Ecosystem Modelling

Introduction Themes People Highlights Publications Softwares Resources Site Map

 Themes
» Biogeochemical Modelling / Physical-Biological Modelling / Inverse Modelling

Biogeochemical Modelling

The biological pump removes CO₂ from the surface ocean during primary production and exports organic carbon to the deep sea. This process effectively lowers atmospheric CO₂ concentration by creating a vertical gradient in dissolved inorganic carbon concentration. In most regions, sinking particulate organic carbon represents the dominant means of transport of organic carbon to the deep ocean. By far the largest pool of organic carbon in the ocean is dissolved organic matter ( DOM ). Marine DOM is released by plankton organisms and accumulates in the surface ocean, as only part of it is recycled by bacteria or photochemical reactions. DOM is exchanged only very slowly with the atmosphere, but could have a significant effect on the global carbon cycle on longer time scales, owing to the large size of this carbon pool.

Deep-water formation in the North Atlantic is thought to be the main mechanism of transferring this accumulated DOM to the deep sea, where it can remain for several thousand years. The Labrador Sea as a primary site for North Atlantic Deep Water formation is a main entry point for dissolved organic matter to the deep ocean. DOM exchange can dominate the total exchange of organic carbon between surface and deep water in the Labrador Sea.

The Labrador Sea is characterized by strong interannual variations in both physical circulation and plankton dynamics. Recent advances in ocean circulation modelling have made it feasible to model interannual variability of plankton ecosystem processes and trophic structure. We use an optimization-based adaptive approach to model plankton-DOM dynamics in 1D in the Labrador Sea forced by temperature, velocity, and mixing coefficients obtained from POP model simulations for this area. We use the plankton-DOM model to analyse the response of the local carbon cycle to climate variations and its translation into changes in the mixing and export of DOM and particulate organic carbon between the surface and deep ocean in the Labrador Sea.

Top of the page

Physical-Biological Modelling

The main goal of our physical-biological modelling work is to assess the behaviour of in-house planktonic ecosystem models when subjected to time dependent physical conditions resulting from climate variability. This work aims at: i) reproducing patterns of various ecosystem variables (e.g. phytoplankton and zooplankton biomass, nutrients) observed at different temporal and spatial scales; ii) providing quantitative estimates of the processes inherent to the ecosystem (e.g. primary production); and iii) providing tools to assist in ecosystem-based management and eventual ecological forecasting.

So far, our physical-biological modelling work has concentrated on three main approaches: i) zero-dimensional (or box) models; ii) one-dimensional (or water column) models; and iii) three-dimensional models.

In the zero-dimensional approach, the ecosystem is essentially contained within a single box corresponding to the surface mixed layer at any given time. This surface box interacts with water underneath for which the properties are known (i.e. boundary conditions). The ecosystem model is physically forced by the surface mixed layer depth as diagnosed from temperature and salinity observations. This modelling approach is being applied in two main projects: i) modelling the interdecadal variability of Nova Scotia's continental shelf based on historical data archived at BIO in the Hydrographic and BioChem databases; ii) modelling the interannual variability of an inshore bay (Lunenburg Bay, Nova Scotia, Canada) based on high frequency data collected as part of the CMEP interdisciplinary initiative.

In the one-dimensional approach, the ecosystem is represented by multiple vertical boxes that span the water column. The boxes interact with each other and are subject to the physical conditions generated by a prognostic ocean model (in our case, GOTM). The ocean model responds to local meteorological forcings (e.g. heat and momentum fluxes) and other physical forcings (e.g. tidal and vertical advection currents). This modelling approach is being applied in two main projects: i) modelling the interannual variability at a fixed shelf station (i.e. Station 2, located ~50 km southeast of Halifax, Nova Scotia, Canada) based on data collected ca. biweekly as part of the AZMP; ii) modelling the intraseasonal variability at shelf Station 2 based on high frequency and low spatial resolution data collected as part of experimental deployments of the SeaHorse profiler.

Finally, our three-dimensional modelling efforts have so far concentrated on the inline coupling of simple NPZ models within the ocean circulation model POP. This work has been applied in assessing the behaviour of different ecosystem parameterizations at the basin scale level (i.e. North Atlantic). Moreover, this work forms the basis for the implementation of biological data assimilation (i.e. chlorophyll) capability within physical-biological models.

Top of the page

Inverse Modelling

Inverse modelling is well established in the physical sciences and is increasingly used in aquatic ecology and biogeochemistry. Inverse modelling can be viewed as simply the reverse of simulation or forward modelling. In simulation modelling, the model is run forward in time from specified initial conditions and parameters. Typically, the simulation is compared to observations to assess its realism or predictive power. In inverse modelling, the observations are used to reconstruct the initial conditions and the parameters. Inverse modelling is used because these simulation parameters are often poorly known and we want to use the observations we have to learn as much about them.

The focus of our research is actually on static models, where there the time dependence is removed by assuming that the system is at or near steady state. Such static models can be used to reconstruct energy and material flows between ecological compartments. Because of the difficulties involved in measuring all flows in complex ecosystems, inverse Modelling can be an efficient way to infer these flows without having to measure them all. However, inverse modelling has its own limitations which we try to address in some of our research.

Click here for a short tutorial on inverse modelling of marime ecosystems.