Status of the research at UCLA on the parameterization of subgrid-scale orographic effects in large-scale models of the atmosphere
Young-Joon Kim
Department of Atmospheric Sciences
University of California, Los Angeles
Outline of the Research
The
main objective of the research has been to verify and improve parameterization
of orographic gravity-wave drag that is needed in numerical models of the atmosphere.
We earlier noted that parameterization schemes for orographic gravity-wave drag
are evaluated mostly from the viewpoint of overall improvement of simulated
fields (e.g., Palmer et al. 1986 ) in which errors from a variety of other sources
may coexist and interact. In order to avoid this situation, we followed a direct
way of assessing parameterization schemes of orographic gravity-wave drag by
using a mesoscale gravity-wave model. We evaluated gravity-wave parameterization
schemes by comparing the results to those simulated by a mesoscale model with
high horizontal resolution ( Kim and Arakawa 1995 ). We constructed a revised
scheme to improve the parameterization based on our findings from the experiments
with the gravity-wave model.
In most large-scale models of the atmosphere, the effects
of subgrid-scale orography are represented by means of parameterizing subgrid-scale
orographic gravity-wave drag and/or enhancing grid-scale orography, such as
"envelope orography" ( Wallace et al. 1983 ), with the use of subgrid-scale
orographic variance. We implemented the revised parameterization scheme and
an envelope orography into the UCLA General Circulation Model (GCM) of the Atmosphere.
Through preliminary experiments with the GCM, we found that the scheme greatly
improved the northern hemisphere winter simulations especially when used in
conjunction with an envelope orography ( Kim 1996 ). It is of further interest
to see the impact of the gravity-wave drag and the envelope orography on the
summer simulations of the northern hemisphere. Also of interest is the interannual
variability of the impact as well as the impact with higher resolution versions
of the model. Research along this line is now underway.
PHASE I: Numerical Simulation of Orographic Gravity Waves
As
the first phase of the research, we developed a gravity-wave model based on
the 2-dimensional, anelastic, nonhydrostatic system of equations with second-order
closure turbulence and a thin, but efficient sponge layer ( Kim 1992; Kim et
al. 1993).
In order to generate a database for assessing gravity-wave
drag parameterization schemes, we performed a series of simulations of mountain
waves with various shapes and sizes of orography ( Kim and Arakawa 1995 ). We
successfully simulated various mountain waves, some of which were verified against
some analytic solutions of the flow over idealized orography.
PHASE II: Development of a New Parameterization Scheme for Orographic Gravity Waves for use in a Large-Scale Model
Observational
studies show that gravity waves generated by orography break not only in the
lower stratosphere but also in the lower troposphere. The potential importance
of nonlinear drag enhancement due to low-level wave-breaking in parameterizing
orographic gravity waves has been recognized by many modelers (e.g. Chouinard
et al. 1986; Palmer et al. 1986; Pierrehumbert 1986; Peltier and Clark 1986;
McFarlane et al. 1987; Bacmeister 1993 ), but it has not been explicitly incorporated
in most existing parameterization schemes, except by, e.g., Miller et al. (1989)
.
As the second phase of our research, we first constructed
a test parameterization scheme by making use of the essential features of the
existing schemes. The reference-level drag is determined using the formulation
by Pierrehumbert (1986) , while the vertical distribution of the reference-level
drag is obtained following the algorithm by Miller and Palmer (1986) , which
incorporates the ideas of wave-saturation and wave-dissipation by Lindzen (1981)
and Eliassen and Palm (1960) , respectively.
We then evaluated the test parameterization scheme with
the aid of the dataset collected from the mountain-wave simulations obtained
for a variety of orographic conditions from the gravity-wave model developed
in Phase I. Through a series of experiments performed with the dataset, we found
that the test scheme, which performs similar to Palmer et al's (1986) scheme,
does not properly treat the enhancement of drag due to low-level wave-breaking.
The experiments we performed indicate that the standard deviation of orography
and the tuning coefficient for the parameterization alone are not sufficient
for properly representing this effect of low-level wave-breaking ( Kim and Arakawa
1995 ).
To overcome this deficiency, we devised more detailed
statistical measures of subgrid-scale orography (i.e., higher moments than the
standard deviation), such as the asymmetry and convexity (or sharpness) of orography,
for use as additional input to the parameterization scheme, and revised the
test scheme ( Kim and Arakawa 1995 ). We performed a series of experiments with
the revised scheme using the dataset obtained from the mountain wave simulations.
The results demonstrate that the revised scheme improves the results through
the selective enhancement of the reference-level drag and the divergence of
drag depending upon the orographic configuration and the corresponding flow
condition.
We performed an "off-line" test of the revised
scheme by applying the scheme to global monthly mean data. We started the 4
deg. lat. x 5 deg. lon. 15-layer version of the UCLA GCM, excluding gravity
wave parameterization, with 12Z 1 Oct. 1982 NCEP (NMC) analyses using 10-year
climatological surface conditions, and integrated for 4 months. We diagnostically
applied our revised scheme to January mean variables obtained from this simulation.
The revised scheme generates substantial drag at low levels especially in the
midlatitude northern hemisphere. Our results is similar to that of Miller et
al. (1989) in that low-level drag divergence is significant (in global average)
and is larger than upper-level drag. The difference is that our result is obtained
through selective enhancement of drag when there is possibility of low-level
wave-breaking as detected by the additional statistical measures of orography
in the revised scheme ( Kim and Arakawa 1995 ). We argued that unlike most other
schemes, our scheme distinguishes between the two contrasting flow situations
in terms of "upstream" and "downstream" effects on the drag;
i.e., decreasing the drag in the upstream region and enhancing the drag in the
downstream region.
PHASE III: Implementation and Evaluation of the New Parameterization Scheme for Orographic Gravity Waves into a Large-Scale Model
The
third phase of the research was to implement and evaluate the revised scheme
in the UCLA GCM. Since the scheme was developed under a 2-D framework based
on our 2-D gravity-wave model, some additional considerations were made to implement
into a 3-D large-scale model. For example, we calculated Orographic Asymmetry,
which is one of the additional statistical measures of orography, for 8 predetermined
wind directions, while we calculated Orographic Convexity for full 3-D orography
in a grid box, both from the 10' Navy high-resolution orographic data. We first
performed sensitivity experiments of the GCM simulation to the parameters of
the scheme that are mainly associated with 3-dimensionality of the model which
could not be determined with a 2-D gravity-wave model. We then cold-started
the model from several dates in October, 1982, and ran using climatological
model boundary conditions until January next year to obtain the mean of January,
during which the atmospheric stability is largest and thus the effect of gravity-wave
drag is largest in the northern hemisphere.
We found from the ensembles of the simulations that
the gravity-wave parameterization scheme and the envelope orography have a qualitatively
similar and beneficial impact on ensemble means of simulated January climate
( Kim 1996 ). The mid-latitude westerlies were weakened at all levels and the
polar atmosphere was warmed in the northern hemisphere. A combination of the
two produced the best results. Sensitivity experiments with the parameterization
scheme also indicated the importance of the selective enhancement of low-level
drag which we argued in our earlier study ( Kim and Arakawa 1995). PHASE IV:
Investigation of the Impact of Parameterized Gravity-Wave Drag and Envelope
Orography on Climate Simulated by a Large-Scale Model
The fourth phase of the research is to further evaluate our parameterization
scheme of orographic gravity waves with emphasis on its impact on the simulated
climate. In our earlier study ( Kim 1996 ), we found that the magnitudes of
gravity-wave drag are systematically different in January simulations using
the two representations of orography in the mid-latitude northern hemisphere
although the overall impact of gravity-wave drag on the mean fields is similar
when using the standard version of orography or the envelope orography. The
modification in the magnitudes of simulated meridional eddy momentum fluxes
by gravity-wave drag with the standard orography is as in earlier studies. It
is, however, not the case with the envelope orography. Whereas the impact of
gravity-wave drag and the envelope orography on the mean fields is similar,
it is not necessarily true in terms of the individual components of simulated
angular momentum budget.
In Kim (1996) , we were limited by the computational
efficiency to obtaining the ensemble means of eight January simulations obtained
from 4 month integrations (Oct. year 0 through Jan. year 1) which do not cover
annual cycles. In order to obtain more robust results, long-term multi-year
simulations are required. It also remains to be seen how the envelope orography
and gravity-wave drag parameterization perform in northern hemisphere summer
as some studies report degradation of summer simulations with envelope orography.
To this end, we recently developed a version of the UCLA GCM code suitable for
massively parallel processors ( Mechoso et al. 1993; Mechoso 1995; Wehner et
al. 1995 ). It is of considerable interest to perform higher horizontal resolution
(2 deg. lat. x 2.5 deg. lon.) experiments since the gravity-wave drag scheme
was designed to parameterize the effects of subgrid-scale gravity waves in domains
smaller than those corresponding to the current resolution of the model, not
to mention that higher resolution better resolves the nature. These simulations
are currently being performed ( Kim et al. 1999). Besides, we have also increased
the vertical resolution (no. of layers) of the model from 15 to 29 and further
to 58 in terms of the number of model layers.
Furthermore, we are verifying the scheme with various
other physics parameterization routines of the model. The atmospheric component
of the UCLA coupled atmosphere-ocean general circulation model has recently
been upgraded from the tropospheric to stratospheric version. Since the presence
of the stratosphere introduces additional physical processes (e.g., ozone and
photo chemistry) and since the stratosphere is where the gravity-wave drag is
largest in magnitude, we need to study the response of the model to the introduction
of gravity-wave drag with and without those physical processes. We have recently
experienced interesting sensitivity of the simulation to some details of the
radiation parameterization. For example, after we prescribed the ozone mixing
ratio rather than let the model predict, we found that the upper part of the
model atmosphere (model top is at 1 mb) became excessively warmer. Then, we
improved the short-wave radiation scheme and obtained significantly cooled-down
upper atmosphere, close to that with predicted ozone. We are currently working
on the improvement of long-wave radiation parameterization as well. Since a
modest change in radiation parameterization can significantly change the thermodynamics
and/or dynamics of the model atmosphere and therefore change the structure and
magnitude of the polar night jet in a way similar to gravity-wave drag, we need
to systematically investigate the separate and combined impact of gravity-wave
drag and radiation. A study along this line is reported in Kim et al. (1998).
References