2.1
RADIATIVE TRANSFER
U.S. Dept. of Commerce /
NOAA /
OAR /
ERL /
GFDL
*Disclaimer
To investigate the characteristics of convection-cloud-radiative interactions on a variety of space and time scales, leading to the understanding of their role in weather, climate and climate change.
To use satellite and other meteorological observations for diagnostic analyses of climate processes, and for evaluating and improving physical parameterizations employed in general circulation models.
To study fundamental aspects of atmospheric radiative transfer, and to investigate the climatic effects of natural and anthropogenic radiatively-active trace gases and aerosols.
S.M. Freidenreich V. Ramaswamy
J. Haywood M.D.
Schwarzkopf
The
availability of extensive computational resources on the GFDL Cray T3E
allowed a re-examination of the catalog of overcast sky cases previously
determined by the binning method (starting in A89/P90). While the previous
sets of cases considered the effects of water vapor and drops on the near-infrared
solar absorption up to 18000 cm-1, the updated set contains
the effects of the additional gases (CO2, O3 and
O2 ) and Rayleigh scattering. It also extends the computations
to cover the entire solar spectrum, and employs the "exact" line-by-line
+ doubling-adding (LBL+DA) method. LBL+DA calculations for ice clouds based
on optical properties of hexagonal crystals have begun.
The
shortwave "benchmark" computations of sulfate aerosol direct
radiative forcing (A96/P97) have been completed as part of an intercomparison
project. A multi-authored, multi-institutional paper is in preparation.
2.1.2
Characteristics of Solar Fluxes
Analyses
have been made of the solar flux disposition in the atmosphere based on
the benchmark results of an extensive number of overcast sky situations.
The results substantiate the hypothesis, first made in A92/P93 based on
limited calculations, that the surface solar flux is approximately invariant
with cloud height. A more dramatic conclusion from the current analyses
is the fact that this near-invariance of the surface solar flux with cloud
height occurs throughout the near-infrared spectrum. The reason for this
near-invariance is that the transmission of solar radiation to the surface
occurs mainly in spectral regions that are nearly transparent, i.e., where
water vapor absorption is weak. Further, the absorption by cloud drops
tends to coincide with bands of water vapor absorption. Thus, no matter
where the clouds are placed, radiation is either reflected or absorbed
by the drops or is absorbed by the water vapor above, in or below cloud,
with the result being that the same amount of flux is transmitted to the
surface.
Further
analysis of the complete solar spectral disposition of the fluxes will
continue, with particular focus on the quantitative dependence of absorption
in clear versus cloudy skies. The treatment of explicit aerosol/cloud effects
in the longwave spectrum will be considered, from both "benchmark"
and parameterization perspectives.
2.2
CONVECTION-CLOUDS-RADIATION-CLIMATE INTERACTIONS
2.2.1
Cumulus Parameterization
L. Donner
A
cumulus parameterization has been developed that places unique emphasis
on the statistical aspects of convective-scale vertical velocities and
microphysics (1349). The overall intensity of parameterized convective
systems is related to the properties of the large-scale flows in which
the systems develop by a hypothesis referred to as a closure. The key closure
assumption is that the vertically integrated forcing provided by the convective
system balances large-scale destabilization. This provides an equilibrium
between large-scale forcing and convective response, but neglects the details
of the interactions between convective subensembles. A one-dimensional
column model has been developed using the closure and the parameterization
described in (1133). The parameterization has also been coded into the
SKYHI GCM, where it is undergoing preliminary testing.
Testing
of the parameterization in the SKYHI GCM will be completed and long-term
integrations to assess its role in climate will be initiated. The role
of deep convection in thermodynamics, hydrology, radiative forcing, and
tracer transport will be studied.
2.2.2
Limited-Area Nonhydrostatic Models
C. Andronache V. Ramaswamy
L. Donner T.
Reisin
J. Haywood C. Seman
R. Hemler
Deep
convection and its associated mesoscale circulations were modeled using
the Lipps-Hemler (885) three-dimensional cloud-system model. The energy
and moisture budgets, large-scale heat sources and moisture sinks, microphysics,
and radiation have been examined. The modeled cloud system undergoes a
life cycle dominated by deep convective towers in its early stages, followed
by an upper-tropospheric mesoscale circulation. The large-scale heat sources
and moisture sinks associated with the convective system agree broadly
with diagnoses from field programs, but the modeled upper-tropospheric
moisture (Fig. 2.1) exceeds observed values. Strong radiative cooling at
the top of the mesoscale circulation can produce overturning there. Radiation
exerts a strong influence on the microphysical properties
of the system (dq). Three-dimensional integrations exhibit considerably less sporadic temporal behavior than corresponding two-dimensional integrations. They also produce stronger interactions between radiation and dynamics in the upper-tropospheric mesoscale circulation.
A
new approach to microphysics in the model has been implemented, consisting
of prognostic equations for several moments of the particle-size distributions.
During the past year, most of the focus of this project has been on improving
the computational performance of this method.
The
evolution of sulfate from sulfur dioxide has been studied with two- and
three-dimensional versions of the cloud-system model. Absorption of sulfur
dioxide by condensed water, Brownian diffusion and nucleation scavenging
of dry sulfate by condensate, oxidation of absorbed sulfur dioxide to sulfate,
transfer of sulfate among different forms of condensed water, and removal
of sulfate by precipitation were treated.
The
distribution of relative humidity from the model was compared to analyzed
relative humidities from the UK Meteorological Research Flight for approximately
similar atmospheric conditions (dg). Reasonable agreement was found between
the two distributions, reaffirming earlier conclusions (A96/P97) that coarser
grid resolutions typical of GCMs will likely underestimate aerosol forcing
because they will miss the sub-grid scale variation of relative humidities
and the effects of those variations on particle sizes.
Analysis
of the three-dimensional cloud-system model will focus on mechanisms important
in the behavior of the cumulus parameterization (2.2.1). Examples include
the relative roles of the convective and stratiform components of the system
and the relationships between cloud vertical motions, microphysics, and
radiation. The physical and numerical controls on the ice content in the
stratiform circulation will be examined, particularly the lateral boundary
conditions, ice sedimentation, and small-scale convection in the anvil
circulation. Analysis of the interactions between microphysics and radiation
will continue with an emphasis on a microphysical parameterization which
predicts particles sizes. The transport and transformation of sulfur and
its impact on radiative properties will be evaluated. Preliminary planning
will begin regarding use of the cloud-system model in a field experiment
in the Arabian Sea which will study aerosol chemistry and transport.
L. Donner B. Soden
R. Hemler J. Warren
C. Seman
Ice
clouds influence both the shortwave and longwave radiation balance of the
earth-atmosphere system and thereby play a delicately balanced role in
climate. A study of ice-clouds associated with large-scale atmospheric
processes has been completed (ef) using the SKYHI GCM and parameterizations
for the microphysical and radiative properties of ice clouds. The ice source
was deposition from vapor, and the ice sinks were gravitational settling
and sublimation. Particle sizes were related empirically to temperature.
Radiative properties were evaluated as functions of ice path and effective
size using approximations to detailed radiative-transfer solutions. The
distributions of atmospheric ice and their impact on climate and climate
sensitivity were evaluated. Most of the major climatological cirrus regions
revealed by satellite observations appeared in the SKYHI GCM. The radiative
forcing associated with the ice clouds acted to warm the Earth-atmosphere
system, i.e., longwave forcing exceeded shortwave forcing. Relative to
a SKYHI integration without these clouds, zonally averaged temperatures
are warmer in the upper tropical troposphere with ice clouds.
The
incorporation of the latest cumulus parameterization (2.2.1) into the SKYHI
GCM will provide a method for linking upper-tropospheric ice clouds to
deep convection. Studies of this interaction will begin in FY98.
2.2.4
Radiative-Convective Equilibria with Explicit Moist
Convection
V. Balaji R. Hemler
I. Held O.
Pauluis
Simulations
of radiative-convective equilibrium in a horizontally homogeneous atmosphere
are being conducted with a cloud-resolving nonhydrostatic model. These
integrations address fundamental issues in the theory of moist convective
turbulence and cloud-radiative feedbacks. The studies are being carried
out with both two- and three-dimensional models at various resolutions
with a vertically stretched grid and cyclic lateral boundary conditions.
In the past year, the focus has been on studying the sensitivity of preliminary
three-dimensional simulations to resolution and domain size, and on studies
of the moist enthalpy, kinetic energy, and entropy budgets. Work has also
begun on a comparison of moist radiative-convective equilibrium with the
dry analog.
The
three-dimensional model was found to generate uniformly distributed convective
events over the domain. At a horizontal resolution of 2 km, there is little
or no tendency toward the clumping of convection and the spontaneous generation
of domain-scale circulations that is seen in two-dimensions. This homogeneity
of the convection allowed a reduction in the size of the domain (to 128
x 128 km) in the latest integrations. The cloud distribution at low levels
is sensitive to vertical resolution, with a tendency to produce near total
cloud cover when this resolution is low. The introduction of a stretched
grid has resulted in a more realistic, broken cloud field with little additional
computational expense (Fig. 2.2). Radiative-convective equilibrium calculations
are now underway with two different values of surface temperature (25
and 30C)
and with two different treatments of the ice microphysics.
Recent
theories of CAPE (convective available potential energy) and kinetic energy
dissipation are qualitatively supported by the current preliminary integrations.
A detailed study of kinetic energy dissipation in the model shows it to
be occurring, in part, near the convective cores, but a significant fraction
also occurs when the gravity waves generated by the convection are absorbed
by the model's sponge layer in the stratosphere. The total kinetic energy
dissipation is sensitive to resolution in two-dimensions, increasing rapidly
as the grid size is reduced from 5 to 2 km and then increasing more slowly
for even smaller grid size. This dissipation appears to be a useful measure
of the convergence of the model's deep convective eddies.
The
sensitivity of the three-dimensional model of radiative-convective equilibrium
to surface temperature and microphysical assumptions will be the main focus
in the coming year, with particular attention given to the sensitivity
of the relative humidity and cloud cover distributions. The comparison
of dry and moist equilibria will also be pursued. In addition, the study
of inhomogeneous statistically steady states will begin with a model of
a Walker cell in a relatively small domain in which the surface temperature
possesses an east-west gradient.
2.3
DIAGNOSTIC ANALYSES USING SATELLITE OBSERVATIONS
2.3.1
A Lagrangian Analysis of Upper Tropospheric Water
Vapor
B. Soden
Hourly
observations of 6.7 
m
"water vapor" radiances from geostationary satellites were used
in conjunction with an objective pattern-tracking algorithm to trace upper
tropospheric water vapor features from time-lapsed satellite imagery. From
this analysis, climatological diagnostics of the moisture and circulation
of the upper troposphere were derived (ff). A close correlation is observed
between upper tropospheric relative humidity and the water vapor pattern
displacements, reflecting the strong dependence of relative humidity upon
the atmospheric circulation. A new analysis technique was developed which
permits study of the evolution of the upper tropospheric moisture field
from a Lagrangian perspective. This analysis demonstrates that the clear-sky
upper troposphere in both tropical and subtropical regions becomes drier
with time, reflecting the impact of large-scale subsidence in drying the
upper troposphere. It was further found that cloud cover has a substantial
influence on the Lagrangian drying rate of the upper troposphere, presumably
due to the re-evaporation of cloud condensate.
Trajectories
of upper tropospheric moisture were also constructed by tracking water
vapor patterns from sequential satellite images. These trajectories reveal
two distinct paths for water vapor entering the subsidence regions of the
subtropics. One path highlights the poleward propagation of convective
outflow from the tropics while the other path reflects the injection of
moisture from westward propagating extratropical disturbances. This result
suggests that a complete picture of the processes regulating the driest
and most radiatively transparent regions of the subtropics may require
an understanding of the role of both tropical and extratropical convective
systems.
The
relationship between upper tropospheric water vapor and the tropical circulation
will be examined on a variety of time-scales, including diurnal, seasonal,
and interannual. A similar tracking algorithm will also be applied to global,
high resolution satellite measurements of ocean surface temperature to
characterize the motion and Lagrangian evolution of SST features.
2.3.2
Characteristics of Clear-Sky Solar Absorption
B. Soden V. Ramaswamy
To
better understand the discrepancy between theory and observations regarding
the amount of solar radiation absorbed by the atmosphere, the variability
in solar absorption over the global oceans was examined. Satellite observations
of the clear-sky, top-of-atmosphere solar absorption from the Earth Radiation
Budget Experiment were compared to that simulated by a GFDL R30 climate
GCM. Systematic discrepancies between the observations and GCM simulations
were noted. In particular, the observed zonal and interannual variations
in clear-sky solar absorption are substantially larger than those predicted
by the GCM. The greater observed variability is closely associated with
changes in column integrated water vapor (Fig. 2.3) and, to a lesser extent,
aerosol concentrations and surface wind speed. Three possible explanations
for the discrepancies are: 1) deficiencies in our current understanding
of water vapor absorption in the solar spectrum; 2) the absence of aerosols
in the model; 3)the
absence of a wind-speed dependent ocean surface albedo.
Investigation
of the discrepancies between the observed and GCM-simulated solar absorption
will continue, with special emphasis on the influence of aerosols and ocean
surface roughness on the absorbed solar radiation.
2.3.3
Sensitivity of a GCM to Observed Cloud Properties
M. Crane R. Wetherald
B. Soden
Monthly
climatologies of cloud amount and cloud optical properties were created
using measurements from the International Satellite Cloud Climatology Project
(ISCCP). The GFDL R30 climate GCM was then integrated using prescribed
cloud properties from the ISCCP climatology in place of model-predicted
quantities. When compared to satellite observations, the top-of-atmosphere
radiative fluxes obtained from model integrations using the ISCCP-prescribed
clouds show notable improvement relative to the predicted cloud runs (Fig.
2.4). Greatest improvement is realized for the reflected solar radiation,
particularly in regions dominated by extensive low cloud cover. This improvement
is most evident over subtropical oceans where boundary layer clouds are
often poorly simulated in GCMs. The outgoing longwave radiation also exhibits
better agreement with the satellite observations, with the most notable
improvement occurring in the tropics.
A
better climatology of ISCCP cloud properties will be created by expanding
the data set to include multiple years of observations. The sensitivity
of the GCM simulations to assumptions made regarding cloud geometric depth
will be considered, with particular emphasis on the impact of clouds on
infrared heating rates. The impact of the prescribed clouds upon surface
radiative fluxes will also be examined, with special emphasis given to
their possible impact on heat flux adjustments in coupled ocean-atmosphere
models.
2.4
CLIMATIC EFFECTS DUE TO ATMOSPHERIC SPECIES
C-T. Chen R. Orris
J. Haywood V. Ramaswamy
J. Mahlman M.D. Schwarzkopf
2.4.1
"Predicted" Cloud Distributions in SKYHI
The
new shortwave and longwave radiation schemes, along with a cloud prediction
scheme similar to that used in the Climate Dynamics Group's GCM, have been
incorporated in the SKYHI GCM. Integrations with one version of the model
(latitude-longitude resolution= 3x3.6)
have been carried out. The new cloud scheme works by inferring the existence
of a cloud in a vertical layer within a horizontal grid box based on the
relative humidity in that layer. If a cloud is predicted, it is assumed
to entirely fill the grid box in that layer. The cloud amount, number of
clouds, and altitude distribution may thus vary in four dimensions. The
radiative properties in layers with clouds are determined by the appropriate
cloud optical properties. By contrast, the standard prescribed-cloud distribution
assumes high, middle and low clouds, which only partially fill the horizontal
grid box and are fixed in longitude, altitude and time.
Comparison of GCM simulations (described below) using the two schemes shows that the predicted-cloud approach results in significant improvements.
Figure
2.5 displays the differences in the zonally averaged fractional cloud amount,
temperature, and relative humidity between two simulations for January,
the first employing the older prescribed-cloud distribution and the other
using the new predicted cloud scheme with "part black" high clouds
(see below). The cloud fraction for the predicted-cloud simulation is less
than that for prescribed clouds at altitudes of 700850 hPa. Increases are
observed near the surface and in the tropical upper troposphere (150250hPa).
The predicted cloud simulation produces zonally averaged cloud distributions
(not shown here) which appear far more realistic than those from the prescribed
cloud formulation. The temperature response to the change in cloud distribution
is highlighted by a large (~12 K) increase in the tropical tropopause region.
This is significant in light of the "cold bias" of older SKYHI
GCM simulations in that region. Lower tropospheric temperatures increase
(decrease) in response to the cloud fraction decrease (increase). The H2O
mixing ratio in the tropical upper troposphere increases by a factor of
up to ~8, and also shows smaller increases throughout most of the troposphere.
GCM
results have been compared to satellite measurements of outgoing longwave
radiation (OLR), the "upper tropospheric humidity" parameter
(a measure of the water vapor amount averaged over the upper troposphere),
and the precipitable water (a measure of the column of lower tropospheric
water vapor). Fig. 2.6 compares the deviation of the January zonal averages
from measured values for three different GCM simulations: 1) prescribed-clouds;
2)predicted-clouds
with "black" high clouds (emissivity of unity); and 3) predicted-clouds
with "part black" high clouds (emissivity of 0.6). The predicted-clouds
simulations use the new shortwave parameterization (A95/P96) and the longwave
"CKD2.1" and "9gas" "parameterizations (A96/P97).
The prescribed cloud simulation does not use the new shortwave parameterization.
The
OLR calculations indicate that the "part black" simulation is
a substantial improvement over both the prescribed cloud and "full
black" cloud simulations in the tropics. In northern midlatitudes,
the prescribed cloud simulation appears to be the best, probably due to
appropriate (fixed) values of the cloud parameters. The upper tropospheric
humidity and the precipitable water calculations using either of the predicted
cloud schemes show an increase in tropical water vapor amount.
2.4.2
Observed Ozone Loss and Stratospheric Temperature
Change
Two
additional SKYHI GCM simulations were performed to test the sensitivity
of the lower stratospheric temperature change to the vertical distribution
of the ozone loss in the lower stratosphere. Both the simulations produced
a substantial cooling of the lower stratosphere, confirming earlier conclusions
(A96/P97) that the ozone loss initiates a strong radiatively-induced perturbation
that leads directly to a temperature decrease. All the simulations agree
qualitatively with the corresponding MSU satellite-derived lower stratospheric
temperature changes, with the zonal pattern of cooling captured reasonably
well. However, considerable quantitative uncertainties remain, affirming
the need for an accurate assessment of the global ozone changes near the
tropopause region, which is currently monitored very poorly.
2.4.3
Tropospheric Aerosol Radiative Effects
The
R30 climate model was used in conjunction with newly developed anthropogenic
tropospheric sulfate aerosol (1421) and tropospheric black carbon aerosol1
distributions to investigate the direct radiative forcing of each species
(fc). The new solar radiative transfer code (A95/P96) was modified to explicitly
include the radiative properties of tropospheric aerosols. A top-of-the-atmosphere
radiative forcing diagnostic was developed for the R30 GCM. The annual
mean present day radiative forcing is estimated to be -0.82 W m-2
for sulfate aerosol and +0.40 W m-2 for black carbon aerosol.
The calculated spatial distribution of the radiative forcing due to sulfate
alone and an external mixture of sulfate and black carbon combined are
shown in Fig. 2.7 and Fig. 2.8, respectively. Additional investigations
into the effects of sub-grid scale variations of relative humidity based
upon the relative humidity distribution from the limited area non-hydrostatic
model (AP96/97, dg, 2.2.2) reveal a systematic underestimate of the radiative
forcing if sub-grid scale variations of relative humidity are not accounted
for.
Additional
sensitivity calculations (Fig. 2.9) have shown that the radiative forcing
due to sulfate aerosol tends to be strongest close to the surface where
the relative humidity is highest (due to the hygroscopic nature of sulfate
aerosol). Conversely, the radiative forcing due to black carbon aerosol
is strongest when the aerosol is at higher altitudes due to the combined
effects of the aerosol residing above more cloudy layers and the reduction
of scattering and absorption in the overlying atmosphere.
An
assessment of the relative influence of absorbing and nonabsorbing aerosols,
from a modeling and observational perspective, has been completed in collaboration
with an international group of scientists, specifically highlighting the
radiative effects of black carbon.
The
top-of-the-atmosphere clear sky upward solar irradiance was calculated
over the Atlantic Ocean using the new 26-band delta-Eddington solar radiative
transfer code (A95/P96). The results were compared to the same quantity
derived from the Earth Radiation Budget Experiment (ERBE). Systematic differences
in the top-of-the-atmosphere irradiances were found in areas that correspond
to high aerosol concentrations, such as the Saharan dust plume off the
western coast of North Africa, a biomass burning plume off the western
coast of South Africa and South America, and plumes of continental aerosol
off the eastern coast of the United States. These differences in the top-of-the-atmosphere
irradiances can be used to help constrain chemical transport models, and
assumptions about the radiative characteristics of tropospheric aerosols
that are used in current estimates of radiative forcing.
2.4.4
Radiative Forcing and Climate Response
The
relationship between radiative forcing and the accompanying climate response
was investigated in the case of a globally homogeneous greenhouse gas forcing
and a variety of Northern Hemisphere-only forcings characteristic of anthropogenic
sulfate aerosols. It was found (1440) that, while the global-mean climate
sensitivity (i.e., global-mean surface temperature response divided by
the global-mean radiative forcing) is the same regardless of whether the
forcing is global or confined to the Northern Hemisphere, the regional
response in the Northern Hemisphere is quite different. In particular,
the meridional gradient of the surface temperature response in the Northern
Hemisphere is dependent on the spatial confinement of the forcing and its
latitudinal and longitudinal extents. However, even though the global-mean
feedbacks yield this particular result, this cannot be taken to imply that
the individual feedback components behave in an identical manner for all
types of forcings.
Investigations
into the linear additivity of the modeled climate response to combined
greenhouse gas and direct sulfate aerosol forcings (A95/P96 and A96/P97)
were concluded using the Climate Group's R15 GCM coupled to a mixed layer
ocean model (1438). The study reveals that radiative perturbations due
to greenhouse gas and aerosol increases that are small relative to the
short- and longwave radiative fluxes yield a linearly additive response
of an atmosphere-mixed layer ocean climate system. However, the fact that
the total climate system acts to yield this simple behavior does not imply
that the individual components behave in a linear, additive manner.
Further
exploration of this feature has been carried out by analyzing the transient
R15 coupled full ocean-atmosphere GCM integrations for increased greenhouse
gas and sulfate aerosol concentrations (A96/P97, en). In these model investigations,
when temperature change due to increased concentrations of greenhouse gases
is added to that due to increased concentrations of sulfate aerosols, the
resultant temperature change is found to be very similar (spatial correlation
coefficient, r=0.97) to that when both greenhouse gases and sulfate aerosols
are included in the model integrations. Similar results are found for precipitation,
although the correlation coefficient is lower due to the inherent noise
in the precipitation signal. These results indicate that the responses
to many different scenarios (e.g., uncertainties in aerosol direct radiative
forcings) may be addressed by scaling and summing the results from only
a few coupled ocean-atmosphere integrations. This further substantiates
the concept first performed with the mixed-layer version of the model (1438).
2.4.5
Tropospheric Ozone Radiative Forcing
The
preindustrial and present-day climatologies of tropospheric ozone (A96/P97,
1445) have been used in conjunction with the R30 GCM to assess the radiative
forcing associated with changes in concentrations of tropospheric ozone
due to anthropogenic activity. The new solar radiative transfer code (A95/P96)
was included in the GCM, and radiative forcing diagnostics for both solar
and terrestrial radiative forcings were developed. The annual mean global
solar, terrestrial and total instantaneous radiative forcings are +0.07
W m-2, +0.31 Wm-2 and +0.38 W m-2. The
total global-mean forcing estimate is comparable to the black-carbon forcing.
Together, these two forcings nearly offset the direct sulfate aerosol forcing
(2.4.3). Additional sensitivity calculations were performed to investigate
the tropospheric ozone radiative forcing assuming cloud-free conditions.
It was found that the global annual mean instantaneous radiative forcing
was approximately 72% lower in the solar spectrum, but approximately 42%
higher in the terrestrial spectrum, resulting in a cloud-free radiative
forcing of +0.46 W m-2. The effect of stratospheric adjustment
is to reduce the total-sky global, annual-mean instantaneous radiative
forcing by approximately 10%.
2.4.6
Consistency of Ozone and Temperature in the Middle
Atmosphere
A
study of the radiative consistency of ozone and temperature observations
in the middle and upper stratosphere was completed (1418). Temperatures
derived from the fixed dynamical heating (FDH) model calculations, which
employ heating rates from the 1
SKYHI GCM and use the UARS MLS (Microwave Limb Sounder) ozone as input,
are slightly colder than the MLS and LIMS (Limb Infrared Monitor of the
Stratosphere) temperatures, but within the uncertainty estimates of the
measurements and the model. Because of its demonstrated capability to reproduce
well the monthly- and zonal-mean stratospheric temperatures of the 1-degree
SKYHI GCM (within 1 K in the tropics and 6 K in the winter hemisphere mid-
and high-latitudes), the FDH model provides a fast and inexpensive tool
with which to conduct these consistency studies.
An
examination of the uncertainties in the data and model yield the following.
The ozone measurements possess systematic error estimates ranging from
5-40%, and the temperature measurements have systematic error estimates
ranging from 2-10 K. The FDH modeled temperatures also have systematic
biases ranging from 2-10 K, due to uncertainties in the radiative transfer
and interannual variability of the dynamical forcing of the stratosphere.
FDH-modeled temperatures in the tropics using two different years of dynamical
heating rates differ by as much as 5 K. A study of 25 years of temperature
data from the 3
resolution SKYHI GCM reveals a low frequency variability of the model temperatures,
suggesting a significant source of uncertainty in FDH calculations that
use dynamical heating rates from a one-year GCM dataset. Furthermore, most
of the measured datasets are only available for a few years. If equivalent
amplitudes of low frequency variations are also present in the real atmosphere,
then several years and perhaps decades of measurements may be needed to
adequately determine the true climatological means. This result is very
important because it implies that for careful intercomparison studies to
be valuable (even in the tropics), either the dynamical state of the atmosphere
must be properly accounted for, or many years of data must be averaged
to reduce noise due to interannual variability.
 |
In
order to eliminate the uncertainty due to interannual variability, the
area-weighted global means of the FDH-modeled and measured temperatures
are compared on pressure surfaces (Fig.2.10). Among observations, MLS and
LIMS temperatures agree to within 2 K in the global-mean at pressures of
1mb and greater. The Barnett and Corney (BC) temperatures are significantly
warmer (3.5-5.0 K) than either the MLS or LIMS temperatures between 2-10
mb during the Northern Hemisphere winter, suggesting that BC temperatures
do not accurately represent the temperature of the global-mean stratosphere
at those levels. The FDH-modeled temperatures are colder than measured
at all pressure levels for all months. When account is taken of the uncertainties
in the FDH calculation and the uncertainty in the measured temperatures,
the biases in the 210mb region between the FDH-modeled temperatures and
both the MLS and LIMS measurements become marginal although remaining negative.
The reasons for this bias are either that the SKYHI radiative transfer
scheme has a cold bias, or the MLS temperatures are too warm. More likely,
it is a combination of both problems.
PLANS FY98
|
*Portions of this document contain material that has not yet been formally published and may not be quoted or referenced without explicit permission of the author(s).