Global warming, models and language - Climate Change: The Facts - page 34

to consider water vapour and cloud feedbacks separately, but this actually makes little
sense. Water vapour is important forinfrared radiation, while clouds are important forboth
infrared and visible radiation. We will refer to the former as longwave radiation and the
latter as shortwave radiation. The water vapour feedback is an infrared feedback, but it
applies only in regions free of upper level cirrus. The areal coverage of the upper level
cirrus (one type of higher level cloud) is highly variable. We must therefore consider the
total longwave feedback. Similarly, we can look at the total shortwave feedback.
Direct measurement of feedbacks with satellite data
An obvious approach to measuring feedbacks would be to see how outgoing radiation
responds to surface temperature fluctuations. The crucial point about the feedbacks is that
theyrespondtosurfacetemperaturefluctuationsregardlessoftheoriginofthefluctuations.
Not surprisingly, this approach also has difficulties.
Satellite measurements of outgoing radiation (both longwave and reflected shortwave)
are used in this approach. It turns out that the model intercomparison program has the
models used by the IPCC, forced by actual sea-surface temperature, calculate outgoing
radiation. So one can use the same approach with models, while being sure that the models
are subject to the same surface temperature fluctuations that applied to the observations.
In principle, this should be a straightforward task. However, as already noted, there are
difficulties. The first two difficulties involve basic physical considerations.
First, not all time scales are appropriate for such studies. Greenhouse warming
continues until equilibrium is re-established. At equilibrium, there is no longer any
radiative imbalance. If one considers time intervals that are long compared to equilibration
times, then one will observe changes in temperature without changes in radiative forcing.
The inclusion of such long time scales thus biases results inappropriately toward high
sensitivity. Equilibration times depend on climate sensitivity. For sensitivity on the order
of 0.5°C for a doubling of CO 2 , it is on the order of years, and for higher sensitivities it is
on the order of decades. In order to avoid biasing sensitivity estimates, one should restrict
oneself to time intervals less than a year. This problem is particularly acute for approaches
which simply regress outgoing radiation on temperature without concern for time scales
(viz.the2006paperbyPiersForsterandJonathanGregory,andthe2010paperbyAndrew
Dessler). 9 Moreover, given that the response time is shorter for low sensitivity, one has the
ironic result that the bias toward high sensitivity is greater when the actual sensitivity is
lower.
There is also the need to consider time intervals long enough for the relevant feedback
processes to operate. For water vapour and cloud feedbacks, these time scales are typically
on the order of days. For practical time resolution, this is generally not a problem.