8.6.5 Vision for Future Remote Sensing of NPP
The preceding subsections outlined various avenues for advancing space-based
NPP models. Here, an example is given which employs some of these pieces and
allows a glimpse of how the distribution of NPP and our understanding may differ
when taken into consideration. This new approach is termed the Carbon,
Absorption, and Fluorescence Euphotic-resolving (CAFE) NPP model. For this
exercise, the VGPM is used as a prototypical satellite NPP model, and its annual
average NPP rate is shown in Fig. 8.4. In contrast, the CAFE NPP model
assimilates new satellite-derived information into its estimation of NPP rates. First,
the model employs satellite-derived estimates of particulate backscattering that are
used to estimate phytoplankton carbon biomass (C) directly. This parameter also
allows estimation of Chl:C which provides physiological information (i.e.,
photoacclimation) and a link to the phytoplankton growth rate, l (Laws and
Bannister 1980). Thus, we can estimate NPP directly using Eq. 8.1. Using this
approach, the model is able to distinguish physiological changes in cellular pigmentation from changes in biomass. The result is that many high Chl regions (i.e.,
North Atlantic) have reduced NPP as some fraction of the bulk Chl is attributed to
photoacclimation. In contrast, many low Chl regions (i.e., North Pacific Subtropical Gyre) exhibit increased NPP relative to the VGPM as their biomass (and
Chl) may be low, but their growth rates can still be high. Second, chlorophyll
fluorescence from satellite has been shown to register the unique imprint of iron
stress over much of the ocean (Behrenfeld et al. 2009; Westberry et al. 2013). The
reason for this, in part, results from chlorophyll present in phytoplankton and
reflected in satellite-based Chl retrievals, but which is dissociated from photosynthetic electron transport (Behrenfeld and Milligan 2013). Therefore, NPP
models which employ Chl as a biomass indicator will tend to overestimate NPP
where phytoplankton are iron stressed. Satellite estimates of chlorophyll fluorescence efficiency (u f , Behrenfeld et al. 2009) can be used to correct for this effect
(Fig. 8.4). Here, a simple linear correction is applied that assumes the strength of
iron limitation is directly proportional to u f above some threshold value that marks
the onset of iron limitation. The effects are largely irrelevant outside the equatorial
oceans, but can reduce NPP by up to 40 % in some places. This is consistent with
the observation that up to 40 % of the total Chl content in iron stressed cells can be
in a ‘dissociated’ state (Moseley et al. 2002). In the example given, this correction
alone decreases global annual, marine NPP by[3 Pg year
-1 , nearly all of which is
in the tropics between 20°N and 20°S. Third, the CbPM can be recast in terms of
phytoplankton absorption (a ph ) per unit carbon rather than Chl:C. This approach
has the benefit of accounting for all accessory pigments which can play an
important role in light absorption and photosynthesis. Absorption-based NPP
modeling has shown superior predictive ability in some field datasets (Lee et al.
1996). In addition, this approach should also reduce uncertainties arising from
empirical retrievals of Chl, as phytoplankton absorption is more closely tied to the
fundamental satellite measurements of radiance.
222
T. K. Westberry and M. J. Behrenfeld
The preceding subsections outlined various avenues for advancing space-based
NPP models. Here, an example is given which employs some of these pieces and
allows a glimpse of how the distribution of NPP and our understanding may differ
when taken into consideration. This new approach is termed the Carbon,
Absorption, and Fluorescence Euphotic-resolving (CAFE) NPP model. For this
exercise, the VGPM is used as a prototypical satellite NPP model, and its annual
average NPP rate is shown in Fig. 8.4. In contrast, the CAFE NPP model
assimilates new satellite-derived information into its estimation of NPP rates. First,
the model employs satellite-derived estimates of particulate backscattering that are
used to estimate phytoplankton carbon biomass (C) directly. This parameter also
allows estimation of Chl:C which provides physiological information (i.e.,
photoacclimation) and a link to the phytoplankton growth rate, l (Laws and
Bannister 1980). Thus, we can estimate NPP directly using Eq. 8.1. Using this
approach, the model is able to distinguish physiological changes in cellular pigmentation from changes in biomass. The result is that many high Chl regions (i.e.,
North Atlantic) have reduced NPP as some fraction of the bulk Chl is attributed to
photoacclimation. In contrast, many low Chl regions (i.e., North Pacific Subtropical Gyre) exhibit increased NPP relative to the VGPM as their biomass (and
Chl) may be low, but their growth rates can still be high. Second, chlorophyll
fluorescence from satellite has been shown to register the unique imprint of iron
stress over much of the ocean (Behrenfeld et al. 2009; Westberry et al. 2013). The
reason for this, in part, results from chlorophyll present in phytoplankton and
reflected in satellite-based Chl retrievals, but which is dissociated from photosynthetic electron transport (Behrenfeld and Milligan 2013). Therefore, NPP
models which employ Chl as a biomass indicator will tend to overestimate NPP
where phytoplankton are iron stressed. Satellite estimates of chlorophyll fluorescence efficiency (u f , Behrenfeld et al. 2009) can be used to correct for this effect
(Fig. 8.4). Here, a simple linear correction is applied that assumes the strength of
iron limitation is directly proportional to u f above some threshold value that marks
the onset of iron limitation. The effects are largely irrelevant outside the equatorial
oceans, but can reduce NPP by up to 40 % in some places. This is consistent with
the observation that up to 40 % of the total Chl content in iron stressed cells can be
in a ‘dissociated’ state (Moseley et al. 2002). In the example given, this correction
alone decreases global annual, marine NPP by[3 Pg year
-1 , nearly all of which is
in the tropics between 20°N and 20°S. Third, the CbPM can be recast in terms of
phytoplankton absorption (a ph ) per unit carbon rather than Chl:C. This approach
has the benefit of accounting for all accessory pigments which can play an
important role in light absorption and photosynthesis. Absorption-based NPP
modeling has shown superior predictive ability in some field datasets (Lee et al.
1996). In addition, this approach should also reduce uncertainties arising from
empirical retrievals of Chl, as phytoplankton absorption is more closely tied to the
fundamental satellite measurements of radiance.
222
T. K. Westberry and M. J. Behrenfeld