Final Report Summary - SIMPLE (Spatially-implicit modelling of plankton ecosystems)
The SIMPLE project had three objectives:
1. to investigate the effects of spatial variability in plankton concentrations on fluxes of carbon and nutrients in marine ecosystems;
2. to assess how well these effects may be captured by a mathematical approximation known as Second moment closure (SMC);
3. to validate such approximations using observational data.
We addressed objective 1 using a high-resolution ocean model, i.e. a detailed computer simulation like those used for predicting the weather. The computed carbon and nutrient fluxes were compared with those computed in control simulations with the spatial variability artificially removed. We found that spatial variability associated with eddies (the 'weather' of the ocean) had a large impact on overall ecosystem fluxes, such as the total growth or 'primary production' by phytoplankton (microscopic photosynthetic algae) or the total grazing by zooplankton (microscopic herbivores). Whether these fluxes increased or decreased with the removal of spatial variability depended on the overall abundance of critical nutrients for phytoplankton growth, such as nitrogen. The main conclusion is that low-resolution models, such as those used for running Intergovernmental Panel on Climate Change (IPCC) simulations to predict global climatic changes, are liable to make poor predictions of ecosystem fluxes because they do not account for the variability imposed by ocean weather.
To address this problem, one might propose using only models which explicitly resolve ocean weather. However, these tend to be very expensive in terms of computation and lead to practical restrictions on the spatial and temporal coverage of the simulation. We have tested an alternative modelling solution (objective 2). Instead of dividing space into ever-finer meshes of 'model grid cells', the coarse mesh is preserved and temporal variations are computed for both the mean fields and the sub-grid spatial covariances. This SMC approximation accurately reproduced the spatial-mean concentrations and fluxes of the high-resolution simulation at roughly 1 / 10 000th of the computational cost. However, this approximation was less successful in cases where the spatial variability was generated by variations in mixed layer depth. A related 'conditional moment closure' (CMC) approximation, whereby the coarse-resolution cell is partitioned according to values of a forcing variable (such as mixed layer depth), was shown to give more robust performance, with a similar cost to that of the SMC scheme. We conclude that the SMC / CMC schemes offer a practical means of accounting for ocean weather (or 'mesoscale variability') in models used for assessing the response of marine ecosystems to global change (see Wallhead et al., in press for full details). This work should therefore interest climate modellers and policy makers who rely on the projections of low-resolution models to assess e.g. how much of future anthropogenic carbon emissions are likely to be absorbed by the ocean, or how climate change-induced changes in ocean circulation are likely to affect marine plankton abundance and hence fisheries productivity.
To achieve objective 3, we gathered data from the Bermuda Atlantic time series (BATS) in order to demonstrate the benefits of SMC / CMC schemes using observational data. However, a preliminary study using simultaneous BATS point measurements suggested that our simple chlorophyll-based primary productivity models were inadequate to explain the BATS data. We therefore engaged in a statistical study, in collaboration with BATS observationalists, to extend and analyse group-resolved phytoplankton carbon data, as a necessary precursor to a group-resolved primary productivity model on which an effective SMC / CMC scheme could be based. We discovered that the group-resolved carbon biomass could be quite accurately inferred from measurements of pigments, irradiance and other 'core' variables using transform-linear multiple regression models. These allowed us to reconstruct the biomass over the years 1989 - 2012 using the raw data from 2004 - 2012, both in terms of total carbon and group fractions. We then developed a time series model to assess the magnitude of long-term trends in the presence of fixed seasonal effects and stochastic variability. Analysis revealed decadal shifts in the phytoplankton community composition associated with increasing trends in nutrient supply and zooplankton grazing pressure on larger eukaryotes. This study, currently being written up for submission to 'Global Biogeochemical Cycles' (Wallhead et al., in preparation), demonstrates methods of extending group-resolved data sets, and of testing for long-term trends in such data. It thus provides further analytic tools to help researchers and policy-makers to assess the long-term role of marine plankton in the biosphere.
References
Wallhead, P. J., Garçon V. C., Martin, A. P. Efficient Upscaling of Ocean Biogeochemistry. Ocean Modelling, in press. doi: 10.1016/j.ocemod.2012.12.002
Wallhead, P. J., Garçon V. C., Casey J. R., Lomas, M.W. Long-term Variability of Phytoplankton Biomass in the Sargasso Sea, in preparation.
1. to investigate the effects of spatial variability in plankton concentrations on fluxes of carbon and nutrients in marine ecosystems;
2. to assess how well these effects may be captured by a mathematical approximation known as Second moment closure (SMC);
3. to validate such approximations using observational data.
We addressed objective 1 using a high-resolution ocean model, i.e. a detailed computer simulation like those used for predicting the weather. The computed carbon and nutrient fluxes were compared with those computed in control simulations with the spatial variability artificially removed. We found that spatial variability associated with eddies (the 'weather' of the ocean) had a large impact on overall ecosystem fluxes, such as the total growth or 'primary production' by phytoplankton (microscopic photosynthetic algae) or the total grazing by zooplankton (microscopic herbivores). Whether these fluxes increased or decreased with the removal of spatial variability depended on the overall abundance of critical nutrients for phytoplankton growth, such as nitrogen. The main conclusion is that low-resolution models, such as those used for running Intergovernmental Panel on Climate Change (IPCC) simulations to predict global climatic changes, are liable to make poor predictions of ecosystem fluxes because they do not account for the variability imposed by ocean weather.
To address this problem, one might propose using only models which explicitly resolve ocean weather. However, these tend to be very expensive in terms of computation and lead to practical restrictions on the spatial and temporal coverage of the simulation. We have tested an alternative modelling solution (objective 2). Instead of dividing space into ever-finer meshes of 'model grid cells', the coarse mesh is preserved and temporal variations are computed for both the mean fields and the sub-grid spatial covariances. This SMC approximation accurately reproduced the spatial-mean concentrations and fluxes of the high-resolution simulation at roughly 1 / 10 000th of the computational cost. However, this approximation was less successful in cases where the spatial variability was generated by variations in mixed layer depth. A related 'conditional moment closure' (CMC) approximation, whereby the coarse-resolution cell is partitioned according to values of a forcing variable (such as mixed layer depth), was shown to give more robust performance, with a similar cost to that of the SMC scheme. We conclude that the SMC / CMC schemes offer a practical means of accounting for ocean weather (or 'mesoscale variability') in models used for assessing the response of marine ecosystems to global change (see Wallhead et al., in press for full details). This work should therefore interest climate modellers and policy makers who rely on the projections of low-resolution models to assess e.g. how much of future anthropogenic carbon emissions are likely to be absorbed by the ocean, or how climate change-induced changes in ocean circulation are likely to affect marine plankton abundance and hence fisheries productivity.
To achieve objective 3, we gathered data from the Bermuda Atlantic time series (BATS) in order to demonstrate the benefits of SMC / CMC schemes using observational data. However, a preliminary study using simultaneous BATS point measurements suggested that our simple chlorophyll-based primary productivity models were inadequate to explain the BATS data. We therefore engaged in a statistical study, in collaboration with BATS observationalists, to extend and analyse group-resolved phytoplankton carbon data, as a necessary precursor to a group-resolved primary productivity model on which an effective SMC / CMC scheme could be based. We discovered that the group-resolved carbon biomass could be quite accurately inferred from measurements of pigments, irradiance and other 'core' variables using transform-linear multiple regression models. These allowed us to reconstruct the biomass over the years 1989 - 2012 using the raw data from 2004 - 2012, both in terms of total carbon and group fractions. We then developed a time series model to assess the magnitude of long-term trends in the presence of fixed seasonal effects and stochastic variability. Analysis revealed decadal shifts in the phytoplankton community composition associated with increasing trends in nutrient supply and zooplankton grazing pressure on larger eukaryotes. This study, currently being written up for submission to 'Global Biogeochemical Cycles' (Wallhead et al., in preparation), demonstrates methods of extending group-resolved data sets, and of testing for long-term trends in such data. It thus provides further analytic tools to help researchers and policy-makers to assess the long-term role of marine plankton in the biosphere.
References
Wallhead, P. J., Garçon V. C., Martin, A. P. Efficient Upscaling of Ocean Biogeochemistry. Ocean Modelling, in press. doi: 10.1016/j.ocemod.2012.12.002
Wallhead, P. J., Garçon V. C., Casey J. R., Lomas, M.W. Long-term Variability of Phytoplankton Biomass in the Sargasso Sea, in preparation.