Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4375521 | Ecological Modelling | 2016 | 13 Pages |
•Parameters in ecological models vary across different locations.•Parameter variation is related to biological factors and environmental conditions.•Our study indicates acclimation of plants and microbes to environment.
Terrestrial ecosystem models have been extensively used in global change research. When a model calibrated with site-specific parameters is applied to another site, how and why the parameters have to be adjusted again in order to fit data well are pervasive yet underexplored issues. In this exploratory study, we examined how and why model parameters of a Flux-Based Ecosystem Model (FBEM) varied across different sites. Parameters were estimated from data at 12 eddy-covariance towers in the conterminous USA using the conditional inversion method. Results showed that optimized values of these parameters varied across sites. For example, the estimated coefficients in the Leuning model, glgl and D0D0, exhibited high cross-site variation, but the ratio of internal to air CO2 concentration (fCifCi) and canopy light extinction coefficient (knkn) varied little among these sites. Parameters greatly varied with ecosystem types at adjacent sites where climate conditions were similar. Five parameters (activation energy of carboxylation, EKcEKc; activation energy of oxygenation, EVmEVm; ecosystem respiration, Reco0; temperature sensitivity of respiration, Q10Q10; and stomatal conductance coefficient, D0D0) were highly correlated with mean annual temperature and precipitation across sites, which were distributed in different climate regions of conterminous US. Our results indicate that individual parameters vary to different degrees across sites and parameter variation can be related to different biological factors (e.g., ecosystem types) and environmental conditions (e.g., temperature and precipitation). It is essential to further examine magnitudes of and mechanisms underlying the parameter variation in ecosystem models so as to improve model prediction.