “fast” processes and their cross-scale interaction with the slowly varying processes on long-term carbon sequestration remains a subject of investigation. Here, we compared four existing process-based stand-level models of varying complexity (3-PG, PnET II, Biome-BGC, and SECRETS-3PG) and a newly proposed nested model with 4 years of eddy-covariance water vapor (LE) and CO2 (Fc) fluxes measured above a maturing loblolly pine forest near Durham, North Carolina, USA. The nested model resolves the “fast” CO2 and H2O exchange processes using canopy turbulence theories and radiative transfer principles while slow evolving processes were resolved using standard carbon allocation methods modified to improve leaf phenology. The model comparisons showed strong linkages between carbon production and LAI variability, which necessitates the use of multi-layer models to reproduce the seasonal dynamics of LAI, Net Ecosystem Exchange (NEE) and LE. However, our findings suggest that increasing model complexity, often justified for resolving faster processes, does not necessarily translate into improved predictive skills at all time scales, especially annual and longer. To address this spectral discrepancy, we performed a variance component analysis of NEE at annual time scales that revealed that most of the inconsistency seems to originate from different model responses to drought. None of the models tested here adequately captured drought effects on water and CO2 fluxes. Furthermore, the good spectral performance of some models on inter-annual time scales appears to stem from erroneously capturing LAI dynamics and from over sensitivity to droughts that injects unrealistic variability at longer time scales.