SURFACE FLUXES
267
7.
Discussion and conclusion
The uncertainties in air-sea fluxes are difficult to quantify, with the
lack of direct measurements a major problem, but they decrease as the
time and space scales expand. Best known are the global, long term
mean heat and freshwater fluxes, which ocean inventories show are near
zero. At the other extreme are turbulent fluxes whose sign at a point
can be uncertain on time scales less than about 10 minutes. Perhaps
most relevant to GODAE are hourly fluxes on a spatial scale of 10 km
or more. Some of these might be tractable from satellite measurements,
but whenever bulk aerodynamic formulae are involved there is at least
a factor of 2 uncertainty due to transfer coefficient variability on these
scales.
The temptation for GODAE to simply utilize global fields of Q, F
and wind stress should be resisted, because too much good physics is
lost. Notable examples include the penetration of solar radiation into
the upper ocean, the relationship between evaporation and the latent
heat flux, sea-ice coverage and the strong dependence of fluxes on SST.
However, the price is high because of the large number of forcing fields
necessary to prescribe the flux components.
Unfortunately, there is no single flux data suitable for GODAE. Instead, various data sets should be merged within an assimilation data
stream. However, a crucial procedure within the stream should be the
correction of most fields for known biases, as revealed by comparisons
with more limited, but more accurate and better understood observations. These adjustments should result in global heat and freshwater
fluxes that are nearly in balance, and these budgets should not be overly
upset by any increments made to the fluxes during the assimilation cycle. In addition the incremented fluxes ought to display the observed
features of global variability, especially those portrayed by large scale
climate indices.
The forcing of a truly global ocean data assimilation system that
incorporates all the details of the known physics would be extremely
complicated. Therefore, simplifications should be sought, but implemented only if well understood and demonstrated to be tolerable from
the viewpoint of assimilation products. A prime example is high latitude
forcing where representation of much of the physics requires a coupled
sea-ice model. Another is forcing near coastlines, where some satellite
products are contaminated by the land and continental runoff is complicated. Also, some fluxes may not need to be accounted for; such as
the temperature of precipitation and runoff, the latent heat of fusion of
glacier and ice-sheet runoff, and the melting of snow.
267
7.
Discussion and conclusion
The uncertainties in air-sea fluxes are difficult to quantify, with the
lack of direct measurements a major problem, but they decrease as the
time and space scales expand. Best known are the global, long term
mean heat and freshwater fluxes, which ocean inventories show are near
zero. At the other extreme are turbulent fluxes whose sign at a point
can be uncertain on time scales less than about 10 minutes. Perhaps
most relevant to GODAE are hourly fluxes on a spatial scale of 10 km
or more. Some of these might be tractable from satellite measurements,
but whenever bulk aerodynamic formulae are involved there is at least
a factor of 2 uncertainty due to transfer coefficient variability on these
scales.
The temptation for GODAE to simply utilize global fields of Q, F
and wind stress should be resisted, because too much good physics is
lost. Notable examples include the penetration of solar radiation into
the upper ocean, the relationship between evaporation and the latent
heat flux, sea-ice coverage and the strong dependence of fluxes on SST.
However, the price is high because of the large number of forcing fields
necessary to prescribe the flux components.
Unfortunately, there is no single flux data suitable for GODAE. Instead, various data sets should be merged within an assimilation data
stream. However, a crucial procedure within the stream should be the
correction of most fields for known biases, as revealed by comparisons
with more limited, but more accurate and better understood observations. These adjustments should result in global heat and freshwater
fluxes that are nearly in balance, and these budgets should not be overly
upset by any increments made to the fluxes during the assimilation cycle. In addition the incremented fluxes ought to display the observed
features of global variability, especially those portrayed by large scale
climate indices.
The forcing of a truly global ocean data assimilation system that
incorporates all the details of the known physics would be extremely
complicated. Therefore, simplifications should be sought, but implemented only if well understood and demonstrated to be tolerable from
the viewpoint of assimilation products. A prime example is high latitude
forcing where representation of much of the physics requires a coupled
sea-ice model. Another is forcing near coastlines, where some satellite
products are contaminated by the land and continental runoff is complicated. Also, some fluxes may not need to be accounted for; such as
the temperature of precipitation and runoff, the latent heat of fusion of
glacier and ice-sheet runoff, and the melting of snow.