Estimates of Surface Heat Fluxes
75
Assuming (as I do) that the advice given to the Perth Water Board was as good
as could be expected given the present state of the coupled-modelling art, then I
think this Western Australian experience is likely to be the first of many like it. I say
this because all the greenhouse models agree on two things: (a) they all (from first
generation through to the present ones) have predicted that climate warming should
have already occurred since the 1970s, and most get the order of magnitude of the
warming rate about right; and (b) the rate of warming is about to increase further, and
to continue high for many decades into the future—even if we somehow drastically
reduce greenhouse emissions. (The reason is that carbon dioxide has a lifetime of
about 100 years in the atmosphere, so our present temperatures are a function of the
last 100 years of CO 2 emissions.)
USE OF BETTER HEAT FLUXES TO VALIDATE GREENHOUSE
COUPLED CLIMATE MODELS
If we regard the result (a) as a reason for taking the prediction (b) seriously, then
I expect that in coming decades situations like that in Perth will crop up in many
locations around the world, often simultaneously. We therefore need better tools for
sorting out whether a decadal trend (in rainfall or evaporation, say) at a given location
is part of a natural, interannual variation, or a harbinger of a more serious, long-term
trend which may force major societal disruption at the place where it is occurring.
If the WHOI heat flux product—or improvements on it, based on a global
verification program—proves to live up to its initial promise, I think it will prove an
extremely valuable tool in such work. I will use the tropical Indian Ocean, as being
the part of the ocean I have dealt with most in recent years, to illustrate what I mean.
One widely acknowledged weakness in global coupled models is their inability
to predict the Asian monsoons well. At least as of 6 years ago, the atmospheric
components of these models differed widely in where the monsoon rains fall; what its
all-Asian average is; and what its interannual variability is, when forced by observed
SSTs (Kang et al., 2002). There are many reasons why this may be so. It may be due
to issues such as representation of topography like India’s Ghats mountains; or the
fact that the atmospheric models do not get the (very large) diurnal cycle of rainfall
at all accurately, suggesting some defects in underlying physics. These models also
have great difficulty in simulating the observed strength and mean properties of the
Madden–Julian Oscillations, referred to earlier.
One possible cause of such problems is that the mean heat fluxes they deliver to
the ocean may be wrong (note that it is only the ocean that we need to be concerned
about—the annual mean heat flux into land should be close to zero, since the heat
cannot be mixed downwards or advected, as in the ocean). If the evaporative heat
flux (the largest term in the tropics, after the incoming solar radiation) is wrong in a
model, then the moisture transport towards rain areas—and therefore the rain quantity
itself—must also be wrong.
75
Assuming (as I do) that the advice given to the Perth Water Board was as good
as could be expected given the present state of the coupled-modelling art, then I
think this Western Australian experience is likely to be the first of many like it. I say
this because all the greenhouse models agree on two things: (a) they all (from first
generation through to the present ones) have predicted that climate warming should
have already occurred since the 1970s, and most get the order of magnitude of the
warming rate about right; and (b) the rate of warming is about to increase further, and
to continue high for many decades into the future—even if we somehow drastically
reduce greenhouse emissions. (The reason is that carbon dioxide has a lifetime of
about 100 years in the atmosphere, so our present temperatures are a function of the
last 100 years of CO 2 emissions.)
USE OF BETTER HEAT FLUXES TO VALIDATE GREENHOUSE
COUPLED CLIMATE MODELS
If we regard the result (a) as a reason for taking the prediction (b) seriously, then
I expect that in coming decades situations like that in Perth will crop up in many
locations around the world, often simultaneously. We therefore need better tools for
sorting out whether a decadal trend (in rainfall or evaporation, say) at a given location
is part of a natural, interannual variation, or a harbinger of a more serious, long-term
trend which may force major societal disruption at the place where it is occurring.
If the WHOI heat flux product—or improvements on it, based on a global
verification program—proves to live up to its initial promise, I think it will prove an
extremely valuable tool in such work. I will use the tropical Indian Ocean, as being
the part of the ocean I have dealt with most in recent years, to illustrate what I mean.
One widely acknowledged weakness in global coupled models is their inability
to predict the Asian monsoons well. At least as of 6 years ago, the atmospheric
components of these models differed widely in where the monsoon rains fall; what its
all-Asian average is; and what its interannual variability is, when forced by observed
SSTs (Kang et al., 2002). There are many reasons why this may be so. It may be due
to issues such as representation of topography like India’s Ghats mountains; or the
fact that the atmospheric models do not get the (very large) diurnal cycle of rainfall
at all accurately, suggesting some defects in underlying physics. These models also
have great difficulty in simulating the observed strength and mean properties of the
Madden–Julian Oscillations, referred to earlier.
One possible cause of such problems is that the mean heat fluxes they deliver to
the ocean may be wrong (note that it is only the ocean that we need to be concerned
about—the annual mean heat flux into land should be close to zero, since the heat
cannot be mixed downwards or advected, as in the ocean). If the evaporative heat
flux (the largest term in the tropics, after the incoming solar radiation) is wrong in a
model, then the moisture transport towards rain areas—and therefore the rain quantity
itself—must also be wrong.