Articles | Volume 18, issue 8
Biogeosciences, 18, 2727–2754, 2021
https://doi.org/10.5194/bg-18-2727-2021
Biogeosciences, 18, 2727–2754, 2021
https://doi.org/10.5194/bg-18-2727-2021

Research article 30 Apr 2021

Research article | 30 Apr 2021

Optimal model complexity for terrestrial carbon cycle prediction

Caroline A. Famiglietti et al.

Data sets

NEE and LAI prediction metrics for DALEC model suite (COMPLEX experiment) Caroline Famiglietti https://doi.org/10.6084/m9.figshare.13409096.v1

Model code and software

COMPLEX Analysis Code Caroline Famiglietti https://doi.org/10.5281/zenodo.4716391

Download
Short summary
Model uncertainty dominates the spread in terrestrial carbon cycle predictions. Efforts to reduce it typically involve adding processes, thereby increasing model complexity. However, if and how model performance scales with complexity is unclear. Using a suite of 16 structurally distinct carbon cycle models, we find that increased complexity only improves skill if parameters are adequately informed. Otherwise, it can degrade skill, and an intermediate-complexity model is optimal.
Altmetrics
Final-revised paper
Preprint