When is a biogeochemical model too complex? Objective model reduction and selection for North Atlantic time-series sites

The degree of structural complexity that should be incorporated in marine biogeochemical models is unclear. We know that the marine ecosystem is complex, and that its observed behaviour is attributable to the interaction of a large number of separate processes, but observations are scarce and often insufficient to constrain more than a small number of model parameters. This issue is addressed using a novel algorithm that systematically removes model processes that are not constrained by observations. The algorithm is applied to a one-dimensional, eight component ecosystem-biogeochemistry model at two North Atlantic time-series sites. Between 11 and 14 of the 30 model parameters can be removed at each site with no significant reduction in the model’s ability to fit upper ocean (0–200 m) biogeochemical tracer and productivity data. The statistically optimal model structures and parameters provide estimates of the most likely state variables and fluxes at each site. Differences in these estimates between the two sites indicate that the optimal models are specialised to both the physical environment and the assimilated observations. At each site the heavily reduced models may thus be suitable for diagnostic purposes but may not be sufficiently complex for more general applications, such as in global ocean general circulation models, or for predicting the response of marine systems to environmental change.

[1]  John Skilling,et al.  Data analysis : a Bayesian tutorial , 1996 .

[2]  H. Ducklow,et al.  Plankton succession and carbon cycling at 47°N 20°W during the JGOFS North Atlantic Bloom Experiment , 1993 .

[3]  James W. Murray,et al.  Functional responses for zooplankton feeding on multiple resources: a review of assumptions and biological dynamics , 2003 .

[4]  T. D. Dickey,et al.  Influence of mesoscale eddies on new production in the Sargasso Sea , 1998, Nature.

[5]  Mark R. Abbott,et al.  Configuring an ecosystem model using data from the Bermuda Atlantic Time Series (BATS) , 2001 .

[6]  J. Toggweiler,et al.  Downward transport and fate of organic matter in the ocean: Simulations with a general circulation model , 1992 .

[7]  K. Buesseler,et al.  Carbon and nitrogen export during the JGOFS North Atlantic Bloom experiment estimated from 234Th: 238U disequilibria , 1992 .

[8]  James G. Richman,et al.  Data assimilation and a pelagic ecosystem model: parameterization using time series observations , 1998 .

[9]  C. Garside,et al.  The “f-ratio” on 20°W during the North Atlantic Bloom Experiment , 1993 .

[10]  Andrew J. Watson,et al.  Ecosystem dynamics based on plankton functional types for global ocean biogeochemistry models , 2005 .

[11]  C. L. Leonard,et al.  An Iron-Based Ecosystem Model of the Central Equatorial Pacific , 1999 .

[12]  Thomas R. Anderson,et al.  Plankton functional type modelling : running before we can walk? , 2005 .

[13]  Andreas Oschlies,et al.  Simultaneous data-based optimization of a 1D-ecosystem model at three locations in the North Atlantic Ocean: Part 2) Standing stocks and nitrogen fluxes , 2003 .

[14]  David R. Anderson,et al.  Model selection and multimodel inference : a practical information-theoretic approach , 2003 .

[15]  Jens Schröter,et al.  Modelling carbon overconsumption and the formation of extracellular particulate organic carbon , 2007 .

[16]  Cory P. McDonald,et al.  Using a model selection criterion to identify appropriate complexity in aquatic biogeochemical models , 2010 .

[17]  J. Ryther,et al.  Annual variations in primary production of the Sargasso sea off Bermuda , 1961 .

[18]  Jerald B. Johnson,et al.  Model selection in ecology and evolution. , 2004, Trends in ecology & evolution.

[19]  Thomas R. Anderson,et al.  Parameter optimisation techniques and the problem of underdetermination in marine biogeochemical models , 2010 .

[20]  G. Hurtt,et al.  A pelagic ecosystem model calibrated with BATS data , 1996 .

[21]  Fredrik Gustafsson,et al.  Twenty-one ML estimators for model selection , 1995, Autom..

[22]  Andreas Oschlies,et al.  Towards an assessment of simple global marine biogeochemical models of different complexity , 2010 .

[23]  J. Sarmiento,et al.  Ecosystem behavior at Bermuda Station “S” and ocean weather station “India”: A general circulation model and observational analysis , 1993 .

[24]  N. M. J. Crout,et al.  Is my model too complex? Evaluating model formulation using model reduction , 2009, Environ. Model. Softw..

[25]  P. Mayzaud,et al.  The importance of the time factor in the response of zooplankton to varying concentrations of naturally occurring particulate matter 1 , 1978 .

[26]  Marjorie A. M. Friedrichs,et al.  Ecosystem model complexity versus physical forcing: Quantification of their relative impact with assimilated Arabian Sea data , 2006 .

[27]  Walker O. Smith,et al.  Temperature effects on export production in the open ocean , 2000 .

[28]  Scott C. Doney,et al.  Assessment of skill and portability in regional marine biogeochemical models : Role of multiple planktonic groups , 2007 .

[29]  C. McClain,et al.  Biogeochemical modelling of the tropical Pacific Ocean. I: Seasonal and interannual variability , 2001 .

[30]  S. J. Tanner,et al.  Iron, primary production and carbon-nitrogen flux studies during the JGOFS North Atlantic bloom experiment , 1993 .

[31]  G. Evans,et al.  The Use of Optimization Techniques to Model Marine Ecosystem Dynamics at the JGOFS Station at 47 degrees N 20 degrees W [and Discussion] , 1995 .

[32]  Adrian P. Martin,et al.  Accounting for unresolved spatial variability in marine ecosystems using time lags , 2006 .

[33]  T. Kana,et al.  Dynamic model of phytoplankton growth and acclimation: responses of the balanced growth rate and the chlorophyll a:carbon ratio to light, nutrient-limitation and temperature , 1997 .

[34]  Peter G. Challenor,et al.  Addressing the impact of environmental uncertainty in plankton model calibration with a dedicated software system: the Marine Model Optimization Testbed (MarMOT 1.1 alpha) , 2011 .

[35]  Richard J. Matear,et al.  Parameter optimization and analysis of ecosystem models using simulated annealing: a case study at Station P , 1995 .

[36]  V. S. Ivlev,et al.  Experimental ecology of the feeding of fishes , 1962 .

[37]  D. Carroll.  Chemical laser modeling with genetic algorithms , 1996 .

[38]  M. Follows,et al.  Interannual variability of phytoplankton abundances in the North Atlantic , 2001 .

[39]  A. H. Taylor,et al.  Seasonal succession in the pelagic ecosystem of the North Atlantic and the utilization of nitrogen , 1993 .

[40]  Andreas Oschlies,et al.  Eddy-induced enhancement of primary production in a model of the North Atlantic Ocean , 1998, Nature.

[41]  E. Maier‐Reimer,et al.  Ocean-circulation model of the carbon cycle , 1990 .

[42]  Eileen E. Hofmann,et al.  A data assimilation technique applied to a predator-prey model , 1995 .

[43]  Willard Van Orman Quine,et al.  On empirically equivalent systems of the world , 1975 .

[44]  Jacques Monod,et al.  LA TECHNIQUE DE CULTURE CONTINUE THÉORIE ET APPLICATIONS , 1978 .

[45]  A. Wood,et al.  Towards the systematic simplification of mechanistic models , 2006 .