Recommendations

Increase accessibility to numerical models by observationalists

In the past, modelers and observationalists have often worked in isolation. This sometimes results in models that are not consistent with field measurements in either their formulation or their behavior. Often sampling programs do not resolve the appropriate temporal and spatial scales or they focus on specific processes that are either not parameterized in models or have only secondary or tertiary importance. As observations become more sophisticated in terms of both the processes that can be measured and the scales that can be resolved (both microscales and global scales), models have assumed new importance as a framework within which data may be interpreted. Moreover, the increasing focus on studies of coupled biological/physical processes and the need for scientific research to focus on the prediction of ecosystem response to climate change has also elevated the role of numerical modeling. Thus the complexity of both models and observations require a much closer interaction between those who build and operate models and those who collect and analyze data.

Large-scale coupled biological/physical models are still in a state of infancy. It would be premature to develop a single model of the Southern Ocean as it would not resolve critical physical and biological processes. Rather, efforts should be focused on a variety of models that are limited either in the scales or processes that are resolved. We expect that eventually researchers may wish to link some of these models together; this will require that the underlying assumptions be clearly stated so that inherent contradictions are avoided.

Specific Recommendations

  1. Encourage Southern Ocean JGOFS and GLOBEC activities that have both a modeling and a field component

  2. Develop a variety of models focusing on specific processes or hypotheses but with clearly defined interfaces and documented assumptions so that other researchers can understand and evaluate the models

  3. Archive output of numerical models much as field and satellite observations are archived and distributed

  4. Encourage the development of models that are structured as a set of testable hypotheses that can be addressed by appropriately designed sampling strategies

Improve modeling capabilities in advance of Southern Ocean field studies for use in designing sampling programs and analyzing data

Although present models are not capable for developing detailed sampling strategies, they can be used to develop specific hypotheses for the field component of JGOFS and GLOBEC. Emerging research areas, such as data assimilation and nested models, would benefit by expanded research in advance of the JGOFS and GLOBEC field programs. Various diagnostic techniques, such as estimating advective fluxes, could be used to design specific sampling strategies at the Southern Ocean station sites.

Specific Recommendations

  1. Encourage modelers to work with researchers participating in Southern Ocean JGOFS and GLOBEC field studies

  2. Encourage development of data assimilation techniques for biogeochemical modeling

  3. Continue development of embedded or nested models which incorporate high resolution models within lower resolution models

  4. Use models to simulate advective fluxes around planned Southern Ocean stations and compare with observations as part of model diagnostics

Improve observing capabilities to take advantage of and test numerical models

Although many observationalists consider them primitive, present models can be used to guide field programs in the types of observations that should be made to improve parameterizations and constrain model behavior. For example, many of the core JGOFS measurements are standing stock observations, yet most of the uncertainties in existing models concern rate parameters. There was a strong consensus that data must be collected on size and functional classes of the ecosystem. Although there is controversy regarding the level of detail that must be included in models, clearly there are profound implications on carbon cycling as the structure of the ecosystem changes. As assimilation techniques improve, it is essential that we develop a scientific basis for our estimates of the error fields associated with the data being assimilated into the models. This may require tedious, closely-spaced observations as these error fields may vary seasonally as well as regionally. Lastly, the Southern Ocean will always be difficult to sample using conventional ship techniques. The JGOFS and GLOBEC programs should draw upon developments in other programs to use low-cost or expendable sensors to increase the scales that can be sampled.

Specific Recommendations

  1. Evaluate present JGOFS core observations in context of the needs of existing numerical models

  2. Develop models that resolve critical time and space scales as identified in field measurements

  3. Collect information on size and functional groups

  4. Quantify error covariances for data fields that are assimilated into models

  5. Continue to encourage the development of new automated and low-cost sensors to extend sampling coverage of the Southern Ocean

Establish a regular program to further the development of coupled physical/biogeochemical models

Models of the Southern Ocean ecosystem must resolve complex physical dynamics as well as complicated chemical and biological interactions. Because of the nature of the circulation in this vast region of the ocean, these models must have high spatial resolution as well. However, our overall goal should be the development of a closer alliance between models (which will always be gross simplifications of reality) and observations (which will always provide a biased and undersampled view of reality). We should continue activities that strengthen the links between these two complementary ways of looking at a complex system.

Specific Recommendations

  1. Regularly assess the state of our knowledge and modeling capabilities

  2. Support annual workshops where models can be run and evaluated by both modelers and observationalists


homepage contents previous section next section