152
IAN ROBINSON
a near-surface phytoplankton bloom on the colour of the sea. Other
signatures took many scientists by surprise when they were first discovered
in the satellite images.
For example internal waves, a dynamical
phenomenon centred tens of metres below the sea surface, can sometimes be
revealed in exquisite spatial detail in the images of synthetic aperture radar
(SAR), because of their surface roughness signature.
In order to extract quantitative information about an ocean phenomenon
from satellite data, we need to understand the physical processes in the upper
ocean that control its surface signature in one of the primary detectable
variables.
Several of the derived properties, such as chlorophyll
concentration retrieved from colour sensors, surface wind speed from
scatterometers, wave height from altimetry, wave spectra from SARs and
salinity from microwave radiometry are now being used, or proposed, for
ingestion into ocean models. Figure 3 summarises the different classes and
types of sensors, the primary variables which they detect, and the way in
which they can supply inputs to ocean models.
For many of the applications to ocean models noted above, it is possible
to use ocean data products already produced by the agencies responsible for
the sensors, without the user having to engage themselves in any of the
processing tasks. Nonetheless, it is important for users to be aware of the
calibrations, corrections, analyses and resampling that are applied to data
products before they are distributed, since these processes have impacts on
the quality, accuracy and timeliness of the data. Figure 4 summarises them,
and also indicates what is meant by the different “levels” of data products
that may be available. Robinson (2004) provides a detailed explanation of
what is involved in each of these processes.
2.3
The sampling constraints imposed by satellite orbits
The use of Earth orbiting satellites as platforms for ocean-viewing
sensors offers a number of unique advantages such as the opportunity to
achieve wide synoptic coverage at fine spatial detail, and repeated regular
sampling to produce time series several years long. However, these benefits
are won at the cost of being tied to the unavoidable constraints imposed by
the physical laws of satellite orbital dynamics.
There are just two basic types of orbit useful for ocean remote sensing,
geostationary and near-polar. The geostationary orbit, at a height of about
36000 km, has a period of one sidereal day (~23.93 hr). Placed over the
Equator, the satellite flies West to East at the same rate as the Earth’s
rotation, so it always remains fixed relative to the ground, allowing it to
sample at any frequency. Being fixed it can view only that part of the world
within its horizon, which is a circle of about 7000 km radius centred on the
Equator at the longitude of the satellite. Its great height also makes it
difficult for sensors to achieve fine spatial resolution.
IAN ROBINSON
a near-surface phytoplankton bloom on the colour of the sea. Other
signatures took many scientists by surprise when they were first discovered
in the satellite images.
For example internal waves, a dynamical
phenomenon centred tens of metres below the sea surface, can sometimes be
revealed in exquisite spatial detail in the images of synthetic aperture radar
(SAR), because of their surface roughness signature.
In order to extract quantitative information about an ocean phenomenon
from satellite data, we need to understand the physical processes in the upper
ocean that control its surface signature in one of the primary detectable
variables.
Several of the derived properties, such as chlorophyll
concentration retrieved from colour sensors, surface wind speed from
scatterometers, wave height from altimetry, wave spectra from SARs and
salinity from microwave radiometry are now being used, or proposed, for
ingestion into ocean models. Figure 3 summarises the different classes and
types of sensors, the primary variables which they detect, and the way in
which they can supply inputs to ocean models.
For many of the applications to ocean models noted above, it is possible
to use ocean data products already produced by the agencies responsible for
the sensors, without the user having to engage themselves in any of the
processing tasks. Nonetheless, it is important for users to be aware of the
calibrations, corrections, analyses and resampling that are applied to data
products before they are distributed, since these processes have impacts on
the quality, accuracy and timeliness of the data. Figure 4 summarises them,
and also indicates what is meant by the different “levels” of data products
that may be available. Robinson (2004) provides a detailed explanation of
what is involved in each of these processes.
2.3
The sampling constraints imposed by satellite orbits
The use of Earth orbiting satellites as platforms for ocean-viewing
sensors offers a number of unique advantages such as the opportunity to
achieve wide synoptic coverage at fine spatial detail, and repeated regular
sampling to produce time series several years long. However, these benefits
are won at the cost of being tied to the unavoidable constraints imposed by
the physical laws of satellite orbital dynamics.
There are just two basic types of orbit useful for ocean remote sensing,
geostationary and near-polar. The geostationary orbit, at a height of about
36000 km, has a period of one sidereal day (~23.93 hr). Placed over the
Equator, the satellite flies West to East at the same rate as the Earth’s
rotation, so it always remains fixed relative to the ground, allowing it to
sample at any frequency. Being fixed it can view only that part of the world
within its horizon, which is a circle of about 7000 km radius centred on the
Equator at the longitude of the satellite. Its great height also makes it
difficult for sensors to achieve fine spatial resolution.