US Joint Global Ocean Flux Study
Regional Ecosystem Modeling Testbed Project
First Regional Ecosystem Modeling Testbed Workshop
The first Regional Ecosystem Modeling Testbed Project Workshop was held
31 March – 02 April 2003 at the Center for Coastal Physical Oceanography,
Old Dominion University. Fourteen scientists from eleven different institutions
attended this highly successful meeting, including:
Larry Anderson (Woods
Hole Oceanographic Institution)
Robert Armstrong
(SUNY, Stony Brook)
James Christian
(CCCMA, Victoria)
Bob Daniels (Virginia
Institute of Marine Science)
John Dunne (NOAA/GFDL)
Jeff Dusenberry
(Woods Hole Oceanographic Institution)
Katja Fennel (Rutgers
University)
Marjorie Friedrichs
(Old Dominion University)
Raleigh Hood (HPL,
University of Maryland)
John Klinck (Old
Dominion University)
Dennis McGillicuddy
(Woods Hole Oceanographic Institution)
J. Keith Moore (University
of California, Irvine)
Yvette Spitz (Oregon
State University)
Jerry Wiggert (Old
Dominion University)
The goals of the workshop, as outlined at the JGOFS SMP
Summer 2002 Meeting in Woods Hole, were (1) to increase awareness of the
similarities and differences between the various ecosystem models currently
in use by the attending JGOFS investigators, (2) to compare simulations
from multiple ecosystem models using a standard physical framework and identical
validation data, (3) to develop methodologies for characterizing the strengths/weaknesses
of the different models, and (4) to provide instruction on the use of the
adjoint method as applied to one-dimensional ecosystem models.
Results stemming from this workshop were presented in
a collaborative poster at the JGOFS Open Science Meeting in May 2003. The
possibility of extending this analysis into a submission to the final JGOFS
SMP DSRII volume was also discussed.
Prior to the workshop, participants were provided with
the testbed framework, which included Fortran routines for processes including
advection, diffusion, mixing, sinking, and attenuation. In addition, standard
forcing time series (mixed layer depth, vertical and horizontal advection,
solar radiation), initial conditions, and biogeochemical validation data
(chlorophyll-a, zooplankton biomass, nitrate, sediment trap, primary production)
were provided. Each participant was responsible for plugging his/her ecosystem
model subroutine into the testbed framework, and providing time series of
ecosystem model output (plankton, nutrients, etc...) Friedrichs synthesized
these results and presented the model comparisons for discussion at the
workshop.
A second component of the workshop entailed a tutorial
on the adjoint method, as applied to one-dimensional ecosystem models. On
the third day participants had the opportunity to use on-site computational
facilities to develop the adjoint code for their specific ecosystem model,
or revise their ecosystem model to enhance performance in relation to the
other models presented. The workshop was marked by many lively discussions
addressing topics such as the formulation of the cost function and the utility
of the automatic adjoint compiler. Excellent presentations on the variational
adjoint method, the utility of the automatic adjoint compiler and a posteriori
error analyses were made by workshop participants. In addition, the collaborative
synthesis poster for the JGOFS Open Science Meeting was outlined, the feasibility
of submitting a paper for the final DSRII SMP volume was discussed, and
plans for the second Regional Ecosystem Modeling Testbed Project Workshop
were made. One of the highlights of the meeting was also the fact that by
the conclusion of the workshop, the adjoint for most of the eleven participating
models had been written.
Go to the top of this page.
Written by J. Klinck, April, 2003. Last updated by M. Friedrichs, September,
2004.
Send questions or comments by e-mail to marjy@ccpo.odu.edu