Water vapour feedback continues to be the most consistently important feedback accounting for the large warming predicted by general circulation models in response to a doubling of CO2. Water vapour feedback acting alone approximately doubles the warming from what it would be for fixed water vapour (Cess et al., 1990; Hall and Manabe, 1999; Schneider et al., 1999; Held and Soden, 2000). Furthermore, water vapour feedback acts to amplify other feedbacks in models, such as cloud feedback and ice albedo feedback. If cloud feedback is strongly positive, the water vapour feedback can lead to 3.5 times as much warming as would be the case if water vapour concentration were held fixed (Hall and Manabe, 1999).
As noted by Held and Soden (2000), the relative sensitivity of OLR to water vapour changes at various locations depends on how one perturbs the water vapour profile; the appropriate choice depends entirely on the nature of the water vapour perturbation anticipated in a changing climate. The sensitivity is also affected by cloud radiative effects, which tend to mask the influence of sub-cloud water vapour on OLR. Incorporating cloud radiative effects and a fixed relative humidity perturbation (argued to be most appropriate to diagnosing GCM water vapour feedback), Held and Soden suggest that OLR is almost uniformly sensitive to water vapour perturbations throughout the tropics. Roughly 55% of the total is due to the free troposphere in the "tropics" (30°N to 30°S) with 35% from the extra-tropics. Allowing for polar amplification of warming increases the proportion of water vapour feedback attributable to the extra-tropics. Of the tropical contribution, about two thirds, or 35% of the global total, is due to the upper half of the troposphere, from 100 to 500 mb. The boundary layer itself accounts for only 10% of the water vapour feedback globally. Simulations incorporating cloud radiative effects in a doubled CO2 experiment (Schneider et al., 1999) and a clear-sky analysis based on 15 years of global data (Allan et al., 1999) yield maximum sensitivity to water vapour fluctuations in the 400 to 700 mb layer (see also Le Treut et al., 1994). In a simulation analysed by Schneider et al. (1999) extra-tropical water vapour feedback affected warming 50% more than did tropical feedback.
Most of the free troposphere is highly undersaturated with respect to water, so that local water-holding capacity is not the limiting factor determining atmospheric water vapour. Within the constraints imposed by Clausius-Clapeyron alone there is ample scope for water vapour feedbacks either stronger or weaker than those implied by constant relative humidity, especially in connection with changes in the area of the moist tropical convective region (Pierrehumbert, 1999). It has been estimated that, without changes in the relative area of convective and dry regions, a shift of water vapour to lower levels in the dry regions could, at the extreme, lead to a halving of the currently estimated water vapour feedback, but could not actually cause it to become a negative, stabilising feedback (Harvey, 2000).
Attempts to directly confirm the water vapour feedback by correlating spatial surface fluctuations with spatial OLR fluctuations were carried out by Raval and Ramanathan (1989). Their results are difficult to interpret, as they involve the effects of circulation changes as well as direct thermodynamic control (Bony et al., 1995). Inamdar and Ramanathan (1998) showed that a positive correlation between water vapour, greenhouse effect and SST holds for the entire tropics at seasonal time-scales. This is consistent with a positive water vapour feedback, but it still cannot be taken as a direct test of the feedback as the circulation fluctuates in a different way over the seasonal cycle than it does in response to doubling of CO2.
Other reports in this collection