Ways of Testing Static Province Boundaries in the Open Ocean
111
a unimodal set of data to represent a province by a minor adjustment of its boundaries.
Obviously, if the historical record of oceanographic expeditions and experiments had
been different, the final solution would also have been somewhat different.
Using this process, a set of unimodal distributions of parameter values describing
the chlorophyll profiles was obtained that represented each case (province and season)
and could be used to check that the boundaries between provinces were realistic. This
was done by investigation of the differences between seasonal mean values (for each
parameter) across the boundaries between each pair of adjacent provinces using analysis of
variance and the Bonferroni-Dunn post hoc test. This investigation showed that measures
of chlorophyll biomass and depth of its maximum value (Z p , B(0), B int , and Z m ) differ
between adjacent provinces more significantly than the parameters describing the form
of the chlorophyll peak itself (s, h, and R), which show weaker regional differentiation;
this is evident for both annual and seasonal means for each province.
For all parameters, the most significant boundaries were found to be between provinces
of the coastal and oceanic biomes (Sathyendranath et al., 1995), whereas within the
oceanic provinces the most distinct boundaries lie to the north and south of the subtropical gyres and thus between the provinces of the westerlies and trades biomes. Most
boundaries between oceanic provinces proved to have some significance for the more
variable measures (Z p and Z m ), though even these do not discriminate convincingly
between boundaries within the most northerly and southerly groups of provinces. These
same parameters confirm the reality of the east-west partition of both the subtropical
gyre and the tropical regions of the Atlantic.
Analytical Tests
Such a test was performed by Kawamiya, Kishi, and Suginohara (2000), who embedded an
ecosystem model in a general circulation model of the North Pacific Ocean. The output
represented chlorophyll biomass, obtained by interaction between the ecosystem model
and the relevant physics of the general circulation model. The data used for the ecological
partition of the North Pacific (between 60
N and 10
S) represented the monthly average
integrated 0- to 150-m chlorophyll (strictly speaking, the “phytoplankton” component
in the model) obtained in the sixth year of model integration, when stability had been
achieved. Partition of the data set was obtained by empirical orthogonal function (EOF)
analysis. EOF 1 showed that the amplitude of seasonal variation of chlorophyll was higher
at middle and high latitudes with maxima in April, whereas the peaks in EOF 2 are
located in regions of deep mixed layers in spring. Based on the EOF analysis and annual
mean abundance, boundaries may be located by fronts either in the EOF modes or in
the annual mean abundance. This analysis achieves a delimitation and definition of seven
provinces, for each of which a seasonal cycle of phytoplankton abundance is available.
The authors find a “remarkable coincidence” between this result and the arrangement
of North Pacific provinces proposed here, several being directly comparable in area and
in their seasonal chlorophyll cycle. Several others proposed by them equate, as they
remark, with adjacent pairs of the provinces reviewed in this book. The provinces whose
location and properties are thus confirmed are BERS, PSAG(E) + PSAG(W), KURO,
NPST(W) NPST(E) + NPTG, PNEC + PEQD. Further, by direct comparison with sea
surface chlorophyll obtained from several satellite-borne sensors, Kawamiya and his
coauthors subjectively confirm that their own proposals are valid—thus closing, as it
were, the circle.
Another such test, using the observations made along the repeated Atlantic Meridional
Transect (AMT 1–7, 1995–1998) from the UK to Antarctica, was made by Hooker
et al. (2000). These authors devised an objective methodology for identifying oceanic
provinces from multiparameter data taken along the meridional cruise tracks that covered
111
a unimodal set of data to represent a province by a minor adjustment of its boundaries.
Obviously, if the historical record of oceanographic expeditions and experiments had
been different, the final solution would also have been somewhat different.
Using this process, a set of unimodal distributions of parameter values describing
the chlorophyll profiles was obtained that represented each case (province and season)
and could be used to check that the boundaries between provinces were realistic. This
was done by investigation of the differences between seasonal mean values (for each
parameter) across the boundaries between each pair of adjacent provinces using analysis of
variance and the Bonferroni-Dunn post hoc test. This investigation showed that measures
of chlorophyll biomass and depth of its maximum value (Z p , B(0), B int , and Z m ) differ
between adjacent provinces more significantly than the parameters describing the form
of the chlorophyll peak itself (s, h, and R), which show weaker regional differentiation;
this is evident for both annual and seasonal means for each province.
For all parameters, the most significant boundaries were found to be between provinces
of the coastal and oceanic biomes (Sathyendranath et al., 1995), whereas within the
oceanic provinces the most distinct boundaries lie to the north and south of the subtropical gyres and thus between the provinces of the westerlies and trades biomes. Most
boundaries between oceanic provinces proved to have some significance for the more
variable measures (Z p and Z m ), though even these do not discriminate convincingly
between boundaries within the most northerly and southerly groups of provinces. These
same parameters confirm the reality of the east-west partition of both the subtropical
gyre and the tropical regions of the Atlantic.
Analytical Tests
Such a test was performed by Kawamiya, Kishi, and Suginohara (2000), who embedded an
ecosystem model in a general circulation model of the North Pacific Ocean. The output
represented chlorophyll biomass, obtained by interaction between the ecosystem model
and the relevant physics of the general circulation model. The data used for the ecological
partition of the North Pacific (between 60
N and 10
S) represented the monthly average
integrated 0- to 150-m chlorophyll (strictly speaking, the “phytoplankton” component
in the model) obtained in the sixth year of model integration, when stability had been
achieved. Partition of the data set was obtained by empirical orthogonal function (EOF)
analysis. EOF 1 showed that the amplitude of seasonal variation of chlorophyll was higher
at middle and high latitudes with maxima in April, whereas the peaks in EOF 2 are
located in regions of deep mixed layers in spring. Based on the EOF analysis and annual
mean abundance, boundaries may be located by fronts either in the EOF modes or in
the annual mean abundance. This analysis achieves a delimitation and definition of seven
provinces, for each of which a seasonal cycle of phytoplankton abundance is available.
The authors find a “remarkable coincidence” between this result and the arrangement
of North Pacific provinces proposed here, several being directly comparable in area and
in their seasonal chlorophyll cycle. Several others proposed by them equate, as they
remark, with adjacent pairs of the provinces reviewed in this book. The provinces whose
location and properties are thus confirmed are BERS, PSAG(E) + PSAG(W), KURO,
NPST(W) NPST(E) + NPTG, PNEC + PEQD. Further, by direct comparison with sea
surface chlorophyll obtained from several satellite-borne sensors, Kawamiya and his
coauthors subjectively confirm that their own proposals are valid—thus closing, as it
were, the circle.
Another such test, using the observations made along the repeated Atlantic Meridional
Transect (AMT 1–7, 1995–1998) from the UK to Antarctica, was made by Hooker
et al. (2000). These authors devised an objective methodology for identifying oceanic
provinces from multiparameter data taken along the meridional cruise tracks that covered