Earlier IPCC reports had quoted a value of about -0.1 Wm-2/ decade with a factor of two uncertainty. There have been revisions in this estimate based on new data available on the O3 trends (Harris et al., 1998; WMO, 1999), and an extension of the period over which the forcing is computed. Models using observed O3 changes but with varied methods to derive the temperature changes in the stratosphere have obtained -0.05 to -0.19 Wm-2/decade (WMO, 1999).
Hansen et al. (1997a) have extended the calculations to include the O3 loss up to the mid-1990s and performed a variety of O3 loss experiments to investigate the forcing and response. In particular, they obtained forcings of -0.2 and -0.28 Wm-2 for the period 1979 to 1994 using SAGE/TOMS and SAGE/SBUV satellite data, respectively. Hansen et al. (1998) updated their forcing to -0.2 Wm-2 with an uncertainty of 0.1 Wm-2 for the period 1970 to present. Forster and Shine (1997) obtained forcings of -0.17 Wm-2 and -0.22 Wm-2 for the period 1979 to 1996 using SAGE and SBUV observations, respectively. The WMO (1999) assessment gave a value of -0.2 Wm-2 with an uncertainty of ± 0.15 Wm-2 for the period from late 1970s to mid-1990s. Forster and Shine (1997) have also extended the computations back to 1964 using O3 changes deduced from surface-based observations; combining these with an assumption that the decadal rate of change of forcing from 1979 to 1991 was sustained to the mid-1990s yielded a total stratospheric O3 forcing of about -0.23 Wm-2. Shine and Forster (1999) have revised this value to -0.15 Wm-2 for the period 1979 to 1997, choosing not to include the values prior to 1979 in view of the lack of knowledge on the vertical profile which makes the sign of the change also uncertain. They also revised the uncertainty to ± 0.12 Wm-2 around the central estimate. A more recent estimate by Forster (1999) yields -0.10 ± 0.02 Wm-2 for the 1979 to 1997 period using the SPARC O3 profile (Harris et al., 1998).
There have been attempts to use satellite-observed O3 and temperature changes to gauge the forcing. Thus, Zhong et al. (1996, 1998) obtained a small value of -0.02 Wm-2/decade; and with inclusion of the 14 micron band, a value of -0.05 Wm-2/ decade. It has been noted that the poor vertical resolution of the satellite temperature retrievals makes it difficult to estimate the forcing; in fact, a similar calculation using radiosonde-based temperatures yields a value of -0.1 Wm-2/decade (Shine et al., 1998). The main difficulty is that the temperature change in the vicinity of the lower stratosphere critically affects the emission from the stratosphere into the troposphere. Thus any uncertainty in the MSU satellite retrieval induced by the broad altitude weighting function (see WMO, 1999) becomes an important factor in the estimation of the forcing. Further, the degree of response of the climate system, embedded in the observed temperature change (i.e., feedbacks), is not resolved in an easy manner. This makes it difficult to distinguish quantitatively the part of temperature change that is a consequence of the strato-spheric adjustment process (which would be, by definition, a legitimate component of the forcing estimate) and that which is due to mechanisms other than O3 loss. Thus, using observed temperatures to estimate the forcing may be more uncertain than the model-based estimates. It must be noted though that both methods share the difficulty of quantifying the vertical and geographical distributions of the O3 changes near the tropopause, and the rigorous association of this to the observed temperature changes. In an overall sense, it is a difficult task to verify the radiative forcing in cases where the stratospheric adjustment yields a dramatically different result than the instantaneous forcing i.e., where the species changes affect stratospheric temperatures and alter substantially the long-wave radiative effects at the tropopause. A related point is the possible upward movement of the tropopause which could explain in part the observed negative trends in O3 and temperature (Fortuin and Kelder, 1996).
Kiehl et al. (1999) obtained a radiative forcing of -0.187 Wm-2 using the O3 profile data set describing changes since the late 1970s due to stratospheric depletion alone, consistent with the range of other models (see Shine et al., 1995). Kiehl et al. (1999) also present results using a very different set of O3 change profiles deduced from satellite-derived total column O3 and satellite-inferred tropospheric O3 measurements to arrive at an implied O3 forcing, considering changes at and above the tropopause, of -0.01 Wm-2. The reason for the considerably weaker estimate reflects the increased O3 in the tropopause region that is believed to have occurred since pre-industrial times (largely before 1970) in many polluted areas. How the changes in O3 at the tropopause are prescribed is hence an important factor for the difference between this calculation and those from the other estimates.
Clearly, since WMO (1992), this forcing has been investigated in an intensive manner using different approaches, and the observational evidence of the O3 losses, including the spatial and seasonal characteristics, are now on a firmer footing. In arriving at a best estimate for the forcing, we rely essentially on the studies that have made use of stratospheric O3 observations directly. Based on this consideration, we adopt here a forcing of -0.15 ± 0.1 Wm-2 for the 1979 to 1997 period. However, it is cautioned that the small values obtained by the two specific studies mentioned above inhibit the placement of a high confidence in the estimate quoted.
In general, the reliability of the estimates above is affected by the fact that the O3 changes in the lower stratosphere, tropopause, and upper troposphere are all poorly quantified, around the globe in general, such that the entire global domain from 200 to 50 hPa becomes crucial for the temperature change and the adjusted forcing. Forster and Shine (1997) note that the sensitivity of forcing to percentage of O3 loss near the tropopause is more than when the changes occur lower in the atmosphere.
Myhre et al. (1998a) derived O3 changes using a chemical model in
contrast to observations. As the loss of O3 in the upper stratosphere
in the simulations was large, a positive forcing of 0.02 Wm-2/decade
was obtained (see Ramanathan and Dickinson (1979) for an explanation of the
change of sign for a O3 loss in the lower stratosphere versus the
upper stratosphere). While there are difficulties in modelling the O3
depletion in the global stratosphere (WMO, 1999), this study reiterates the
need to be cognisant of the role played by the vertical profile of O3
loss amounts in the entire stratosphere, i.e., middle and upper stratosphere
as well, besides the lower stratosphere.
An important issue is whether the actual surface temperature responses to the forcing by stratospheric O3 has the same relationship with forcing as obtained for, say, CO2 or solar constant changes. Hansen et al. (1997a) and Christiansen (1999) have performed a host of GCM experiments to test this concept. The forcing by lower stratospheric O3 is an unusual one in that it has a positive short-wave and a negative long-wave radiative forcing. Moreover, it has a unique vertical structure owing to the fact that the short-wave effects are felt at the surface while the long-wave is felt only initially at the upper troposphere (Ramanathan and Dickinson, 1979; WMO, 1992). Compared to, say, CO2 change, the stratospheric O3 forcing is not global in extent, being very small in the tropics and increasing from mid- to high latitudes; the O3 forcing also differs in its vertical structure, since the radiative forcings for CO2 change in both the troposphere and surface are of the same sign (WMO, 1986). The relationship between the global mean forcing and response differs by less than 20% for O3 profiles, resembling somewhat the actual losses (Hansen et al., 1997a). However, serious departures occur if the O3 changes are introduced near surface layers when the lapse rate change, together with cloud feedbacks, make the climate sensitivity quite different from the nominal values. There also occur substantial differences in the climate sensitivity parameter for O3 losses in the upper stratosphere. This is further substantiated by Christiansen (1999) who shows that the higher climate sensitivity for upper stratospheric O3 losses relative to lower stratospheric depletion is related to the vertical partitioning of the forcing, in particular the relative roles of short-wave and long-wave radiation in the surface-troposphere system. It is encouraging that the global mean climate sensitivity parameter for cases involving lower stratospheric O3 changes and that for CO2 changes (viz., doubling) are reasonably similar in Christiansen (1999) while being within about 25% of a central value in Hansen et al. (1997a). An energy balance model study (Bintanja et al., 1997) suggests a stronger albedo feedback for O3 changes than for CO2 perturbations (see also WMO, 1999).
The evolution of the forcing due to stratospheric O3 loss hinges on the rate of recovery of the ozone layer, with special regards to the spatial structure of such a recovery in the mid- to high latitudes. If the O3 losses are at their maximum or will reach a maximum within the next decade, then the forcing may not become much more negative (i.e., it was -0.1 Wm-2 for just the 1980s; inclusion of the 1990s increases the magnitude by about 50%). And, as the O3 layer recovers, the forcing may remain static, eventually tending to become less negative. At this time, there will be a lesser offset of the positive greenhouse effects of the halocarbons and the other well-mixed greenhouse gases (WMO, 1999). Solomon and Daniel (1996) point out that the global mean stratospheric O3 forcing can be expected to scale down substantially in importance relative to the well-mixed greenhouse gases, in view of the former's decline and the latter's sustained increase in concentrations. Note, however, that the evolution of the negativity of the stratospheric O3 forcing may vary considerably with latitude and season i.e., the recovery may not occur at all locations and seasons at the same rate. Thus, the spatial and seasonal evolution of forcing in the future requires as much scrutiny as the global mean estimate.
Other reports in this collection