Future Directions in Bioluminescence Research

---

UNCLASSIFIED WORKGROUP SESSIONS

Modeling and Prediction
Discussion Leader: Dennis J. McGillicuddy, Jr.

An overarching goal of naval research in bioluminescence is to develop a suite of models that can be used to predict the distribution and variability of light producing organisms in the sea, and quantify the light exiting the water column from a bioluminescence event. The goal of bioluminescence prediction is derived in part from two naval needs. The first is to maximize the detection potential of naval bioluminescence sensors. The second is to minimize the risk of detection during nighttime naval operations. Both of these needs require a quantified characterization of the battlespace environment.

Progress toward this goal can be achieved through two interrelated modeling activities:

Objective 1: Statistically based "diagnostic" models

Objective 2: Coupled physical-biological "dynamical" models

The former use empirically-derived relationships amongst a range of observed variables (both climatological and contemporary) to predict bioluminescent activity. Such models exist today and were described in some detail in presentations by D. Lapota and P. Bissett (see Abstracts).

Diagnostic models based on climatologies offer information at the seasonal time scale. Their advantage stems from their incorporation of long-term time series to yield bioluminescence potential in a region of interest. Their disadvantage includes limited applicability to the actual coastal environment as it exists at the moment of interest. Contemporary diagnostic models require initialization of the current fields of bioluminescence potential in the region of interest. These fields may be directly measured, e.g. AUV, or limited measurements may be coupled with nowcast techniques. Their advantage is that they yield a more accurate determination of the in situ bioluminescence potential. Their disadvantage is that they require assets in the field during the time of interest in order to make the requisite nowcast.

It is reasonable to expect significant near-term (i.e. over the next five years) advancement in these types of models as research continues on historical data sets and those to be acquired in the near future.

In particular, nowcast of bioluminescence potential from contemporary diagnostic models may be expanded to incorporate prediction of the three-dimensional current flows in a region of interest. Such a coupling of bioluminescence tracers with the predictions of water flows may yield short-term predictions (24 to 48 hours) of bioluminescence potential in a dynamic coastal region.

Models of the second type are based on mechanistic representations of the underlying ecological principles which control the abundance of bioluminescent organisms. Such systems are created by incorporating the relevant biological processes into realistic circulation models. Given suitable initial conditions, forcing functions, boundary conditions and schemes for the assimilation of incoming data, these coupled models can be used to make predictions of the dynamical evolution of the natural system. Three-dimensional data-assimilative coupled physical-biological models are still in their infancy (Hofmann and Lascara, 1998). A presentation by D. McGillicuddy provided some examples of this kind of approach used in problems involving organisms that do not produce light (see Abstracts). Application of this approach to the prediction of bioluminescent organisms is a long-term prospect that will likely require a decade or more of basic research.

Conceptual Basis:

The extreme patchiness which is characteristic of planktonic populations poses a difficult sampling problem. Practical limitations preclude acquisition of observations which are dense enough to fully resolve all the relevant scales. The combination of observations with models via data assimilation offers an attractive solution which is consistent with observations where they exist and dynamically interpolated across data-sparse regions. Part-and-parcel with such synthetic products are the four-dimensional distributions of expected error, which are critical for analysis and interpretation of the predicted fields. A number of excellent examples of integrated observational and modeling systems exist today in a variety of sustained regional applications, and these are in various stages of development. Oscar Schofield reported on one of the more mature of such programs at the LEO-15 site (see Abstracts).

Most of these integrated observational and modeling systems are focused on issues other than bioluminescence. They seek to develop systems to predict ecological structure in response to changing physical conditions. Many of the issues that they face are the same issues that need to be addressed to predict bioluminescence potential. e.g., phytoplankton community structure, water column clarity, heterotrophic interactions. The naval concerns of bioluminescence detection are also directly dependent on these ecosystem issues, as well as the maximum bioluminescence potential. Integrating bioluminescence research into these larger programs may provide the necessary physical, chemical, biological, and optical information required to predict the ecosystem structure, and concomitant bioluminescence risks and potentials.

Biological formulation:

Light is produced by a tremendously diverse collection of organisms in the sea (see Peter Herring Abstract). Explicit treatment of all species is clearly intractable at this point. However, a large fraction of the total light produced in the near-surface waters of the coastal ocean is associated with two major groups: dinoflagellates and copepods. A consensus emerged that Naval needs could be met in large measure by predictive models of these two "functional groups." (Paul Bissett presented an example of a functional group model, which could be expanded to incorporate bioluminescence organisms. See Abstracts; Bissett et al., 1999a,b).

The functional group approach is particularly attractive by virtue of the leverage that can be achieved through partnership with other ecological modeling efforts which have embraced this concept. A variety of linkages are possible; the following describes three examples:

(1) Ocean biogeochemistry. The ocean carbon cycle modeling community has begun to progress from traditional biogeochemical models that treat all phytoplankton species in aggregate to more ecologically-based formulations with several functional groups. It has become apparent that this partitioning into several groups (e.g. diatoms, dinoflagellates, nitrogen fixing organisms) is needed to properly model the flows of carbon in the sea (U.S. JGOFS, 1998).

(2) Harmful Algal Blooms (HABs). Several recent initiatives in HAB research in the U.S. are focussed on dinoflagellates in particular. Two examples include Alexandrium spp. in the Gulf of Maine, and Gymnodinium breve in the Gulf of Mexico. Both areas currently have large regional efforts as part of the National ECOHAB program (ECOHAB, 1995; ECOHAB-GOM, 2000; ECOHAB-Florida, 2000).

(3) Ecosystem dynamics. The U.S. Globec program is emphasizing trophic levels ranging from zooplankton to fish. Population dynamics of Calanoid copepods is a central issue in that program’s study of Georges Bank (U.S. Globec, 1992); Calanus finmarchicus and Pseudocalanus spp. are two of the target organisms.

Although none of these efforts (1-3 above) focus on species that are prodigious light-producers, it is likely that they will result in important insights into ecological controls on the functional groups in which the most important bioluminescent organisms reside.

Clearly, not all of the organisms comprising the functional groups loosely referred to as dinoflagellates (autotrophic, heterotrophic and mixotrophic) and copepods produce light. Thus it is important to recognize that the development of predictive models for these groups is not in itself sufficient to meet Naval needs. Additional information will be needed in order to predict bioluminescent activity from bulk estimates of the biomass in these two functional groups. At present there does not appear to be a mechanistic basis on which to explicitly represent light-producing and non-light-producing components of these groups in ecological models. If such a basis does not become apparent in the future, operational models for naval purposes will have to rely on empirically derived relationships to convert from functional group biomass to an estimate of the potential for light production.

Alternatively, we may be able to predict the environmental conditions that give rise to bioluminescent organisms, while missing the numbers and location of the these organisms in a three-dimensional, time-dependent, prognostic model. In other words, we may be able to predict total dinoflagellate and copepod biomass, but miss the prediction of bioluminescence fraction of these organisms. If our ecological simulations include the predictions of water clarity at the wavelength of bioluminescence, i.e., the Inherent Optical Properties (IOPs) of the water column, we may be able to couple our IOP predictions to climatological and contemporary diagnostic models of bioluminescence potential. This may yield prediction of horizontal and vertical detection horizons for a pre-determined bioluminescence event.

Needs:

Within the general framework of the functional groups concept described above, there is a clear need to emphasize representative or "keystone" organisms. Suggestions included the dinoflagellates Gonyaulax, Protoperidinium, Noctiluca and the copepods Metridia and Pleuromamma. Model development will require extensive measurements in both the laboratory and the field to better define the autecology of the organisms, their relationships with both predator and prey, and of course space/time variability of their occurrence in the natural system.

Laboratory / Process Studies:

Much is known about the life history strategies, growth, reproduction, and mortality of various dinoflagellates and copepods. However, there are few such organisms for which enough of this kind of information is available to formulate relatively complete mechanistic models of their oceanographic ecology. Success of any effort to develop predictive dynamic models for bioluminescent organisms will depend critically on the quantification of vital rate processes and their dependence on various environmental parameters (e.g. temperature, salinity, ambient light, etc.). The need for this information will likely require a suite of both laboratory measurements and experimental work at sea.

Integrated Field Measurements and Modeling:

As data assimilative coupled physical-biological models of bioluminescent organisms mature, it will of course be necessary to collect four-dimensional (space/time) data sets with which such models can be tested. Acquisition of such data sets is a substantial undertaking that will require significant resources. A fully coupled observational and modeling strategy will be required in order to integrate the tremendous diversity of observations needed for such an endeavor. It is likely that measurements would be delivered from a variety of platforms, including ships, moorings, satellites, aircraft, drifters and AUVs.

Development of a rational basis for the coordinated deployment of these resources could be facilitated by modeling studies that have come to be known as "Observational System Simulation Experiments" (OSSEs). This activity begins with the construction of a model simulation that is characteristic of the natural system. The model run serves as a space/time continuous representation of reality, which is then subsampled in a specified fashion to produce a simulated data set. The simulated data are then fed into an analysis scheme in which they are synthesized into a reconstruction of reality. Direct comparison of the reconstructed field with the "truth" as defined by the original simulation thus provides a quantitative evaluation of that particular sampling strategy. The OSSE concept has its origins in dynamic meteorology (e.g., Charney et al., 1969) and is recognized as an important tool for the development of oceanographic sampling systems (Smith, 1993; Robinson et al., 1998). McGillicuddy et al. (in press) provide a specific example of this technique in which the synopticity of the U.S. Globec Broadscale Survey is evaluated.

Deliverables:

Long-term investment in a research program of this type will likely result in predictive models of the two functional groups which comprise a large fraction of the bioluminescent plankton in the coastal ocean. Biomass estimates of dinoflagellates D(x,y,z,t) and copepods C(x,y,z,t) will facilitate predictions of bioluminescent activity associated with these two groups, C’(x,y,z,t) and D’(x,y,z,t). Such information could then be fed into operational models used to quantify the bioluminescent potential for naval purposes, i.e. B(x,y,z,t) = f(C’(x,y,z,t) , D’(x,y,z,t)).

While our long-term goal is the prediction of bioluminescence and detection horizons, we should be able to provide nowcast/forecast information relevant to naval bioluminescence needs in the near-term. In particular, the simultaneous development of diagnostic and dynamical models should provide a stream of interim products that help characterize the bioluminescence potentials in the battlespace environment.

References

Charney, J., Halem, M. and Jastrow, R., 1969. Use of incomplete historical data to infer the present state of the atmosphere, Journal of Atmospheric Science 26, 1160-1163.

ECOHAB, 1995. The Ecology and Oceanography of Harmful Algal Blooms: A national research agenda. Woods Hole Oceanographic Institution, Woods Hole, MA.

ECOHAB- Florida: http://www.fmri.usf.edu/ecohab/ecofla.htm

ECOHAB- GOM: http://crusty.er.usgs.gov/ecohab/

U.S. JGOFS, 1998. Synthesis and Modeling Project. U.S. JGOFS Planning Report Number 21, S.C. Doney and J.L Sarmiento, Eds. U.S. JGOFS Planning Office, Woods Hole, MA.

Bissett, W.P., Walsh, J.J., Dieterle, D.A., and Carder, K.L. (1999a). Carbon cycling in the upper waters of the Sargasso Sea: I. Numerical simulation of differential carbon and nitrogen fluxes. Deep-Sea Research I 46, 205-269.

Bissett, W.P., Carder, K.L., Walsh, J.J., and Dieterle, D.A. (1999b). Carbon cycling in the upper waters of the Sargasso Sea: II. Numerical simulation of apparent and inherent optical properties. Deep-Sea Research I 46, 271-317.

Hofmann, E. E. and C. M. Lascara, 1998. An Overview of Interdisciplinary Modeling for Marine Ecosystems, In: The Sea, The Global Coastal Ocean:

Processes and Methods, A. R. Robinson and K. H. Brink, eds., Vol. 10, 507-540.

McGillicuddy, D.J., Lynch, D.R., Wiebe, P., Runge, J., Durbin, E.G., Gentleman, W.C., and C.S. Davis, 1999. Evaluating the synopticity of the U.S. Globec Georges Bank Broad-scale sampling pattern with Observational System Simulation Experiments. In press, Deep-Sea Research II.

Robinson, A., Lermusiaux, P. and Sloan, N., 1998. Data assimilation. In: The Sea 10, 541-594.

Smith, N. 1993. Ocean modeling in a global ocean observing system, Rev.

Geophys. 31, 281-317.

U.S. Globec, 1992. Northwest Atlantic Implementation Plan,

Technical Report 6 , U.S. Globec Scientific Steering Committee Coordinating Office, Solomons, Maryland.

   Next summary >>

  [Download printable PDF version of this page]