|
6.3.1. Models for Radiative Forcing
Calculation of radiative forcing from 3-D models is a relatively new endeavor.
The history of this calculation began with one-dimensional models (e.g., Hansen
et al., 1984a) that made use of the basic radiative-convective model approach
initiated by Manabe and Wetherald (1967). Subsequent researchers expanded this
calculation, using GCM output to describe atmospheric lapse rate, water vapor,
and cloud cover and including latitudinal and seasonal changes (e.g., Pollack
et al. 1993). Both the radiative perturbations from aircraft and the background
radiative constituents (e.g., clouds, water vapor, albedo) vary with altitude,
latitude, longitude, and time.
Three-dimensional modeling of radiative forcing introduces substantial complexity.
The different modeling groups cited here have different approaches to calculating
RF, involving choices in spatial and temporal domains.
All models report instantaneous values of radiative forcing at the top of the
atmosphere and at the tropopause; in other words, these RF values have been
calculated with no changes in atmospheric temperature. As discussed in Section
6.2.1, the most appropriate RF value includes allowance for stratospheric
temperatures to readjust to radiative perturbation. Only two groups (Forster
and Haywood, and Ponater and Sausen-see paragraphs below) have models that allow
for such stratospheric adjustment. This correct, adjusted RF usually lies between
the instantaneous values at the top of the atmosphere and the tropopause, and
we correct the RF reported from other groups so that all RF values here refer
to tropopause radiative forcing with stratospheric adjustment.
RF modeling results were contributed by P. Forster and J. Haywood (Forster
and Shine, 1997), A. Grossman (Grossman et al., 1997), J. Haywood (Haywood and
Ramaswamy, 1998), D. Rind (Rind and Lonergan, 1995), W.-C. Wang (Wang et al.,
1995), and M. Ponater and R. Sausen (Ponater et al., 1998).
Forster and Haywood's radiation scheme has been previously used to calculate
ozone and water vapor radiative forcings; it is described in Forster and Shine
(1997). It employs a 10 cm-1 narrowband model (Shine,
1991) in the thermal infrared (IR) and a discrete-ordinate model (Stamnes et
al., 1988) at solar wavelengths with 5-nm resolution in the ultraviolet (UV)
and 10-nm resolution in the visible. As in Forster and Shine (1997), the fixed
dynamic heating approximation (Ramanathan and Dickinson, 1979) is used to calculate
stratospheric temperature perturbations. A zonally and annually averaged, 5°
latitudinal resolution climatology was used as the basis for the forcing calculations.
Temperature and humidity were derived largely from European Centre for Medium-range
Weather Forecasts (ECMWF) analyses, averaged over the period 1980-91. In the
upper stratosphere, temperatures were derived from Fleming et al. (1990). At
pressures less than 300 hPA, humidity was based on a combination of Stratospheric
Aerosol and Gas Experiment II (SAGE-II) and Halogen Occultation Experiment (HALOE)
data. Surface albedos, cloud amounts, and optical depths were 7-year averages
from International Satellite Cloud Climatology Project (ISCCP) (Rossow and Schiffer,
1991). Clouds were specified at three levels. The thermal infrared calculations
included absorption by nitrous oxide, methane, and carbon dioxide. Ozone climatologies
were taken from an observed climatology derived by Li and Shine (1995), a combination
of SAGE-II, Solar Backscatter Ultraviolet (SBUV), Total Ozone Mapping Spectometer
(TOMS), and ozonesonde data. To calculate forcing, the climatological profiles
were perturbed by the absolute annual averages of ozone and water vapor changes.
Grossman (Grossman et al., 1997) uses a set of baseline annual and longitudinal
average atmospheric profiles, resolved by 50 layers between 0 and 60 km, at
latitudes of 60°N/S, 30°N/S, and at the Equator that are scaled to IS92A (IPCC,
1995) composition. Supersonic and subsonic aircraft O3 and H2O perturbation
profiles were added to the baseline atmospheric profiles for RF calculations.
The Lawrence Livermore National Laboratory (LLNL) 16-band solar radiation model
(Grant and Grossman, 1998) and the LLNL 32-band IR radiation model (Chou and
Suarez, 1994) were used to calculate instantaneous tropopause and top-of-atmosphere
RFs at each latitude for the global average value for O3
and H2O.
Haywood's RF calculations for sulfate and black carbon aerosols were made following
the method of Haywood and Ramaswamy (1998). The Geophysical Fluid Dynamics Laboratory
(GFDL) R30 GCM incorporates a 26 band delta-Eddington solar radiative code (Ramaswamy
and Freidenreich, 1997) and includes the cloud parameterization of Slingo (1989)
and aerosol optical properties calculated using Mie theory. RF calculations
are performed at the top of the atmosphere every day using mean solar zenith
angle. No account is made for stratospheric adjustment, the effects of which
are likely to be small for tropospheric aerosol in the solar spectrum.
Ponater and Sausen estimated instantaneous RF using the ECHAM4 GCM (Roeckner
et al., 1996). Radiative transfer calculations (one radiative time step only)
were performed for each grid point with and without local ozone perturbation,
including the actual cloud profile. Several diurnal cycles were calculated for
each calendar month, and the radiative flux change was determined for each individual
grid point at the top of the atmosphere and at the tropopause. The annual global
mean radiative forcing was obtained by averaging over all grid points and over
the seasonal cycle. To calculate the stratosphere-adjusted, tropopause RF, a
"second atmosphere" was implemented into the ECHAM4 GCM. Whereas the primary
atmosphere of the GCM does not "feel" the perturbation of the greenhouse gas,
the second atmosphere experiences an additional radiative heating above the
tropopause, although dynamic heating is identical to that of the first, unperturbed
atmosphere. In the troposphere, the primary and second atmospheres are not allowed
to diverge. In this configuration, the model is run for one annual cycle.
Rind used the Goddard Institute for Space Studies (GISS) Global Climate Middle
Atmosphere Model (Rind et al., 1988). The radiation scheme in the model is the
correlated-k method for modeling non-gray gaseous absorption (Lacis and Oinas,
1991). The procedure involved keeping re-start files and full diagnostics for
the first full day of each month from a control run. Radiation and all other
routines were called each hour, so a full, diurnal average global response was
calculated. Then the first day of each month was re-run with altered atmospheric
composition (e.g., changes in ozone). The global net radiation at the top of
the atmosphere was compared. The assumption is that with only 1 day of running
time, temperatures would not adjust (even in the stratosphere) to the altered
composition; hence, the results are instantaneous values.
Wang's RF calculations use the National Center for Atmospheric Research (NCAR)
Community Climate Model 3 (CCM3) radiative model with monthly mean, latitude-by-longitude
distributions of vertical temperature, moisture, clouds, and surface albedo
simulated from the Atmospheric Model Intercomparison Project (AMIP). The year
1992 of the CCM3-AMIP simulations was used because the corresponding year was
used to simulate ozone in the Oslo 3-D CTM (Isaksen et al., 1999). Because this
State University of New York/ Albany version of CCM3 used the ozone climatology
(Wang et al., 1995), CTM-simulated absolute ozone changes for 1992-2015 and
1992-2050 are mapped onto 1992 ozone climatology to calculate the RF. RFs are
based on fixed temperature treatment rather than fixed dynamic heating treatment
(Wang et al., 1993).
|