The approaches described above all suffer from their reliance on satellite Chl as
an indicator of biomass. As a result, they interpret higher Chl as more biomass, and
by inference, a greater contribution from larger size classes of phytoplankton. One
solution to this problem in the context of remote sensing is to build on the body of
work which employs satellite estimates of particulate backscattering (b bp ) to
quantify phytoplankton carbon directly (Behrenfeld et al. 2005; Westberry et al.
2008; Kostadinov et al. 2009). The global relationship of Westberry et al. (2008)
relating particulate backscattering to C phyto can be improved through (1) routine
field measurements of C phyto (currently there are none) to better constrain the
relationship, (2) improvements in bio-optical inversion schemes that estimate b bp
from satellite radiance, and (3) algorithm development that accounts for anomalous
sources of b bp biasing estimates of C phyto (e.g., coccolithophores). Kostadinov et al.
(2009, 2010) recently introduced a remote sensing method for characterizing particle size distributions (PSD) based on spectral b bp retrievals. Resultant PSDs can be
expressed as a continuous function of size and related to specific phytoplankton size
ranges, such as pico-, nano-, and micro-phytoplankton. Given biovolume-specific
carbon concentrations, which are available from laboratory studies, this method
could yield class-specific carbon biomass explicitly. Perhaps an even more compelling avenue would be to combine the approaches of Uitz et al. (2006) and
Kostadinov et al. (2009) to partition both Chl and C phyto individually to characterize
size-class specific Chl:C phyto ratios. Any existing NPP model that accounts for
Chl:C phyto variability would surely benefit from this added information.
Another means of incorporating taxonomic information is through the use of
phytoplankton absorption, a ph , rather than Chl concentration. NPP models
employing Chl implicitly assume a fixed Chl-specific absorption capacity, a
Ã
ph ,
despite order of magnitude variability that exists in a
Ã
ph (Bricaud et al. 1995, 1998).
Much of this variability can be related to the size distribution of extant phytoplankton and presumably taxonomic composition (Bricaud et al. 2004). Lee et al.
(1996) provide a NPP model amenable to remote sensing which is cast in terms of
a ph , rather than Chl. For the dataset these authors investigated, a ph -based NPP
models were far superior to Chl-based analogs. In addition to the conceptual
advantages of using a ph over Chl for estimating NPP, retrieval of a ph from satellite
reflectance has also been suggested to be preferable to direct estimation of biogeochemical quantities such as Chl (Lee et al. 2002).
8.6.4 New Tools
The various avenues discussed above for improving global assessments of ocean
NPP are largely focused on advances that can be made with currently available
observational and model-derived data sets. Major future advancements, however,
may also be realized through engineering developments, both in the ocean and in
space. Autonomously collected data by free-drifting and profiling floats and gliders
220
T. K. Westberry and M. J. Behrenfeld
an indicator of biomass. As a result, they interpret higher Chl as more biomass, and
by inference, a greater contribution from larger size classes of phytoplankton. One
solution to this problem in the context of remote sensing is to build on the body of
work which employs satellite estimates of particulate backscattering (b bp ) to
quantify phytoplankton carbon directly (Behrenfeld et al. 2005; Westberry et al.
2008; Kostadinov et al. 2009). The global relationship of Westberry et al. (2008)
relating particulate backscattering to C phyto can be improved through (1) routine
field measurements of C phyto (currently there are none) to better constrain the
relationship, (2) improvements in bio-optical inversion schemes that estimate b bp
from satellite radiance, and (3) algorithm development that accounts for anomalous
sources of b bp biasing estimates of C phyto (e.g., coccolithophores). Kostadinov et al.
(2009, 2010) recently introduced a remote sensing method for characterizing particle size distributions (PSD) based on spectral b bp retrievals. Resultant PSDs can be
expressed as a continuous function of size and related to specific phytoplankton size
ranges, such as pico-, nano-, and micro-phytoplankton. Given biovolume-specific
carbon concentrations, which are available from laboratory studies, this method
could yield class-specific carbon biomass explicitly. Perhaps an even more compelling avenue would be to combine the approaches of Uitz et al. (2006) and
Kostadinov et al. (2009) to partition both Chl and C phyto individually to characterize
size-class specific Chl:C phyto ratios. Any existing NPP model that accounts for
Chl:C phyto variability would surely benefit from this added information.
Another means of incorporating taxonomic information is through the use of
phytoplankton absorption, a ph , rather than Chl concentration. NPP models
employing Chl implicitly assume a fixed Chl-specific absorption capacity, a
Ã
ph ,
despite order of magnitude variability that exists in a
Ã
ph (Bricaud et al. 1995, 1998).
Much of this variability can be related to the size distribution of extant phytoplankton and presumably taxonomic composition (Bricaud et al. 2004). Lee et al.
(1996) provide a NPP model amenable to remote sensing which is cast in terms of
a ph , rather than Chl. For the dataset these authors investigated, a ph -based NPP
models were far superior to Chl-based analogs. In addition to the conceptual
advantages of using a ph over Chl for estimating NPP, retrieval of a ph from satellite
reflectance has also been suggested to be preferable to direct estimation of biogeochemical quantities such as Chl (Lee et al. 2002).
8.6.4 New Tools
The various avenues discussed above for improving global assessments of ocean
NPP are largely focused on advances that can be made with currently available
observational and model-derived data sets. Major future advancements, however,
may also be realized through engineering developments, both in the ocean and in
space. Autonomously collected data by free-drifting and profiling floats and gliders
220
T. K. Westberry and M. J. Behrenfeld