413
temperature of the thermocline is justified in long scale climate experiments if episodes with
time scales less than a few decades are to be filtered out.
Short time predictability of the atmospheric weather is simply a reflection of the fact that
the atmosphere stratification has a very short turnover time. On the contrary, turnover times for
the ocean waters are acknowledged to be of the order of hundreds or thousands of years: A
well known case is the Black Sea. Thus the stratification of the water column keeps a record of
climate fluctuations, because the cold water generated in cold climate episodes sinks down
rather quickly to the bottom, but during hot episodes warm surface waters can not transfer heat
to the bottom.
The above discussion shows that SST is the main boundary constraint in atmospheric
GCM's. In other words, it can be shown that the air temperature is forced by and follows the
slower oscillations of the SST, thus providing an argument for the implementation of "marine
only" climate models (Hughes and Weaver, 1994; Weaver and Hughes, 1994).
The elimination of atmospheric dynamics in climate models could only be achieved if
the atmosphere would prove to be an ergodic system. Ergodicity of the atmosphere is an open
problem (Feistel,I993; Nagai et at., 1992) and it is only in the last years that large-scale
experiments are starting their way. Theoretical modelling is still far from today's state of the
art.
In the Canary Islands several studies are in due course to identify the local mechanisms
of heat storage and transport in the Eastern-Central Atlantic Ocean area (Pe1egrf et at., 1995).
The conclusions show quite surprisingly that there is a seasonal heat accumulation well under
the sea surface, a fact that could be interpreted as a signal of a slow manifold. On the other
hand, the Canaries water column has been systematically explored in several oceanographic
campaigns in the years 1957, 1981 and 1992, but no climate studies have been carried on with
this data base.
3. On Climate Modelling.
It was pointed out in the Introduction that climate modelling has a predominantly
diagnostic character. Nevertheless, prognostic studies can also be envisaged with the aim of
making short and medium term forecasts, in order to assess the expected tendencies: This is but
one way of understanding the structure of climatic changes.
A global climatic model expresses some kind of energy budget. If an average
temperature u(x,t) is employed as the main descriptor, the already classical Budyko-Sellers
energy-balance model (with X= sinrp ,rp = latitude) for the closed interval [-1, 1] contaIns the·
nonlinear parabolic equation:
au
a [
au aU]
at - ax D(x, ax) ax = Ra-Re
and appropriate free boundary conditions.
More local models, on the other hand, tend to be more complicated. It is now time to
recall that the main features in the dynamics of geophysical fluids are captured in the so-called
Navier Stokes system: First of all, a set of equations with the same structure is considered.
These are models for the time-space evolution of several different magnitudes X , either vector
or scalar ones. Among them there are the flux components (wind, marine currents),
temperatures, water vapor and salt concentrations (atmosphere or sea), various pollutant
temperature of the thermocline is justified in long scale climate experiments if episodes with
time scales less than a few decades are to be filtered out.
Short time predictability of the atmospheric weather is simply a reflection of the fact that
the atmosphere stratification has a very short turnover time. On the contrary, turnover times for
the ocean waters are acknowledged to be of the order of hundreds or thousands of years: A
well known case is the Black Sea. Thus the stratification of the water column keeps a record of
climate fluctuations, because the cold water generated in cold climate episodes sinks down
rather quickly to the bottom, but during hot episodes warm surface waters can not transfer heat
to the bottom.
The above discussion shows that SST is the main boundary constraint in atmospheric
GCM's. In other words, it can be shown that the air temperature is forced by and follows the
slower oscillations of the SST, thus providing an argument for the implementation of "marine
only" climate models (Hughes and Weaver, 1994; Weaver and Hughes, 1994).
The elimination of atmospheric dynamics in climate models could only be achieved if
the atmosphere would prove to be an ergodic system. Ergodicity of the atmosphere is an open
problem (Feistel,I993; Nagai et at., 1992) and it is only in the last years that large-scale
experiments are starting their way. Theoretical modelling is still far from today's state of the
art.
In the Canary Islands several studies are in due course to identify the local mechanisms
of heat storage and transport in the Eastern-Central Atlantic Ocean area (Pe1egrf et at., 1995).
The conclusions show quite surprisingly that there is a seasonal heat accumulation well under
the sea surface, a fact that could be interpreted as a signal of a slow manifold. On the other
hand, the Canaries water column has been systematically explored in several oceanographic
campaigns in the years 1957, 1981 and 1992, but no climate studies have been carried on with
this data base.
3. On Climate Modelling.
It was pointed out in the Introduction that climate modelling has a predominantly
diagnostic character. Nevertheless, prognostic studies can also be envisaged with the aim of
making short and medium term forecasts, in order to assess the expected tendencies: This is but
one way of understanding the structure of climatic changes.
A global climatic model expresses some kind of energy budget. If an average
temperature u(x,t) is employed as the main descriptor, the already classical Budyko-Sellers
energy-balance model (with X= sinrp ,rp = latitude) for the closed interval [-1, 1] contaIns the·
nonlinear parabolic equation:
au
a [
au aU]
at - ax D(x, ax) ax = Ra-Re
and appropriate free boundary conditions.
More local models, on the other hand, tend to be more complicated. It is now time to
recall that the main features in the dynamics of geophysical fluids are captured in the so-called
Navier Stokes system: First of all, a set of equations with the same structure is considered.
These are models for the time-space evolution of several different magnitudes X , either vector
or scalar ones. Among them there are the flux components (wind, marine currents),
temperatures, water vapor and salt concentrations (atmosphere or sea), various pollutant