An offline implicit solver for simulating prebomb radiocarbon

Ann Bardin, François Primeau, Keith Lindsay

Research output: Contribution to journalArticlepeer-review

27 Scopus citations

Abstract

It takes several thousand years for the deep-ocean concentration of natural radiocarbon to come to equilibrium with surface fluxes, making it computationally too expensive to routinely simulate it with moderate- to high-resolution ocean models. We present an implicit solver for computing prebomb δ14C that requires the equivalent of only a few tens of model years to reach equilibrium. The solver uses a Newton-Krylov algorithm with a preconditioner based on a coarse-grained annually-averaged tracer-transport operator. Coarse-graining provides a general approach for developing preconditioners for models of increasing resolution. We implemented and tested the solver for the ocean component of the Community Earth System Model (CESM) with a nominal horizontal resolution of 1° × 1° and with 60 vertical levels. Simulated δ14C values are in good agreement with observations at the surface and in the North Atlantic, but the deep North Pacific simulated values show a substantial bias, with prebomb radiocarbon δ14C values translating to ages that are twice the observationally based estimate. This bias is substantially larger than published simulations obtained with coarser resolution models, suggesting that increasing model resolution does not automatically improve the fidelity of the deep ocean ventilation processes. We therefore recommend that natural δ14C be used as a deep-ocean ventilation metric for critically evaluating deep ocean circulation.

Original languageEnglish
Pages (from-to)45-58
Number of pages14
JournalOcean Modelling
Volume73
DOIs
StatePublished - Jan 2014

Keywords

  • Global ocean modeling
  • Implicit solver
  • Newton-Krylov
  • Prebomb radiocarbon
  • Preconditioner

Fingerprint

Dive into the research topics of 'An offline implicit solver for simulating prebomb radiocarbon'. Together they form a unique fingerprint.

Cite this