SATELLITE MEASUREMENTS
173
Microwave radiometers cannot be used within about 100 km of the coast
because of the side-lobe contamination of microwave sources on land
leaking into the antenna reception. This, with their low spatial resolution,
severely limits their usefulness in coastal and shelf seas.
3.3.4
The character of the ocean surface thermal structure
The ocean modeller requiring measurements of SST, for assimilation or
to validate the temperatures in the top layer of an ocean GCM, may conclude
that their task is greatly simplified by the wide choice of different types of
observations of SST now available from both in situ and satellite platforms.
However, there is a pitfall for the unwary user of SST data, arising from the
detailed character of the thermal structure in the top few metres of the ocean.
Two distinct factors create near-surface vertical temperature gradients.
Firstly on sunny calm days a diurnal thermocline tends to develop above
which a top layer is found, a metre or so thick and up to about 1 K warmer
than below (although exceptionally it can be several K warmer). At night
the warm layer collapses. Secondly (and independently of the first effect)
the top skin layer of the sea, a fraction of a millimetre thick, tends to be a
few tenths of a Kelvin cooler than the water immediately below. Both these
effects, and especially the first, may be horizontally variable, leading to
spatial patchiness of SST.
Neither of these processes is normally represented in the physics of ocean
models for which the topmost layer typically corresponds to the upper mixed
layer assumed to be uniform above the seasonal thermocline. The different
types of measurement of SST also sample at different levels of the nearsurface thermal structure. In other words the definition of “SST” is different
for the thermometer on a buoy’s hull, for a sensor in a ship’s cooling water
intake, for an infrared radiometer, for a microwave radiometer, and for an
ocean model. These differences are important when accuracies of a few
tenths of a Kelvin are required. They may also vary considerably during the
day so that a single daily measurement used may be aliased depending on
the time in the diurnal cycle at which it is sampled. It is therefore necessary
to harmonise SST data from different sources before they are introduced to
an ocean model. This is one of the issues discussed in section 4 of this
paper.
It is certainly worth taking the trouble to resolve these issues because
SST observations can provide a very useful constraint on models. Surface
ocean dynamical features often have thermal signatures. Major ocean
currents are normally associated with thermal fronts. Ocean eddies are often
visible in satellite SST images. Thus the assimilation of SST should in
principle help to constrain the modelled evolution of mesoscale variability.
In the case of coupled ocean-atmosphere models the interface temperature
gains even more importance for constraining the model. In this case
173
Microwave radiometers cannot be used within about 100 km of the coast
because of the side-lobe contamination of microwave sources on land
leaking into the antenna reception. This, with their low spatial resolution,
severely limits their usefulness in coastal and shelf seas.
3.3.4
The character of the ocean surface thermal structure
The ocean modeller requiring measurements of SST, for assimilation or
to validate the temperatures in the top layer of an ocean GCM, may conclude
that their task is greatly simplified by the wide choice of different types of
observations of SST now available from both in situ and satellite platforms.
However, there is a pitfall for the unwary user of SST data, arising from the
detailed character of the thermal structure in the top few metres of the ocean.
Two distinct factors create near-surface vertical temperature gradients.
Firstly on sunny calm days a diurnal thermocline tends to develop above
which a top layer is found, a metre or so thick and up to about 1 K warmer
than below (although exceptionally it can be several K warmer). At night
the warm layer collapses. Secondly (and independently of the first effect)
the top skin layer of the sea, a fraction of a millimetre thick, tends to be a
few tenths of a Kelvin cooler than the water immediately below. Both these
effects, and especially the first, may be horizontally variable, leading to
spatial patchiness of SST.
Neither of these processes is normally represented in the physics of ocean
models for which the topmost layer typically corresponds to the upper mixed
layer assumed to be uniform above the seasonal thermocline. The different
types of measurement of SST also sample at different levels of the nearsurface thermal structure. In other words the definition of “SST” is different
for the thermometer on a buoy’s hull, for a sensor in a ship’s cooling water
intake, for an infrared radiometer, for a microwave radiometer, and for an
ocean model. These differences are important when accuracies of a few
tenths of a Kelvin are required. They may also vary considerably during the
day so that a single daily measurement used may be aliased depending on
the time in the diurnal cycle at which it is sampled. It is therefore necessary
to harmonise SST data from different sources before they are introduced to
an ocean model. This is one of the issues discussed in section 4 of this
paper.
It is certainly worth taking the trouble to resolve these issues because
SST observations can provide a very useful constraint on models. Surface
ocean dynamical features often have thermal signatures. Major ocean
currents are normally associated with thermal fronts. Ocean eddies are often
visible in satellite SST images. Thus the assimilation of SST should in
principle help to constrain the modelled evolution of mesoscale variability.
In the case of coupled ocean-atmosphere models the interface temperature
gains even more importance for constraining the model. In this case