Most numerical models of the upper ocean ecosystem are based on coupled partial differential equations with growth, loss, interaction, and diffusion terms. The basic model has been used in oceanography for many decades, although there have been many enhancements such as size classes, complex grazing and nutrient uptake terms, sophisticated mixed layer models, etc. As these models have grown in complexity, there are more adjustable parameters that must be estimated and more uncertainty about the exact forms of the parameterizations. Simple changes in parameters can have dramatic effects on model behavior. Several studies are investigating methods to reduce the number of parameters to those that capture most of the possible model behaviors.
As ocean models move towards a closer coupling with observations through data assimilation, it becomes essential that we know far more about the various parameters and functional forms than simply their mean and variance. Assimilation models require that we characterize the temporal and spatial variability of these parameters in order to fill in the gaps in time and space. This is a daunting task. For example, we know decorrelation scales of phytoplankton biomass in only a few locations in the world ocean; little is known about the decorrelation scales of phytoplankton growth rates.
Numerical models, including data assimilation models, also require experimental design and sampling strategies directed towards the specific questions being addressed. Much of the field data that have been used to provide model parameters and functional forms was gathered to solve specific scientific questions and hypotheses that are not always related to the questions being addressed by the model. For example, models of the relationship between chlorophyll concentration and diffuse attenuation may be based on field measurements from tropical waters, and it is not appropriate to apply such functional forms in models of high latitude processes.
The Southern Ocean will be the site of major field campaigns for both JGOFS and GLOBEC. There is still great uncertainty about the regulation of primary productivity in the Southern Ocean; iron limitation, grazing, and light limitation have been invoked. Near the ice edge, processes are even more complicated. Existing coupled biological/physical models must contend with a wide range of processes, many of which (such as iron limitation) have not yet been incorporated into existing models.
Given the expanse of the Southern Ocean and its isolation, field programs are by necessity both limited and costly. The upcoming JGOFS and GLOBEC Southern Ocean projects represent a unique opportunity to collect data on Southern Ocean biogeochemistry and ecological processes. Campaigns by other countries, including the United Kingdom, Australia, France, Germany, Japan, and South Africa, will also provide important data sets along with long-term studies such as the Palmer Long Term Ecological Research (LTER) program. It is unlikely we will be able to assemble these resources again. Given the predicted sensitivity of the Southern Ocean to climate change (and the resulting feedbacks), we must improve our ability to make predictions about the functioning of the Southern Ocean with only limited data sets in the future.
Physical forcing is particularly intense in the Southern Ocean. Strong wind forcing, large seasonal (and interannual) variations in ice cover, and mesoscale eddies are some of the processes that play critical roles in Southern Ocean dynamics. Weak stratification (relative to waters at lower latitudes) gives rise to short dynamical scales. The internal radius of deformation decreases towards the south, ranging from 20 km to 8 km. Bottom topography has a much stronger effect on the flow than at mid-latitudes because weak stratification allows bottom disturbances to penetrate to the surface. Coupled with the smaller dynamical scale, this means that small topographic features can have large-scale dynamical effects. This physical environment has strong links with biological processes that must be accounted for in both our field and modeling programs.
The focus of the workshop was an assessment of our present state of knowledge from both observations and models. We assessed where our greatest uncertainties lie and where small improvements in observation strategies and models would result in large increases in understanding. We estimated the time and space scales over which we can make useful predictions about the Southern Ocean. As part of this assessment, we explored the needs of the observational community in terms of models. We also sought to outline the type of measurement program that would lead to significantly improved models.
The workshop developed the following recommendations.