Ecological Provinces in the Open Ocean
107
But, before we get to that, this problem raises another one that we must deal with here.
The partition of the ocean required by biogeochemists must be useful for the deployment
of models of (say) primary production of phytoplankton to analyze a field of (say)
satellite observations of chlorophyll. For such tasks, it may be more important to locate
the areas and periods over which a certain parameterization of a model is appropriate,
than to enable coherent regional statements about the scientific natural history of the
sea to be made. But it is really the latter that is my object in this book so that, despite
the use of the edge-adjustment routine “MakeShift” of George White (see Chapter 1),
I have somewhat distanced the provinces used here from the concept of the dynamic
biogeochemical province.
The partition of the Arabian Sea of Brock et al. (1998) done to support the deployment
of a production model illustrates how others have established provinces within which
to partition computations of ocean productivity. These authors merged climatologies of
incident light, mixed-layer depth, and chlorophyll so as to identify three fundamental
types of “biohydro-optical class” defined by the form of the chlorophyll profile: (i) Typical
Tropical Profile and (ii) Mixed Layer Bloom, plus (iii) a Transitional type between
the first two. These classes were discussed as if they could serve as the biogeochemical
provinces in a partition of this region even though, at different seasons, different profile
types might occur at any given location. The seasonal output from interaction between
the three climatologies, computed pixel by pixel across the whole Arabian Sea, indeed
showed a seasonal, meridional march of the three classes that was imposed by changing
irradiance and wind speeds.
This partition was later used to model nitrate-based production in the same region
that was partitioned by the use of a decision tree for water depth, water temperature,
and rate of primary production to allocate each pixel to one of six provinces (Watts et al.
1999); thus, the concept of a “province” was equated with the location of a particular
phase in a seasonal plankton calendar. This is very different from the concept of bounded
geographical entities that are modified by seasonal or between-year anomalies to their
average hydrographic situation. Such, of course, is the concept of an ecological province
defined and used in this book, and that I have described elsewhere; this concept can
perhaps best be explained by a description of how the partition was actually achieved.
So, what follows is an account of the steps that were taken toward defining a global set
of ecological provinces and their boundaries. Things are often tidier in the telling than
in the doing, and the following description implies a logical progression that is rather
different from what actually occurred: most of the activities described in this section were
undertaken more or less simultaneously, and it was only the testing of boundaries that
logically had to follow the other work, and did so. Furthermore, the work was performed
in the early 1990s by reference to CZCS data, prior to the availability of the much-superior
SeaWiFS and MODIS images, and prior to the general availability of sea-surface elevation
images at the basin scale. Modern data have been used to make minor modifications to,
and clarifications of, the boundaries originally proposed.
The first step was an examination of all available regional and seasonal images of the
surface CZCS chlorophyll field in a variety of formats for characteristic, observable, and
repetitive regional patterns, both spatial and temporal. Where necessary, the individual
scenes for critical regions were scanned to clarify the nature of blooms observed in the
monthly and seasonal composites. This subjective technique of interrogating the images
to locate boundaries was a proxy for the objective and numerical technique that would
be the method of choice.
The second step was to examine the regional oceanography of all parts of the ocean
not only by bibliographic search but also by consulting data archives. The physical
oceanographic literature for each ocean basin was reviewed extensively to compare the
seasonal and regional distribution of chlorophyll values found in the SeaWiFS images with
107
But, before we get to that, this problem raises another one that we must deal with here.
The partition of the ocean required by biogeochemists must be useful for the deployment
of models of (say) primary production of phytoplankton to analyze a field of (say)
satellite observations of chlorophyll. For such tasks, it may be more important to locate
the areas and periods over which a certain parameterization of a model is appropriate,
than to enable coherent regional statements about the scientific natural history of the
sea to be made. But it is really the latter that is my object in this book so that, despite
the use of the edge-adjustment routine “MakeShift” of George White (see Chapter 1),
I have somewhat distanced the provinces used here from the concept of the dynamic
biogeochemical province.
The partition of the Arabian Sea of Brock et al. (1998) done to support the deployment
of a production model illustrates how others have established provinces within which
to partition computations of ocean productivity. These authors merged climatologies of
incident light, mixed-layer depth, and chlorophyll so as to identify three fundamental
types of “biohydro-optical class” defined by the form of the chlorophyll profile: (i) Typical
Tropical Profile and (ii) Mixed Layer Bloom, plus (iii) a Transitional type between
the first two. These classes were discussed as if they could serve as the biogeochemical
provinces in a partition of this region even though, at different seasons, different profile
types might occur at any given location. The seasonal output from interaction between
the three climatologies, computed pixel by pixel across the whole Arabian Sea, indeed
showed a seasonal, meridional march of the three classes that was imposed by changing
irradiance and wind speeds.
This partition was later used to model nitrate-based production in the same region
that was partitioned by the use of a decision tree for water depth, water temperature,
and rate of primary production to allocate each pixel to one of six provinces (Watts et al.
1999); thus, the concept of a “province” was equated with the location of a particular
phase in a seasonal plankton calendar. This is very different from the concept of bounded
geographical entities that are modified by seasonal or between-year anomalies to their
average hydrographic situation. Such, of course, is the concept of an ecological province
defined and used in this book, and that I have described elsewhere; this concept can
perhaps best be explained by a description of how the partition was actually achieved.
So, what follows is an account of the steps that were taken toward defining a global set
of ecological provinces and their boundaries. Things are often tidier in the telling than
in the doing, and the following description implies a logical progression that is rather
different from what actually occurred: most of the activities described in this section were
undertaken more or less simultaneously, and it was only the testing of boundaries that
logically had to follow the other work, and did so. Furthermore, the work was performed
in the early 1990s by reference to CZCS data, prior to the availability of the much-superior
SeaWiFS and MODIS images, and prior to the general availability of sea-surface elevation
images at the basin scale. Modern data have been used to make minor modifications to,
and clarifications of, the boundaries originally proposed.
The first step was an examination of all available regional and seasonal images of the
surface CZCS chlorophyll field in a variety of formats for characteristic, observable, and
repetitive regional patterns, both spatial and temporal. Where necessary, the individual
scenes for critical regions were scanned to clarify the nature of blooms observed in the
monthly and seasonal composites. This subjective technique of interrogating the images
to locate boundaries was a proxy for the objective and numerical technique that would
be the method of choice.
The second step was to examine the regional oceanography of all parts of the ocean
not only by bibliographic search but also by consulting data archives. The physical
oceanographic literature for each ocean basin was reviewed extensively to compare the
seasonal and regional distribution of chlorophyll values found in the SeaWiFS images with