Sensitivity of ecosystem parameters to simulated satellite ocean color data using a coupled physical-biological model of the North Atlantic

Abstract

A means of assimilating simulated satellite ocean color data with a coupled physical-biological model of the North Atlantic Ocean is implemented, allowing the relative sensitivities of different biological parameters to those data to be investigated. The model consists of an eddy-permitting general circulation model derived from the WOCE Community Modeling Effort and a nitrogenbased, four-compartment NPZD marine ecosystem model. Many of the parameters in marine ecosystem models are poorly known and via assimilation, we hope to better constrain their values. The control parameters chosen for the variational assimilation are the model parameters involved in parameterizationsof recycling as these are the most poorly known. Simulated observations are taken while following several oats seeded in varying dynamical biogeochemical provinces of the North Atlantic model domain over a six-month period. Twin experimental results show that, for the given functional forms of growth, mortality and grazing, the following parameters can be successfully recovered from simulated satellite ocean color data: nitrate and detrital recycling parameters in the trade wind domain, zooplankton parameters at higher latitudes (westerly wind and polar domains), and the phytoplankton mortality rate in all regions. By simultaneously assimilating ocean color data in different biological provinces, it becomes possible to successfully constrain all ecosystem parameters at once.

Constraining parameters in marine pelagic ecosystem models – is it actually feasible with typical observations of standing stocks?

“…The advantages of this approach have been appreciated in previous studies already (e.g. Friedrichs, 2001;Gunson et al, 1999;Lawson et al, 1996;Schartau et al, 2001;Spitz et al, 1998). One major difference here, however, is the usage of reddish noise.…”

“…Here, we apply this constant ratio in order to stay consistent with typical N-P-Z-D modelling approaches (e.g. Chai et al, 1995;Gunson et al, 1999;Löptien et al, 2009;Spitz et al, 1998). Note that some recent model developments include an explicit representation of chlorophyll (e.g.…”

Reviews and syntheses: parameter identification in marine planktonic ecosystem modelling

“…Prior uncertainty can be represented by prior probability densities in Bayesian approaches or plausible ranges in non-Bayesian approaches. To account for nonnegativity constraints, prior distributions typically include lognormal (Parslow et al, 2013), square-root normal (Gunson et al, 1999), or beta distributions (Dowd and Meyer, 2003), although normal distributions may yet be applicable for parameters that are well constrained above zero (Parslow et al, 2013). To our knowledge no application has yet incorporated prior correlations between parameters in (i.e.…”

Assimilation of ocean colour data into a Biogeochemical Flux Model of the Eastern Mediterranean Sea

“…Such techniques basically seek for the unknown parameters that minimize the misfit between model simulations and data, e.g. the adjoint method (Fennel et al, 2001;Friedrichs, 2001;Gunson et al, 1999;Lawson et al, 1996;Spiitz et al, 1998;Vallino, 2000), direct minimization methods (Fasham and Evans, 1995;Prunet et al, 1996), and Monte Carlo methods (Harmon and Challenor, 1996;Matear, 1995). Recently, focus shifted toward the use of Kalman filter based sequential assimilation techniques to directly compute estimates of the system state, as these methods allow for efficient handling of the model uncertainties while intermittently adjusting the model trajectory each time new observations are available (Ghil and Malanotte-Rizzoli, 1991).…”