Saltzman B. (editor) Anomalous Atmospheric Flows and Blocking
Подождите немного. Документ загружается.
29
-
2 4x5
215
-
2 4x5
29
-
215
4x5
FIG.
29.
As
in Fig.
27,
except for the
IOday
forecast.
29
-
z
2
4x3
Z15
-
Z
2
4x3
29
-
215
2
4x3
FIG.
30.
As
in
Fig.
27,
except for the
5day
forecast from
15
February
with
high-resolution
versions.
2s
-
z
2
4x3
Z15
-
Z
2 4x3
29
-
215
FIG.
3
1.
As
in Fig.
30,
except for the
7day
forecast.
L9
-
z
2
4x3
215
-
Z
2
4x3
29
-
215
2.4X3
FIG.
32.
As
in Fig.
30,
except for the
I0-day
forecast.
NUMERICAL FORECASTS
OF
EVENTS
405
ary can become comparable to the errors due to other sources, at least for
some cases. The large difference between the
15-
and 9-layer forecasts for the
2
1
January case and the small difference for the 15 February case will be
interpreted in the following discussions.
6.3.
Ultralong Wave Forecasts
Figures
33
and
34
show the observed and predicted amplitude and phase
of zonal wavenumber
1
in geopotential height at
60"N
for the
5-
and 10-day
US
UO
35
30
-
zc
25
E
+
I
l3
w
x
-
20
I5
10
5
0
-2
-3
-5
-
10
-20
g
-
-30
w
U
3
-50
$
-70
e
-
100
-
200
300
500
7
700
-
-
2
U
6
8
1012lUl6
180W
120W
60W
0
RHPLITUOE
(100Ml
LONGITUDE
OF
RIDGE
FIG.
33.
Amplitude and phase (position
of
the ridge)
of
zonal
wavenumber
1
in geopotential
height at
60"N
on
26
January (thick-solid lines) and 5-day
forecasts
with the low-resolution
9-layer version
(dasheddotted
line), low-resolution 15-layer version (thin solid lines),
high-res-
olution 9-layer version (dotted lines), and high-resolution 15-layer version
(dashed
lines).
406
CARLOS
R.
MECHOSO
ET
AL.
2
11
6
8
1012
1416
180W
120W
60W
0
QMPL
I
TUDE
I1
OOMI
LONGITUDE
OF
RIDGE
FIG.
34.
As
in
Fig.
33,
except
for
31
January
and
IO-day
forecasts.
forecasts from
21
January, respectively.
At
day 5, the wave amplitude
is
generally well predicted in the troposphere by all forecasts and underpre-
dicted
in
the stratosphere by both 15-layer forecasts. The 9- and 15-layer
forecasts for the wave phase exhibit significant differences: the 9-layer fore-
casts show a tendency toward reducing the vertical tilt with height, particu-
larly above
300
mbar, while the 15-layer forecasts are quite accurate. At day
10,
the wave amplitude is underpredicted
in
the lower stratosphere but
overpredicted in the upper stratosphere by both 15-layer forecasts. The
forecast errors for wave phase are generally larger than at day
5.
However, the
tendency of the 9-layer forecasts toward reducing the vertical tilt with height
remains apparent. The significant retrogression
of
the wave below
500
mbar
during the second half of the forecast period is qualitatively predicted by the
low-resolution 15-layer forecast and ignored
by
all other versions.
?
3
3
10
?O
z
r
a
30
w
50
2
3
70
100
a.
'00
300
500
700
350
NUMERICAL FORECASTS OF EVENTS
407
The ultralong wave with zonal wavenumber
2
was predominant during
the period of the forecast from
15
February. Figures
35
and
36
show the
observed and predicted amplitude and phase of zonal wavenumber
2
at
60"N
for that case. At day
5,
the wave amplitude is underpredicted by all
forecasts in the troposphere and lower stratosphere. The high-resolution
forecasts seem to
be
the most accurate there. Contrary to the previous case,
there are no major differences between the
9-
and 15-layer forecasts for the
wave phase. All of them show the same tendency toward reducing the verti-
cal tilt with height in the middle and upper troposphere and in the lower
stratosphere. At day
10,
the errors are larger than at day
5
and the high-reso-
lution versions show superior performance in predicting the wave ampli-
tude.
uc
40
3s
30
-
=
25
5
+
I
c3
w
I
-
20
15
10
5
0
2
3
5
10
!O
iz
E
30
w
70
&
(r
3
50
a
I00
!OO
300
ioo
100
)SO
200
YO0
600
800
GOW
0
'
GOE
RMPLITUDE
(MI
LONGITUDE
OF
RIDGE
FIG.
35.
As
in Fig.
33,
except
for
20
February and
forecasts
from
15
February
of
zonal
wavenumber
2.
408
115
110
35
30
-
= 25
5
+
I
c3
W
I
-
20
15
10
5
0
-
-
-
-
-
-
-
CARLOS
R.
MECHOSO
ETAL.
-2
-3
-5
10
-20
G
E:
-30
W
cc
3
-50
$
-70
W
100
200
300
500
700
850
1
#
-
200
400
600
800
60W
0
60E
AMPL
I
TUOE
(MI
LONGITUDE
OF
RIOGE
FIG.
36.
As
in Fig.
35,
except
for
25 February
and
IO-day
forecasts.
6.4.
On
the Diflerent Impact
of
the Lowered Upper Boundary for the
Forecastsfrom
21
January and
15
February
We have shown that the
9-
and 15-layer forecasts at 500 mbar are signifi-
cantly different at day
10
in
the
2
I
January case, while they are similar in the
15
February case.
To
interpret this, we first analyze the differences and
similarities between the forecasts in the framework of the theory on the
vertical propagation of planetary waves. According to the theory (Matsuno,
1970),
the propagation characteristics of a linear, steady, planetary wave
with zonal wavenumber
k
are given by the values
of
its refractive index
NUMERICAL FORECASTS
OF
EVENTS
409
squared,
Qk,
defined by
Here
u
is the zonal component of the horizontal velocity,
q
is the quasi-geo-
strophic potential vorticity,
(b
is latitude,fis the Coriolis parameter, Nis the
Brunt-VGsiilti frequency,
H
is the scale height, and
a
is the radius of the
earth. Butchart
et
al.
(1982) showed that higher (negative) values of
Qk
increase (prevent) propagation of the planetary wave with zonal wave-
number
k
even in unsteady cases.
The
Q,
field for the 2day, low-resolution 15-layer forecast from 2
1
Jan-
uary and the corresponding observed field are shown in Fig.
37.
The broad
regions with positive values in both fields indicate the possibility of vertical
propagation for wavenumber
1.
The upward propagation of this wave is
predicted by the 15-layer forecast as indicated by the successful forecast of
the minor stratospheric warming that occurred during this period. On the
other hand, the structure of the wave in the 9-layer forecast is affected in the
lower stratosphere according to the constraints imposed by the model's top,
and those effects can propagate downward, reaching eventually the middle
and lower troposphere.
The
Q2
field for the 2-day, high-resolution 15-layer forecast from
15
Feb-
a
b
"
90
80N
70
GO
50
110
30N
20
10
DAY
2
4x5
90
80N
70
GO
50
110
30N
20
10
LRT
I
TUDE
LRT
I
TUOE
FIG.
37. Refractive index squared
Q,
for
(a) 23 January and
(b)
2-day forecast
from
21
January with the low-resolution 15-layer version. Values greater than 100
or
less than -100
are
not contoured. For
this
figure
H
=
7
km.
410
CARLOS
R.
MECHOSO
ET
AL.
ruary and the corresponding observed field, shown
in
Fig.
38,
suggest differ-
ent situations. The observed field exhibits a region of weak negative values
centered at about
57
“N
and
500
mbar, and negative values at polar latitudes.
The predicted field shows a broader region of negative values with large
magnitudes centered at about 62”N and
400
mbar, and negative values at
high latitude. These differences in
Q,
are associated with small errors in the
prediction of the zonal wind and have a crucial effect on the characteristics of
upward wave propagation, as indicated by the failure in predicting the major
stratospheric warming that occurred during this period.
If,
for a 15-layer
forecast, planetary waves forced near the surface cannot propagate through
the upper troposphere, the forecast should not differ from another per-
formed with a model that has its top in the lower stratosphere.
In Section 6.2, we have shown that the 15- and 9-layer forecasts for
500
mbar from 2
1
January are significantly different at day 10, while those from
15
February are similar even at day
10.
This can now be interpreted as a
consequence of the differences and similarities in the forecasts of ultralong
waves during the early part
of
the forecast periods. In the “forecast” experi-
ment performed by Mechoso
et
al.
(
1982), the error due to a lowered upper
boundary appears first in ultralong waves (wavenumber
1
in that case) in the
uppermost layers, and then propagates downward and progressively spreads
to shorter waves. It is reasonable to assume that similar processes happened
in the forecasts from
2
1
January. In the forecasts from
15
February, on the
other hand, wavenumber
1
is not dominant and the vertical propagation of
a
b
x
Y
DAY
2
2.4x3
10
19
37
72
100
150
200
300
1100
ti
850
90
80N 70
60
50
110
30N
20
10
W
(r
3
m
m
W
(r
o_
LRT
I
TUDE LATITUDE
FIG.
38.
As
in Fig.
37,
except
for
QZ
for (a)
17
February and (b) 2-day forecast
from
15
February with the high-resolution IS-layer version.
NUMERICAL FORECASTS OF EVENTS
41
1
wavenumber
2
is underpredicted even in the 15-layer forecast.
As
a conse-
quence, the 15- and 9-layer forecasts for
500
mbar remain similar.
7.
CONCLUSIONS
We have compared 10-day forecasts with four versions of the
UCLA
GCM.
The versions differ in horizontal resolution (the low-resolution is 4"
of latitude by
5
of longitude, while the high-resolution is 2.4" of latitude by
3
of
longitude) or in vertical extent (the 9-layer has the top at
5
1.8 mbar,
while the 15-layer has the top at
1
mbar). In this article we have considered
four cases with starting dates 16 and 2
1
January 1979 and 15 and 17 Febru-
ary 1979. These cases are characterized by episodes of tropospheric blocking
and stratospheric sudden warming in the Northern Hemisphere.
We have shown 5-day forecasts for
500
mbar which are generally accurate,
particularly in midlatitudes. We have presented time means for the second
5-day period for selected forecasts which show the model's skill in predicting
quasi-stationary features
of
the flow. We have shown very accurate
5-
and
10-day forecasts for
10
mbar of stratospheric warming events during the
winter of 1979. We conclude that the
UCLA
GCM
has the ability to predict a
variety of atmospheric phenomena at
5
days and beyond.
The impact of increasing the horizontal resolution (from 4" of latitude by
5
O
of longitude to 2.4" of latitude to
3
O
of longitude) on the accuracy of the
forecasts at
500
mbar was not the same in all cases. The forecast from 16
January is quite insensitive to the increase in resolution. Also, there is no
clear indication of impact in the 2
1
January and 17 February cases. On the
other hand, the high-resolution forecasts from
15
February are more accu-
rate than those with the low-resolution version, particularly in midlatitudes.
The most visible impact of the increase in horizontal resolution was found in
the stratosphere. In particular, the forecast of the stratospheric major warm-
ing of February 1979 was drastically improved when the horizontal resolu-
tion was increased. This case is discussed in detail by Mechoso
et
al.
(1 985),
who argue that the reasons for improvement are not the more accurate
numerics in the stratosphere, but the more accurate forecast of the zonal-
mean wind in the upper troposphere. The slightly higher errors in the zonal-
mean wind with the low-resolution version in that region can have a large
impact on the characteristics of upward wave propagation. We conclude that
increased horizontal resolution can significantly improve both tropospheric
and stratospheric forecasts in some cases.
The impact of the upper boundary on forecasts was analyzed by studying
rms forecast errors and by comparing fields predicted with the 9-layer fore-
cast of the model, which has the top at
5
1.8 mbar, to those predicted with the
412
CARLOS
R.
MECHOSO
ETAL
15-layer version, which has the top at
1
mbar. The 9- and 15-layer versions
are identical below
5
1.8 mbar.
As
expected, the differences between 9- and
15-layer forecasts increased rapidly at the uppermost levels of the 9-layer
version in all cases. However, the differences for
500
mbar were case depen-
dent. They reached significant values in the 2
1
January case, and remained
generally small in the 15 February case. We interpreted this different behav-
ior on the framework of the theory on vertical propagation of planetary
waves. In the first case, the 15-layer version predicts correctly the upward
propagation of wavenumber 1,
as
indicated by the accurate forecast of the
minor stratospheric warming that occurred during the period. The wave
propagation through the lower stratosphere is altered in the 9-layer forecasts
by the presence of the model’s top. Furthermore, the effects of the top can
propagate downward, eventually reaching the middle and lower tropo-
sphere. In the second case, the 15-layer version does not predict correctly the
upward propagation of wavenumber
2,
as
indicated by the failure in predict-
ing the major warming that occurred during the period. The poor forecast of
wave propagation is associated with small errors in the predicted zonal wind
in the upper troposphere.
If
upward propagation is not well predicted in the
lower stratosphere by the 15-layer version because of the particular configu-
ration of the flow, and in the 9-layer version for the effects of the model’s top,
then the 15- and 9-layer forecasts should be similar in the middle and lower
troposphere. We conclude that, although the impact of the model’s top on
forecasts is largest at the uppermost model layers in all
cases,
it can be
significant in the middle and lower troposphere in some cases.
The limited number of
cases
does not allow us to estimate the overall gain
in predictability achieved by adding
6
more layers above the top of the
9-layer model. Such a gain might be small for this model, since the 15-layer
version might lose its useful predictability before the differences between
15-
and 9-layer forecasts became significant. In addition, the predictability of
some specific phenomena might be artificially increased by the lowered
model top. However, as models become more accurate, the presence of the
unavoidable errors due to artificial boundaries will become more apparent.
ACKNOWLEDGMENTS
Mr.
Joseph Spahr provided invaluable programming help. We are grateful
to
Ms.
Julia
Lueken and
Mr.
David Silverfarb
for
technical assistance
in
preparing the manuscript. This
research
was
supported by the
office
of Naval Research, through the Naval Environmental
Prediction Research Facility under Contract NOOO14-80-K-0947, and jointly by the National
Science Foundation and the National Oceanic and Atmospheric Administration under Grant
ATM 82 182
15
and the Naval Environmental
Prediction
Research Facility, Monterey, Califor-
nia, under Program Element 62759N,
Project
WF59-55
1,
“Meteorological Models
and
Predic-
tions.”
NUMERICAL FORECASTS
OF
EVENTS
413
REFERENCES
Arakawa, A., and Schubert, W. R. (I 974). Interaction of a cumulus cloud ensemble with the
Bengtsson,
L.
(198
1).
Numerical prediction of atmospheric blocking: A case study.
Tellus
33,
Butchart, N., Clough,
S.
A., Palmer,
T.
N., and Trevelyan, P. J. (1982). Simulations of an
observed stratospheric warming with quasigeostrophic refractive index
as
a
model diag-
nostic.
Q.
J. R. Meteorol.
Sm.
108,475
-
502.
Derome, J. (198 I). On the average errors of an ensemble of forecasts.
Amos.-Ocean
19,103-
127.
Katayama, A. (1972). A simplified scheme for computing radiative transfer in the troposphere.
“Numerical Simulation
of
Weather and Climate.” Tech. Rep. No. 6, Dept. of Meteorol-
ogy,
University of
California,
Los
Angeles.
Lambert,
S.
J. (1980). The sensitivity of tropospheric numerical weather forecasts to increased
vertical resolution and the incorporation of stratospheric
data.
Amos.-Ocean
18,53-64.
Lindzen, R.
S.,
Batten, E.
S.,
and Kim, J. W. (1968). Oscillations in atmospheres with tops.
Mon. Weather Rev.
96,
133
-
140.
McIntyre, M.
E.
(
1982). How well do we understand the dynamics of stratospheric warmings?
J.
Meteorol.
Soc.
Jpn.
60,37-65.
Matsuno,
T.
(1970). Vertical propagation of stationary planetary waves in the winter Northern
Hemisphere.
J. Atmos.
Sci.
27,871-883.
Mechoso,
C.
R., Suarez, M. J., Yamazaki, K., Spahr, J. A., and Arakawa, A. (1 982). A study of
the sensitivity of numerical forecasts to an upper boundary in the lower stratosphere.
Mon.
Weather Rev.
110, 1984- 1993.
Mechoso, C.
R.,
Yamazaki,
K.,
Kitoh, A., and Arakawa,
A.
(1985). Numerical forecasts of
stratospheric warming events during the winter of 1979.
Mon.
Weather Rev.
113,
1015-
1029.
Palmer, T. N.
(1
98
1).
Diagnostic study of
a
wavenumber-2 stratospheric sudden warming in a
transformed Eulerian-mean formalism.
J.
Atmos.
Sci.
38,844-855.
Quiroz, R.
S.
(1979). Tropospheric-stratospheric interaction in the major warming event of
January-February 1979.
Geophys. Res. Lett.
6,645 -648.
Schlesinger, M. E. (1976). A numerical simulation of the general circulation of atmospheric
ozone. Ph.
D.
thesis, Dept. Atmos. Sci., University
of
California,
Los
Angeles.
Schlesinger, M. E., and Mintz, Y. (1979). Numerical simulation ofozone production, transport
and distribution with a global atmospheric general circulation model.
J. Atmos.
Sci.
36,
Simmons,
A.
J., and Striifing, R. (1983). Numerical forecasts of stratospheric warming events
using a model with a hybrid vertical coordinate.
Q.
J.
R. Meteorol.
SOC.
109,8
I
-
1
1
1.
Suarez, M. J., Arakawa, A., and Randall, D.
A.
(1983). The parameterization of the planetary
boundary layer in the UCLA general circulation model: Formulation and results.
Mon.
Weather Rev.
111,2224-2243.
large-scale environment, Part
I.
J. Atmos.
Sci.
31,674-701.
19-42.
1325-1361.