Parameter optimisations of a marine ecosystem model in the Southern Ocean

Download files
Access & Terms of Use
open access
Copyright: Kidston, Mehera
Altmetric
Abstract
The Southern Ocean plays a central role in global biological production (Sarmiento et al., 2004). Quantifying the mechanisms that regulate magnitude and variability in plankton biomass in the Southern Ocean is a key challenge in understanding controls on global climate (Busalacchi, 2004). Numerical models are an integral means of understanding large scale views of seasonal plankton cycles (Fasham et al., 1990). Marine biogeochemical models are characterised by non-linear dynamics and the annual model trajectories are highly sensitive to the parameters used to run them. Experiments were performed to select a stochastic inverse method to objectively estimate the parameters of a simple four component nitrogen based mixed-layer marine ecosystem model for the Southern Ocean. Twin experiments using the Metropolis-Hastings algorithm and simulated annealing show that simulated annealing holds promise as a standard means of assigning the parameters of marine biogeochemical models in a way that improves the model agreement to available observations. SeaWiFS surface chlorophyll estimates in the Sub-Antarctic Zone show low concentrations south west of Tasmania and high concentrations south east of Tasmania. Simulated annealing was used to estimate the model parameters at two locations in the Sub-Antartic Zone (station P1 at , and station P3 at , ) through assimilation of SeaWiFS chlorophyll observations. Model parameter estimates were compared to in situ parameter estimates from the SAZ-Sense (Sub-Antarctic Zone Sensitivity to environmental change) project stations P1 and P3 in the austral summer of 2007. The parameter estimates suggest that different ecosystems are present within the Sub-Antarctic Zone. Station P3 has higher regenerated production with an f-ratio of 0.57 compared to P1 which has an f-ratio of 0.70, indicating larger size fractionated phytoplankton at P1 with a greater capability to sink and smaller size fractioned organisms at P3 with greater recycling ability. Different biological processes rather than different physical conditions between the two sites are responsible for the difference in ecosystem function in our experiments. At P3 the optimisation results in satisfactory estimates of recycling rates but underestimates primary production, zooplankton biomass and zooplankton grazing. We conclude that the same ecosystem model structure is not applicable at both stations and we need additional processes at P3 to reproduce the observed seasonality of phytoplankton and the observed primary productivity. We hypothesize that the missing processes in the ecosystem model at P3 are iron limitation of phytoplankton and the seasonal variations in atmospheric deposition of iron. SeaWiFS surface chlorophyll estimates and Levitus nitrate estimates in the Southern Ocean south of Australia (140 E) show that this region is characterised by a high-nitrate low-chlorophyll (HNLC) regime typical of Southern Ocean waters. The HNLC conditions become more prominent moving south from the Sub-Antarctic Zone, with surface chlorophyll generally decreasing and nitrate increasing with latitude. Simulated annealing was used to fit the ecosystem model to SeaWiFS surface chlorophyll data in the Sub-Antarctic Zone, Polar Frontal Zone and Antarctic Zone. We hypothesise that bio-availability of iron limits phytoplankton growth in this region. The most reliable physiological indicator of iron availability was investigated by optimising the model parameters for maximum photosynthetic growth and maximum photosynthetic efficiency of phytoplankton. The parameter optimisations indicate that phytoplankton growth rates in the Polar Frontal Zone and Antarctic Zone are limited by some process not explicitly included in this model, with iron availability being the most likely candidate. In our experiments variations in ecosystem functioning caused by iron availability are more significant than differences caused by light availability. Based on these optimisations we support the contention that micronutrient availability, such as iron, is the primary cause of the HNLC conditions in the Australian sector of the Southern Ocean. Unification of the information provided by observationalists and modellers is a valuable approach for improving understanding of marine ecosystem dynamics. Measurements of atmospheric deposition of iron, phytoplankton growth rates, primary production, f-ratio and processes contributing to regenerated production throughout the year are needed to support Southern Ocean modelling studies.
Persistent link to this record
Link to Publisher Version
Link to Open Access Version
Additional Link
Author(s)
Kidston, Mehera
Supervisor(s)
Baird, Mark
Creator(s)
Editor(s)
Translator(s)
Curator(s)
Designer(s)
Arranger(s)
Composer(s)
Recordist(s)
Conference Proceedings Editor(s)
Other Contributor(s)
Corporate/Industry Contributor(s)
Publication Year
2010
Resource Type
Thesis
Degree Type
PhD Doctorate
UNSW Faculty
Files
download whole.pdf 1.57 MB Adobe Portable Document Format
Related dataset(s)