Anthropogenic climate change for 1860 to 2100 simulated with the HadCM3 model under updated emissions scenarios

Abstract.

In this study we examine the anthropogenically forced climate response over the historical period, 1860 to present, and projected response to 2100, using updated emissions scenarios and an improved coupled model (HadCM3) that does not use flux adjustments. We concentrate on four new Special Report on Emission Scenarios (SRES) namely (A1FI, A2, B2, B1) prepared for the Intergovernmental Panel on Climate Change Third Assessment Report, considered more self-consistent in their socio-economic and emissions structure, and therefore more policy relevant, than older scenarios like IS92a. We include an interactive model representation of the anthropogenic sulfur cycle and both direct and indirect forcings from sulfate aerosols, but omit the second indirect forcing effect through cloud lifetimes. The modelled first indirect forcing effect through cloud droplet size is near the centre of the IPCC uncertainty range. We also model variations in tropospheric and stratospheric ozone. Greenhouse gas-forced climate change response in B2 resembles patterns in IS92a but is smaller. Sulfate aerosol and ozone forcing substantially modulates the response, cooling the land, particularly northern mid-latitudes, and altering the monsoon structure. By 2100, global mean warming in SRES scenarios ranges from 2.6 to 5.3 K above 1900 and precipitation rises by 1%/K through the twenty first century (1.4%/K omitting aerosol changes). Large-scale patterns of response broadly resemble those in an earlier model (HadCM2), but with important regional differences, particularly in the tropics. Some divergence in future response occurs across scenarios for the regions considered, but marked drying in the mid-USA and southern Europe and significantly wetter conditions for South Asia, in June–July–August, are robust and significant.

Appendices

Appendix A: Calibration of indirect sulfate aerosol forcing

Annual-mean distributions of sulfate aerosol were generated from HadAM3 model runs which included the sulfur cycle. Only natural emissions were used for the pre-industrial era, namely emissions of DMS (Kettle et al. 1999; Liss and Merlivat 1986) and volcanic emissions (Andres and Kasgnoc 1998). Anthropogenic emission scenarios were added to the natural emissions for the years 1860, 1900, 1950, 1975, 2000, 2050 and 2100 (see Appendix C). Successive pairs of these sulfate distributions (e.g. pre-industrial and 1860; 1860 and 1900) were then used as input to 1-year runs of HadAM3 with climatological (present-day) sea surface temperatures and sea ice. The version of HadAM3 used was identical to the atmospheric component of HadCM3 except that it used the relation of Jones et al. (1994) to predict cloud droplet number concentration (N _d ) rather than the fixed N _d values used in the HadCM3 control run. Concentrations of sulfate and (over the oceans) sea-salt aerosols were used as input to this relation to determine N _d (Jones et al. 1999), subject to a minimum value of 5 cm^–3. For each model run the shortwave radiation scheme was called twice, once with the earlier sulfate distribution and once with the later. In each case a simple measure of cloud albedo α was calculated (Bohren 1980):

$$ \alpha = \tau (1 - g)[2 + \tau (1 - g)]^{- 1} $$

(1)

This assumes conservative scattering, and in approximating the indirect effect we also assume a fixed asymmetry parameter g (=0.85) and use the fact that the optical (extinction) cross section of a particle much larger than the wavelength is twice its geometrical cross section (Bohren and Huffman 1980).

For each 1-year HadAM3 model run using a pair of sulfate scenarios, the distribution of the change in α caused by the changes in aerosol was stored and used to build up a history of the change in cloud albedo (Δα). If dα₁, dα₂ etc. represent the changes in cloud albedo for the sulfate distributions pairs {pre-industrial,1860}, {1860, 1900} etc. respectively, then the cloud albedo change at each "marker" year is calculated by summation:

$$\eqalign{& \Delta \alpha (1860) = d\alpha _1 \cr & \Delta \alpha (1900) = d\alpha _1 + d\alpha _2 \cr & etc.} $$

Cloud albedo changes between these dates were obtained by linear interpolation.

These cloud albedo changes were used to simulate indirect forcing in the HadCM3 runs as follows. The change in α due to the change in aerosol distribution corresponds to changing τ by a factor λ:

$$ \Delta \alpha = \lambda \tau (1 - g)[2 + \lambda \tau (1 - g)]^{- 1} - \tau (1 - g)[2 + \tau (1 - g)]^{- 1} $$

(2)

Re-arranging Eq. (1) to give an expression for τ and then substituting into Eq. (2) yields:

$$\Delta \alpha = \lambda \alpha [(1 - \alpha) + \lambda \alpha ]^{- 1} - \alpha $$

(3)

Following Platnick and Twomey (1994), we define a factor χ, the fractional change of N _d corresponding to the fractional change λ in τ. As τ is inversely proportional to the cloud droplet effective radius r _e , and as r _e is proportional to N _d ^–1/3 (Martin et al. 1994), then we can replace λ by χ^1/3. Re-arranging Eq. (3) then gives:

$$\chi = [\Delta \alpha (\alpha - 1) - \alpha (1 - \alpha)]^3 [\alpha (\Delta \alpha + \alpha - 1)]^{- 3} $$

(4)

At each time step in the HadCM3 run the fixed values of N _d and the predicted cloud water content were used to derive an initial value for r _e using the parametrization of Martin et al. (1994). Cloud optical depth τ was then calculated and used to derive an initial distribution of α using Eq. (1). The distribution of Δα for the appropriate time was available from the offline history produced from the HadAM3 runs and used in conjunction with α to calculate χ using Eq. (4). Note that at this point in the model r _e has only been used to determine the initial distribution of α. Before r _e was used in the radiation code for a full calculation of cloud radiative properties it was modified to r _e ′ using χ to approximate the indirect albedo effect:

$$r^{\prime}_e = r_e \chi ^{- 1/3} $$

(5)

It is possible for χ to be infinite, zero or negative, implying either: (1) that the decrease in cloud albedo required is greater than (χ < 0) or equal to (χ = 0) the current value of α, and in such cases the cloud was made as dark as possible by setting r _e to the maximum value allowed (37 μm); or (2) that the increase in cloud albedo required to produce the desired forcing, when added to α, produces a result which is greater than (χ < 0) or equal to (χ → ∞) unity, and in such cases the cloud was made as bright as possible by setting r _e to the minimum permitted value (0.35 μm). Both these factors may exert a limiting influence on the forcing actually produced in the coupled run, as may differences between the cloud distribution of the coupled run and of the atmosphere-only runs used to generate the albedo-change history. It is too computationally expensive to calculate the actual forcing experienced by the coupled model as the simulation evolves as this would require multiple calls to the (computationally intensive) radiation scheme. Consequently, 1-year portions of the simulations were re-run with extra radiation calls and diagnostics for the scenario years 1860, 1900, 1950, 1975, 2000, 2050 and 2100. The ideal forcing for these years was available from the separate HadAM3 runs used to generate the albedo-change history, and this was compared with the actual forcing produced in HadCM3 in these years.

Appendix B: Calculation of ozone trends

In the troposphere the ozone trends are constructed by means of off-line 3-D chemical transport (STOCHEM, Collins et al. 1997, 1999) model simulations for specific epochs 1990, 2030, 2060 and 2100; each simulation being run for 15 months. Intermediate results are obtained by linear interpolation in time. Initial methane concentrations are provided from an earlier run of the 2-D TROPOS model. Emissions for 1990 are based on those from the EDGAR database (Olivier et al. 1996), with future emissions taken from the SRES scenario definitions themselves. The 3-D STOCHEM model runs are also meteorologically and dynamically forced with six-hourly data from a previous HadCM3 coupled experiment which to first order reflects the climate conditions (e.g. global warming) expected at the appropriate time.

All SRES scenarios assume the Montreal protocol is followed, the Cl and Br concentrations being determined via box modelling techniques. The resulting stratospheric ozone concentrations are then scenario-independent. For the purpose of extrapolating stratospheric ozone depletion into the future, 'effective equivalent stratospheric chlorine' (EESC = Cl + 40*Br) contributions were estimated for Cl and Br compounds not already included in the 2-D model used for computing well-mixed minor GHGs by means of a simple box model driven using SRES and natural emissions where appropriate. Tropospheric mixing ratio estimates from this and the 2-D model for years from 1950 to 2100 were then combined and used to calculate stratospheric ozone as a function of height, longitude and month, assuming:

• 1. That it takes 3 years for tropospheric Cl and Br to get into the stratosphere

• 2. That ozone is depleted linearly by EESC levels above the 1974 stratospheric level, any implied negative ozone values being replaced by a minimum ozone mass mixing ratio of 10^–11 for numerical reasons

Ozone loss is at its largest in 2002 and recovers subsequently as EESC returns to preindustrial values. Beyond about 2060, this should have resulted in a stabilization of stratospheric ozone at preindustrial levels again, but due to a software error ozone in fact continued to increase. This leads to an erroneous small positive net forcing of +0.17 W/m² from stratospheric ozone at 2100 relative to preindustrial in all the scenarios.

Further details of the method and the stratospheric ozone trends up to present day is given by Tett et al. (2002). Zero ozone change was imposed at the tropopause model level, as estimated from the 1871–1990 mean tropopause height of a HadAM3 integration forced by the observed sea surface temperatures and sea ice extents from the GISST3.1 data set, an updated version of that described in Rayner et al. (1996).

Appendix C: Anthropogenic sulfur emissions

Only anthropogenic emissions were included in the experiments. For the 'past' portion of the experiments, the basic datasets we used are the revised version of the 1985 GEIA 1B dataset (released in June 1997 – see http://www.geiacenter.org/ and http://www.ortech.ca/cgeic/ for details) and the University of Stockholm (Orn et al. 1996) historic emissions datasets for 1860, 1870,… 1970. The GEIA 1B dataset includes seasonal variations over some regions, e.g. Europe, but for convenience we employed an annually averaged version. In any case the Stockholm datasets are annual means. The GEIA 1B dataset also splits the emissions into those from near-surface and elevated sources such as power station chimneys; the latter were injected into the third model level above the surface (approximately 550–1000 m), to allow for buoyant plume rise.

For the 'future' portion of the experiments, we used the preliminary versions of two of the four possible marker scenarios produced by the IPCC SRES group, namely A2 and B2. We also used the published versions of the 'fossil fuel intensive' variant of the A1 scenario (known as A1FI) and the B1 marker scenario, both of which were developed at a later stage (Nakicenovic et al. 2000).

The SRES SO₂ emissions are provided on a regular latitude–longitude grid at a resolution of 1°× 1°. Fields are available at ten year intervals from 1990 to 2100. Note that the emissions actually used at 1990 in the HadCM3 experiments B1 and A1FI were identical to those in A2 for simplicity in conducting the experiments, rather than the slightly different values tabulated in Table 2 which reflect revisions between the preliminary and final versions of the respective emissions scenarios. The complete sequence of base dates, between which emissions are found by interpolation, is therefore 1860, 1870, … 1980, 1985, 1990, 2000, 2010, … 2090, 2100.

Define a two-dimensional field ${\cal F}$ by

$${\cal F}(x,y) = {{H(x,y,1985)} \over {H(x,y,1985) + S(x,y,1985)}}$$

where H(x, y, 1985) is the GEIA 1B elevated emission field and S(x, y, 1985) is the GEIA 1B surface emission field. ${\cal F}$ is thus the fraction of the emissions from elevated sources. (Where there are no emissions in 1985 ${\cal F}$ is set to zero.) ${\cal F}$ was used to split the Stockholm historic datasets into surface and elevated components. Clearly the fraction of emissions from elevated sources has varied in time, so introduce a factor φ (t) to describe this: assume that φ(t) is zero before 1950, 1 after 1985 and varies linearly with time between 1950 and 1985. Emissions prior to 1985 were thus split by multiplying them by $\phi (t){\cal F}(x,y)$ to generate an elevated source dataset, the remaining emissions being assigned to the surface source field.

We have no information about possible changes in ${\cal F}$ in the future, so we assumed that no changes take place after 1985. This is not ideal, because it is clear that changes will occur, and the emission height does affect the sulfate distribution.