close

Вход

Забыли?

вход по аккаунту

?

857

код для вставкиСкачать
INTERNATIONAL JOURNAL OF CLIMATOLOGY
Int. J. Climatol. 18: 473–504 (1998)
THE MEAN STRUCTURE AND TEMPORAL VARIABILITY OF THE
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
IAN SIMMONDSa and DAVID A. JONESb,*
School of Earth Sciences, Uni6ersity of Melbourne, Park6ille, Victoria 3052, Australia
b
Bureau of Meteorology Research Centre, GPO Box 1289K, Melbourne, Victoria 3001, Australia
a
Recei6ed 28 March 1997
Re6ised 3 No6ember 1997
Accepted 4 No6ember 1997
ABSTRACT
The semiannual oscillation of the pressure and mid-tropospheric baroclinicity in the southern extratropics has been
studied using 21 years (1973–1993) of numerical Southern Hemisphere analyses and long series of pressure data from
a number of mid- and high latitude stations. Using recent data and techniques not previously documented, this study
has verified that the semiannual oscillation is an important but highly variable feature of the annual cycle of pressure
and meridional temperature gradient. In the pressure, the half-yearly wave attains its greatest amplitudes in the
mid-latitude ocean basins and on the Antarctic periphery, with a minimum near 55°S. The semiannual oscillation of
the temperature gradient is strongest near 60°S, where it explains typically 50% of the mean annual variance of
monthly data, and the harmonic maxima (strongest gradients) occur during the transition seasons.
Analysis of the temporal behaviour of the half-yearly wave has revealed considerable variability on interannual to
decadal time scales. The comparison and correlation of the temporal variability of the mid- and high latitude
components of the semiannual oscillation of the pressure has revealed that these are statistically related and the
strength of the relationship is greatest for variations occurring on longer time scales.
In both the mid- and high latitudes the temporal variability of the semiannual oscillation of pressure has been
found to be statistically related to the variability of the high latitude temperature gradient. These observations suggest
that the differing annual cycles of temperature in the mid- and high southern latitudes not only give rise to the
semiannual oscillation of the pressure in the mean, but, in addition, the variability of this dynamic phenomenon is
linked to the variability of the thermal gradients. These findings suggest that a key to understanding the temporal
variability of pressure in the southern extratropics on annual to decadal time scales lies in the understanding of the
large scale variability of the temperature gradients. © 1998 Royal Meteorological Society.
KEY WORDS: Southern
Hemisphere; Antarctic; half-yearly wave; semiannual oscillation; sea-level pressure; temperature gradient;
variability; Fourier analysis
1. INTRODUCTION
The semiannual oscillation (SAO) of the pressure in the Southern Hemisphere (SH) mid- and high
latitudes is an important and large scale feature of the atmospheric annual cycle, resulting from the
differing annual cycles of temperature in the mid-latitude ocean and Antarctic regions (e.g. van Loon,
1967, 1972; Meehl, 1991). Described as early as the 1930s (see van Loon, 1967), this oscillation is
understood to result from the complex thermodynamics and dynamics of the SH ocean–atmosphere
system, which results in the mid-tropospheric meridional temperature gradients (baroclinicity) between
about 50°S and the Antarctic coast displaying a significant half-yearly wave (van Loon, 1967; Meehl,
1991).
In turn, the SAO of the meridional temperature gradient in the high southern latitudes results from the
differing annual cycles of the temperatures over the Antarctic continent and the mid-latitude oceans. Over
* Correspondence to: Bureau of Meteorology Research Centre, GPO Box 1289K, Melbourne, Victoria 3001, Australia. e-mail:
d.jones@bom.gov.au
CCC 0899–8418/98/050473 – 32$17.50
© 1998 Royal Meteorological Society
474
I. SIMMONDS AND D.A. JONES
the oceans near 50°S cooling in the autumn season is more rapid than warming in spring, while the reverse
is true near 65°S (van Loon, 1967, 1972; Meehl, 1991). This difference results in a marked second
harmonic in the meridional temperature gradient between typically 50 and 65°S which in turn modulates
the annual cycles of the pressure and wind.
Despite the importance of the SAO in the SH, there have been few articles devoted to its study since
that of van Loon (1967), and these have been mostly based on short periods of data. Van Loon (1972)
making use of hemispheric analyses and time series of raw station data studied the annual cycles of the
pressure, temperature, and wind fields across the SH. Hsu and Wallace (1976) and White and Wallace
(1978) used early gridded data to investigate the annual and semiannual cycles of the pressure and surface
temperature across both hemispheres. More recently, Meehl (1991) has examined the mechanisms of the
SAO, supporting the hypothesis of van Loon (1967) that its origins lie in the differing annual cycles of the
temperature in the Antarctic and mid-latitude ocean regions. Furthermore, Meehl observed a marked
SAO in the transient eddy activity in the southern ocean latitudes, an observation independently
confirmed by Trenberth (1991) and Jones (1994). The observed association between the SAOs of the
pressure and transient eddy activity was such that the intensification and poleward shift of the
circumpolar trough during the equinox periods was found to be associated with maxima of the transient
eddy momentum flux convergence. Synoptic evidence for a link between the SAO of the baroclinicity and
the annual cycle of the extratropical eddy activity has been provided by Carleton (1981), Howarth (1983),
and Jones (1994) who have each observed a half-yearly wave in the distribution of cyclones in the midand high southern latitudes. Further, it has been shown that the SAO is evident in precipitation data from
coastal Antarctic stations (Turner et al., 1997), being associated with the half-yearly wave in high latitude
cyclone activity.
The interannual variability of the SAO of the mean sea level pressure (MSLP) and zonal wind has been
examined by van Loon and Rogers (1984) who found that while the half-yearly wave displays considerable interannual variability, and may in individual years be weaker than one or more of the other
harmonics, the SAO dominates the long-term annual cycles of the pressure and zonal geostrophic wind
due to the stability of its phase relative to the other harmonics. Van Loon et al. (1993) and Hurrell and
van Loon (1994) have examined the variability of the MSLP SAO on interannual to decadal time scales
and found significant low frequency variability. The SAO was found to have weakened during the late
1970s as a result of a change in the monthly means of the pressure, chiefly during the second half of the
year. A delay in the weakening of the low level polar vortex until November was reflected in a marked
weakening in the semiannual cycle of the pressure in mid-latitudes, while in the high latitudes the SAO
showed a less dramatic, but nonetheless important, reduction. It is noteworthy that in the studies of the
SAO variability, little attention has been paid to the variability of the thermal gradient.
The purpose of this paper is to consolidate this scientific knowledge about the climatological SAO, and
to examine in a systematic manner its variability using recent digital analyses and raw station data. This
study may be divided into two sections. In the first, we update the present understanding of the time-mean
SAO and fully document the mid-tropospheric temperature component of this phenomenon using time
series considerably longer than those used in the aforementioned climatological studies. Secondly, we
address the stability and temporal variability of the half-yearly wave. Importantly, we comprehensively
examine the variability of both the pressure and baroclinicity components of the SAO, and their
interdependency.
2. THE DATA AND ANALYSIS PROCEDURES
The data employed in this study are the numerical SH analyses of the Australian Bureau of Meteorology
and long-time series of pressure from widely scattered stations in the SH mid- and high latitudes.
The observational gridded analyses are a subset of the twice-daily Australian Bureau of Meteorology
Southern Hemisphere analyses. These analyses (hereafter referred to as the ASH analyses) are analyzed on
a 47× 47 polar stereographic grid over the SH, giving an effective resolution of approximately 500 km at
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
475
60°S (Guymer, 1986). This hemispheric data set has been used in many studies, including those of
Trenberth (1981), Le Marshall et al. (1985), Jones and Simmonds (1993, 1994), van Loon et al. (1993),
Hurrell and van Loon (1994), and Jones (1994). The analysis procedures have remained essentially
unchanged during the period of study (Seaman, personal communication, 1997), and are detailed in Le
Marshall et al. (1985). The quality of the analyses has been discussed by Trenberth (1979, 1981), Le
Marshall et al. (1985), among others, and these data have been found to be generally reliable for the
description of large scale atmospheric features.
We have extracted the 2300 UTC MSLP and 500 hPa temperature analyses for the 21 years 1973–1993.
These data were then interpolated onto a 5°× 5° latitude–longitude grid covering the SH using bicubic
spline interpolation. The monthly averages for each of the calendar months in each of the 21 individual
years were generated from these daily analyses.
The station data used in this study to augment the gridded analyses have consisted of long series of
pressure observations from the Australian Antarctic stations at Casey, Macquarie Island, and Mawson
(see Russell-Head and Simmonds, 1993), while monthly means for Faraday, Halley Bay, Scott Base,
Campbell, Chatham, and Marion Islands were obtained from the Oak Ridge National Laboratory (6ide
Jones and Limbert, 1989; Vose et al., 1992). These data date back as early as 1930, allowing the study of
the SAO prior to the advent of the analyses.
The first (yearly wave y1) and second (half-yearly wave y2) harmonics of the annual cycle have been
determined through the Fourier analysis of the yearly time series of data. This analysis has been
performed for monthly data (monthly averages). The yearly and half-yearly waves of the year long data
series have thus been defined as,
y1 =A1 cos[t −f1]
(1)
y2 =A2 cos[2 ×t −f2]
(2)
where A1 and A2 are the amplitudes of the harmonics. f1 and f2 are the phase angles which determine
the times (values of t) at which the extremes of the first and second harmonics occur, while t varies from
0 to 2p over the course of a year. For presentation purposes, these phases have been converted into
months such that the value gives the month in which the maximum of the given harmonic occurs (0
indicates first of January etc.).
3. THE SEMIANNUAL OSCILLATION OF THE MONTHLY MEAN PRESSURE
To highlight the cycle of the pressure across the SH, Figure 1 shows the annual cycle of the zonal average
MSLP, averaged for the 21 years 1973 – 1993, together with the cycles of the first and second harmonics.
The SAO is clearly apparent in the annual cycle of the MSLP in both the mid- and high southern
latitudes. In the high latitudes (60 – 75°S) the monthly pressure attains its lowest values during the
transition seasons, and it is highest during mid-summer with a secondary maximum in mid-winter (Figure
1(a)). The half-yearly wave has an amplitude of more than 2 hPa in the high latitudes, with the extremes
of the oscillation centred on the equinox (March and September) and solstice (June and December)
months. In the mid-latitudes the amplitude of the SAO is somewhat smaller, while the phase is shifted
from an equinox minimum, to a solstice minimum. The comparison with the first harmonic of the mean
annual cycle (Figure 1(b)) reveals that the second harmonic is the stronger of the two between 35 and 75°S
(except near 60°S), while near the pole and in the low latitudes the reverse is true.
To highlight the meridional structure of the half-yearly wave of the MSLP, in Figure 2 is shown the
amplitude, percentage of the annual variance explained, and the phase of the second harmonic of the
mean annual cycle of the zonally averaged monthly pressure. Overlaid on the figures for the amplitude
and percentage variance explained is the zonal averages of these as computed at each individual spatial
(5° × 5°) point. It will be appreciated that these will differ if there is variation of the phase with longitude,
and hence these curves provide a measure of the variability of the phase of the SAO with longitude.
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
476
I. SIMMONDS AND D.A. JONES
The amplitude of the second harmonic has a mid- and high latitude peak with a marked minimum near
55°S. In the high latitudes the amplitude of the second harmonic of the zonally averaged pressure and the
zonal average of the amplitude are very similar and both approach 3 hPa. This similarity indicates that
there is little longitudinal variation of the phase in the high latitudes. In the mid-latitudes the amplitude
of the SAO is near 1.5 hPa, and, again, the amplitude of the SAO of the zonally averaged pressure is
similar to the zonal average of the amplitude.
Figure 2(b) reveals that the SAO of the zonally averaged MSLP explains more than 50% of the mean
annual variance in the mid- and high latitudes. In contrast with the amplitude, there is a sizeable
difference between the percentage of the variance explained by the SAO of the zonally averaged MSLP
and the zonal average of the variance explained in the high and, in particular, the mid-latitudes. The
increased importance of the second harmonic in explaining the variance of the annual cycle of the zonal
average pressure compared to the zonal average of the variance explained indicates that the phase of the
second harmonic varies considerably less with longitude than the other harmonics.
The phase of the SAO of the zonally averaged pressure (Figure 2(c)) reveals that the half-yearly wave
attains its maxima in the solstice periods in the high latitudes and during the transition seasons in the
mid-latitudes with a phase reversal near 55°S. Both the mid- and high latitude components of the SAO
show little variation in phase with latitude, varying by less than 2 weeks within these two zones.
To highlight the spatial characteristics of the half-yearly wave of the pressure, the geographical
distributions of its amplitude, percentage of the mean annual variance explained, and the phase in the
21-year mean annual cycle of the monthly averaged MSLP are shown in Figure 3.
The amplitude and percentage of the mean annual variance explained by the SAO attain their largest
values on the Antarctic periphery between 65 and 75°S, and in the mid-latitude ocean basins between 30
and 45°S. A general minimum (particularly of the amplitude) lies between 55 and 60°S, marking the
region where the phase of the harmonic shifts from an equinox maximum in the mid-latitudes to an
equinox minimum at high latitudes (Figures 1 –3(c)). The comparison of these figures indicates a weaker
SAO than that documented in van Loon (1972) and Xu et al. (1990), consistent with the findings of
Hurrell and van Loon (1994).
Figure 1. The mean annual cycle (a), first harmonic (b), and second harmonic (c) of the zonally monthly averaged MSLP. The
isopleth intervals are 2.5, 0.5, and 0.5 hPa, respectively
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
477
Figure 1 (Continued)
On a regional scale, the SAO at high latitudes is strongest near the Ross Sea. It is noteworthy that this
area is also characterized by large interannual variability of pressure, geopotential height, and cyclone
activity (e.g. Le Marshall et al., 1985; Jones, 1994). In the mid-latitudes the amplitude is greatest in the
central Atlantic and Indian Oceans. In the Pacific Ocean the mid-latitude component of the SAO is
comparatively weaker. This weakness is related to the secondary storm track in the Pacific Ocean (e.g. van
Loon, 1967; Jones and Simmonds, 1993; Jones, 1994; Sinclair, 1994), which shows a half-yearly
modulation which is in approximate phase with that of the cyclone activity and pressure at high latitudes.
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
478
I. SIMMONDS AND D.A. JONES
The phase of the SAO of the monthly average MSLP (Figure 3(c)) has a marked shift between 50 and
60°S, with the first peak occurring between mid-December and mid-January in the high latitudes, and
during March or April in the mid-latitudes. In both zones the phase of the second harmonic displays little
longitudinal structure, typically varying by less than 1 month around a given latitude circle.
Figure 2. The meridional distribution of the amplitude (a), percentage of the annual variance explained (b), and phase (c) of the
second harmonic of the mean annual cycle of the zonally monthly averaged MSLP. The units are hPa, %, and months, respectively
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
479
Figure 2 (Continued)
4. THE SEMIANNUAL OSCILLATION OF THE 500 hPa MERIDIONAL TEMPERATURE
GRADIENT
While the SAO in the meridional gradient of the mid-tropospheric temperature fields has been recognised as being responsible for the SAO in the MSLP and the zonal geostrophic wind for a number of
years (e.g. van Loon, 1967, 1972; Meehl, 1991), the structure of this feature has received little attention.
Figure 4 shows the amplitude, percentage of the mean annual variance explained, and the phase of the
SAO of the mean annual cycle of the monthly zonally averaged meridional temperature gradient at 500
hPa (computed using centred differencing over 5° latitude zones). The temperature gradient shows a
prominent half-yearly wave between 50 and 65°S which accounts for about 50% of its annual variation.
Within this zone we deduce the phase of the harmonic is relatively constant with longitude, as the
amplitude of the SAO of the zonally averaged gradient and the zonal average of the amplitude of the
SAO are similar.
The phase of the SAO of the zonally average meridional temperature gradient reveals that between 50
and 65°S, the half-yearly wave has its maxima during the transition seasons. This phase is consistent
with the second harmonic of the pressure in the mid-latitudes, and broadly phase shifted from that of
the pressure at high latitudes. During the transition seasons the maxima of the SAO of the meridional
temperature gradient are reflected in decreased pressures in the high latitudes and increased pressure in
the mid-latitudes, while the reverse is the case during the extreme seasons (van Loon, 1972).
The geographical distribution of the amplitude and the percentage variance explained by the second
harmonic of the 21-year average 500 hPa meridional temperature gradient (Figure 5) exhibits a
maximum in the high latitudes. The peak lies near 55°S in the Indian Ocean and between 60 and 65°S
in the Pacific and Atlantic Oceans, and generally explains some 40–70% of the annual variance. It is of
interest that the maximum of the second harmonic is confined to a relatively narrow zone of typically
5 –10°, thinner than the 15° latitude range (50–65°S) which has historically been used to diagnose the
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
480
I. SIMMONDS AND D.A. JONES
high latitude temperature gradient (e.g. van Loon, 1967; Meehl, 1991). It is the case, however, that while
the maximum of the SAO of the meridional temperature gradient at a given longitude is confined to a
rather narrow latitude zone, the latitude of this maximum varies by some 10° between the eastern and
Figure 3. The amplitude (a), percentage of the annual variance explained (b), and phase (c) of the second harmonic of the mean
annual cycle of the monthly averaged MSLP. The isopleth intervals are 0.5 hPa, 20%, and 1 month, respectively. Shading in (a) and
(b) indicates values greater than 2 and 40, respectively. Shading in (c) indicates a transition season maximum
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
481
Figure 3 (Continued)
western hemispheres, making the definition of a single zonal index to measure the baroclinicity component
of the SAO difficult.
The comparison of Figure 5(a) with that for the MSLP (Figure 3(a)) reveals a tendency for local high
latitude maxima in the MSLP SAO to occur to the southeast of amplitude maxima of those of the
temperature gradient. This suggests that the influence of local maxima of the temperature gradient SAO
have most impact downstream in the MSLP field, presumably through baroclinic eddy activity. There is
rather little evidence to link the mid-latitude amplitude maxima of the MSLP to maxima in the
temperature gradient. This may reflect the impact of the subtropical continents in weakening the SAO in
their vicinity.
5. THE SENSITIVITY OF THE HARMONIC ANALYSIS ON THE RECORD LENGTH
The earlier mentioned studies of the SAO in the southern extratropics have made use of data of varying
temporal lengths. However, the SAO is variable in time, meaning that its importance and characteristics
are dependent on the length of the data on which the analysis is performed. Hence, it would appear that
before we can speak sensibly of the amplitude of the temporal harmonics and, especially, of the variance
each explains, we must know how many years have been considered in the averaging process. For
example, the variance explained by the SAO would be expected to increase as the length of the record
increases and the level of noise decreases. One purpose of our study is to quantify the sensitivity of the
Fourier analysis of the annual cycle of pressure in the SH to the length of the averaging period over which
the annual cycles have been derived. This dependence has potentially important consequences for studies
such as those of Xu et al. (1990), Meehl (1991), and Tzeng et al. ( 1993) for which model verification is
being made against observations, and for which the model and observational data may have different
lengths.
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
482
I. SIMMONDS AND D.A. JONES
Figure 4. The meridional distribution of the amplitude (a), percentage of the annual variance explained (b), and phase (c) of the
second harmonic of the mean annual cycle of the zonally monthly averaged 500 hPa meridional temperature gradient. The units are
K/degree latitude, %, and months, respectively
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
483
Figure 4 (Continued)
We have computed the mean of the amplitude and the percentage of the variance explained by the first
four harmonics of the pressure cycle at the representative stations of Casey, Faraday, Macquarie Island,
and Mawson using monthly data. The mean annual cycle of the pressure has been generated for all
possible consecutive periods for each of these stations and then harmonically analyzed. The mean
amplitude and percentage of the variance explained has then been taken to be the average of these for
each of the individual mean annual cycles. Figure 6 shows the mean amplitude of the first four harmonics
of the annual cycle of the pressure at these stations as a function of the averaging period over which the
mean annual cycle has been calculated. One of the consequences of interannual variability in phase is that
for each harmonic the mean amplitude is greatest when the analysis is performed on the individual annual
cycles, and the amplitudes of these meaned. Beyond about 5 years, the amplitude of the first and second
harmonics is only weakly dependent on the averaging period, while those of the third and fourth
harmonics show a general decrease with increasing averaging period, indicating much variability in their
phase. These figures reflect the fact that all four harmonics may have a significant amplitude in an
individual year, but that the higher harmonics are considerably less important in the mean, due to the
interannual variability of their phase.
The dependence of the percentage of the variance explained by the first four harmonics of the annual
cycle on the averaging period used to derive the mean annual cycle confirms the relatively rapid
convergence of the first and second harmonics towards near fixed values (Figure 7). The percentage of the
variance explained varies little beyond an averaging period of about 10 years.
6. THE TEMPORAL VARIABILITY OF THE SEMIANNUAL OSCILLATION OF PRESSURE
IN THE GRIDDED DATA
While the SAO is a climatologically important feature of the annual cycle of pressure in the southern
extratropics it is known to display considerable temporal variability (van Loon and Rogers, 1984; van
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
484
I. SIMMONDS AND D.A. JONES
Figure 5. The amplitude (a) and percentage of the annual variance explained (b) by the second harmonic of the mean annual cycle
of the monthly averaged 500 hPa meridional temperature gradient. The isopleth intervals are 0.05 K/degree latitude and 20%,
respectively
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
485
Figure 6. The dependence of the mean amplitude of the first four harmonics of the annual cycle of the monthly pressure at Casey
(a), Faraday (b), Macquarie Island (c), and Mawson (d) on the length of the period used to derive the mean annual cycle. The units
are hPa
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
486
I. SIMMONDS AND D.A. JONES
Figure 6 (Continued)
Loon et al., 1993; Hurrell and van Loon, 1994). To further document this interannual variability, in
Figure 8 is shown the running time series (extracted from a 12-month ‘window’ of pressure) of the
amplitude, percentage of the annual variance explained, and the phase (at 50 and 70°S only) of the SAO
of the zonally averaged pressure.
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
487
Figure 7. The dependence of the mean percentage of the annual variance explained by the first four harmonics of the annual cycle
of the monthly pressure at Casey (a), Faraday (b), Macquarie Island (c), and Mawson (d) on the length of the period used to derive
the mean annual cycle. The units are %
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
488
I. SIMMONDS AND D.A. JONES
Figure 7 (Continued)
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
489
The SAO exhibits considerable variability with its amplitude in the mid-latitudes varying from near 3
hPa during 1976 and 1982 to less than 0.5 hPa during short periods in 1980, 1985/1986, and 1991. In the
high latitudes the SAO has exhibited even greater variability with the amplitude having varied from more
than 6 hPa during 1976 and 1977 to less than 0.5 hPa in 1979 and 1988. Evident too is a consistency of
the variations with latitude. The time synchronous correlation between the amplitudes at 50 and 70°S is
0.41 which is significantly different from zero at the 95% confidence level.
Figure 8. Time series of the amplitude (a), percentage of the variance explained (b), and phase (at 50 and 70°S only) (c) of the
second harmonic of the annual cycle of the zonally monthly averaged MSLP. The isopleth intervals are 2 hPa and 25%, respectively,
with an additional 1 hPa isopleth added in (a). The units in (c) are months
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
490
I. SIMMONDS AND D.A. JONES
Figure 8 (Continued)
The time series for the percentage of the annual variance explained by the SAO (Figure 8(b)) displays
similar variability on interannual and longer time scales. In both the mid- and high latitudes the SAO
explains in excess of 75% of the variance in some 12-month periods. On other occasions the percentage
of the variance explained has been less than 25%, and during brief periods both components of the SAO
have virtually disappeared. Again, there is a significant correlation between the time series at 50 and 70°S
(0.45), with the strength of the SAO during the mid/late 1970s, weakness around 1980, and many of the
prominent variations during the remaining years reflected in both zones.
The time series of the phase (in months) at 50 and 70°S (Figure 8(c)) reveal that the SAO, while
generally peaking during the solstice periods in the high latitudes, and the equinox periods at mid-latitudes, has varied considerably during the 21 years of this study. Again, many of the features of the two
series are shared at both latitudes, suggesting that the variations of the phase in these zones frequently
occur in sympathy. It is of particular interest that the marked perturbations to the phase during the early
1980s, during which time the SAO was very weak, is shared by both latitudes.
To highlight the lower frequency variability of the SAO in the ASH data, the time series of the 5-year
average (rather than the individual 12 month periods considered above) annual cycles of the zonally
averaged pressure have been harmonically analyzed. The time series of the amplitude, percentage of the
variance explained, and phase of the second harmonic of these data are shown in Figure 9. These reveal
considerable low frequency variability during the 21 years of this study. In general, the SAO was strongest
during the 1970s and weakened to a minimum near 1980. In the high latitudes the amplitude recovered
in the mid-1980s, before again dropping to low values during the final years considered here. In the
mid-latitudes the amplitude has shown modest variability subsequent to 1980 about a reduced mean
amplitude. The cursory comparison of the series in the mid- and high latitudes suggests a link, with the
amplitudes at 50 and 70°S possessing a correlation of 0.76.
The series for the percentage of the variance explained (Figure 9(b)) highlights the significant change to
the annual cycle of the MSLP during the late 1970s, with the strength of the SAO decreasing dramatically,
particularly in mid-latitudes. This marked weakening has been noted previously by van Loon et al. (1993)
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
491
and Hurrell and van Loon (1994) while, importantly, our data reveal that the general weakness has not
been significantly reversed in more recent years. The time series of the phase of the SAO of the 5-year
average zonal mean pressure at 50 and 70°S (Figure 9(c)) reveal that the characteristic solstice and
equinox maxima of the high and mid-latitude components of the SAO have been broadly preserved, with
little systematic variation during the study period.
Figure 9. Time series of the amplitude (a), percentage of the variance explained (b), and phase (at 50 and 70°S only) (c) of the
second harmonic of the 5-year mean annual cycle of the zonally monthly averaged MSLP. The isopleth intervals are 2 hPa and 25%,
respectively, with an additional 1 hPa isopeth added in (a). The units in (c) are months
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
492
I. SIMMONDS AND D.A. JONES
Figure 9 (Continued)
7. THE TEMPORAL VARIABILITY OF THE SEMIANNUAL OSCILLATION OF THE
MERIDIONAL TEMPERATURE GRADIENT
There is strong evidence to link the SAO of the pressure in the southern extratropics to the half-yearly
cycle of the high latitude baroclinicity (van Loon, 1967, 1972; Xu et al., 1990; Meehl, 1991; Jones, 1994).
A powerful test of this relationship is to determine whether the interannual variations of the SAOs of the
MSLP and temperature gradient are related.
As has been shown previously, the selection of a temperature index for the diagnosis of the thermal
component of the SAO is problematic due to the concentration of the second harmonic of the meridional
temperature gradient in a relatively narrow latitude band and the tendency for this zone to vary by some
10° of latitude between the eastern and western halves of the SH. With these considerations in mind, the
thermal component of the SAO has been studied using the zonally averaged 500 hPa meridional
temperature gradients at 57.5 and 62.5°S and by the temperature difference between 50 and 65°S.
In Figure 10 is shown the time series of the amplitude, percentage of the variance explained, and the
phase of the SAO of the yearly 500 hPa meridional temperature gradients at 57.5 and 62.5°S. Supporting
the earlier observations of considerable variability in the equivalent series for the MSLP, the SAO of the
thermal gradient is highly variable. The amplitude at both latitudes has exceeded 0.15 K/degree latitude
during individual 12-month periods during which times it has been responsible for in excess of 60% of the
variance, while on other occasions it has been very weak and virtually disappeared. As might be expected,
the time series of the amplitudes at 62.5 and 57.5°S are similar, though the amplitude at the lower latitude
is typically a little larger.
The temporal correlation between the amplitudes of the baroclinicity components of the SAO
(including that for the 50 – 65°S belt) and those of the pressure at 50 and 70°S for yearly data are at least
0.57 (see Table I) and are as high as 0.64, coefficients which are all significant at the 95% level. These
structures and correlations are very strong evidence suggesting that the variability of the second harmonic
of the zonally averaged pressure is related to the variability of the semiannual cycle of the large scale
temperature gradient.
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
493
Figure 10. Time series of the amplitude (a), percentage of the variance explained (b), and phase (c) of the second harmonic of the
annual cycle of the zonally monthly averaged 500 hPa meridional temperature gradient at 57.5 and 62.5°S. The units are K/degree
latitude, %, and months, respectively
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
494
I. SIMMONDS AND D.A. JONES
Figure 10 (Continued)
The series of the phase of the SAO of the 500 hPa meridional temperature gradient at 57.5 and 62.5°S
(Figure 10(a)) indicate rather little variability with the variations being mostly less than 1 month. An
interesting observation is that the marked perturbations to the phase of the second harmonic of the MSLP
which occurred during 1980/1981 and 1988/1989 have corresponding features in the series for the
temperature gradient at 62.5 and, to a lesser extent, 57.5°S.
To document the lower frequency variability of the SAO of the temperature gradient, Figure 11 shows
the time series of the amplitude, percentage of the variance explained, and phase of the 5-year running
mean annual cycles. The amplitude series shows high values during the 1970s, a decrease to a relative
minimum near 1980, and a partial recovery during the early/mid 1980s. Subsequently, the amplitude has
shown modest variability. The marked weakening of the SAO during the late 1970s is consistent with that
evident in the MSLP (Figure 9(a)).
Table I. The temporal correlation of the amplitude of the SAO of the MSLP at 50 and
70°S with the amplitude of the SAO of the meridional temperature gradients at 57.5°S,
62.5°S, and the bulk temperature difference 50°S−65°S
Latitude of MSLP
50°S
70°S
50°S
70°S
50°S
70°S
© 1998 Royal Meteorological Society
1-Year average
2-Year average
500 hPa meridional temperature gradient
0.60
0.63
0.61
0.71
500 hPa meridional temperature gradient
0.64
0.69
0.57
0.68
500 hPa temperature at 50°S−65°S
0.60
0.63
0.66
0.77
5-Year average
at 57.5°S
0.61
0.82
at 62.5°S
0.81
0.86
0.67
0.83
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
495
Figure 11. Time series of the amplitude (a), percentage of the variance explained (b), and phase (c) of the second harmonic of the
5-year mean annual cycle of the zonally monthly averaged 500 hPa meridional temperature gradient at 57.5 and 62.5°S. The units
are K/degree latitude, %, and months, respectively
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
496
I. SIMMONDS AND D.A. JONES
Figure 11 (Continued)
Examining further the relationship between the SAO of the MSLP and that of the temperature gradient
indicates that the temporal correlations of the respective amplitudes at the considered latitudes all exceed
0.60 (Table I). The persistence and strength of the associations suggest that the interannual and,
particularly, low frequency variability of the SAO of the zonally averaged pressure is strongly related to
that of the SAO of the 500 hPa temperature gradient. An important deduction from this is that, even
though the MSLP SAO has undergone variations in strength over the last two decades, the link to the
cycle in the temperature appears to have been largely maintained.
We note that the SAO of the MSLP at 50°S is more highly correlated with that in the temperature
gradient at 62.5°S than that at 57.5°S. This is probably due to the fact that the variability of the
mid-latitude component of the half-yearly wave of the MSLP has been greatest in the Pacific region (see
Hurrell and van Loon, 1994), where the maximum of the SAO of the meridional temperature gradient
occurs furthest south.
The series for the percentage of the variance explained confirm the considerable low frequency
variability of the second harmonic during the 21 years of this study (Figure 11(b)). The percentage of the
variance explained by the SAO has varied from more than 70% to less than 40%. In contrast to the
amplitude and variance explained, the phase of the SAO at both latitudes has been quite stable during the
study period (Figure 11(c)) varying, for the most part, by less than 1 week.
Van Loon (1967) and Meehl (1991) have used the difference between the 500 hPa temperatures at 50
and 65°S as a measure of the thermal component of the SAO and while the time series of these are
structurally similar to those shown for the meridional temperature gradients at 57.5 and 62.5°S, it is of
interest to compare how these are related to those of the second harmonic of the MSLP. In Table I is
shown the correlation coefficients between the amplitude of the SAO of this bulk temperature measure
and the amplitude for the zonally average MSLP for yearly, 2-year average, and 5-year average data in
both the mid- and high latitudes. For the pressure at both latitudes and each averaging period these
correlations are comparable in magnitude to those for the temperature gradient at 62.5°S.
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
497
8. THE TEMPORAL VARIABILITY OF THE SEMIANNUAL OSCILLATION IN STATION
DATA
A primary aim of this study has been to investigate the temporal variability of the large scale SAO on
interannual and longer time scales. To this end, the ASH data have proved valuable, revealing
considerable variability and that the variations of the half-yearly wave in the pressure are statistically
related to those in the high latitude temperature gradient. The ASH data, however, only provide the
opportunity to study this phenomenon over a period of slightly longer than two decades.
To complement the analyses discussed above monthly mean pressure time series have been obtained for
each of the four mid-latitude stations at Campbell, Chatham, Macquarie, and Marion Islands, and for the
four high latitude stations Casey, Faraday, Halley Bay, and Scott Base and these have been subjected to
temporal Fourier analysis. These eight stations are a subset of those which have been analyzed as part of
this study, but have been chosen for presentation because they provide a reasonable spatial coverage.
The time series of the amplitude of the second harmonic of the annual and 5-year mean annual cycles
of the pressure at the four mid-latitude stations are shown in Figure 12. Notably, each exhibits a marked
weakening of the SAO during the late 1970s or early 1980s consistent with that noted previously, though
the exact timing of this reduction varies geographically.
Taken over the entire data record the SAO has shown considerable variability through time at each
station, suggesting that the recent variations and, in particular, the marked weakening of the mid-latitude
component of the SAO around 1979 may not be unprecedented. In particular, each series suggests that
the SAO was similarly weak during the early 1960s, and while the series don’t all extend back beyond
1950, those at Chatham and Campbell Islands suggest the SAO was also weak during the 1940s.
At each of the higher latitude stations (Figure 13) the relative weakness of the SAO during the late
1970s and early 1980s is clearly apparent. It is of interest that for each of these stations there is the
suggestion that the mid 1970s may have, in fact, been a period with an anomalously strong SAO, and that
the SAO during the 1980s may be more typical of the recorded period at these stations. The series at each
station (except Faraday) suggest a further period of large amplitudes during the mid 1960s.
While the generalization from the individual stations to the hemispheric scale is difficult, the considerable temporal variability in these series and the correspondence between the stations and the ASH data
suggests that the variability of the SAO across a range of frequencies is not unique to the 21 years of ASH
analyses used in this study. At lower frequencies, while the weakening of the SAO during the 1970s is
marked, there is the suggestion that there have been other periods during which the SAO has undergone
similar reductions.
These observations highlight the dangers of using relatively short periods of data for the definition of
climate. In addition, the considerable variability of this important climatological feature cautions that
careful consideration be given to the techniques applied to gauging the ability of atmospheric models to
simulate higher order features of the climate such as the SAO, when these features display considerable
long-term variability.
9. DISCUSSION AND CONCLUSIONS
We have examined the time mean characteristics and temporal variability of the SAO of the MSLP and
500 hPa meridional temperature gradient using 21 years of SH analyses. These data cover a period
considerably longer than those used in the aforementioned climatological studies and provide a timely and
reliable update to these studies. The climatological SAO of the pressure has been found to be greatest in
the mid- and high latitudes with a relative minimum between 55 and 60°S. In the high latitudes the
amplitude of the second harmonic is typically greater than 3 hPa and it explains 50% or more of the mean
annual variance. In the mid-latitudes the amplitude of the SAO is between 1 and 2 hPa, while it accounts
for typically 50% and in places more than 70% of the mean annual variance.
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
498
I. SIMMONDS AND D.A. JONES
Figure 12. Time series of the amplitude of the second harmonic of the annual and five-year mean annual cycles of the pressure at
Campbell (a), Chatham (b), Macquarie (c), and Marion Islands (d). The units are hPa
The 500 hPa meridional temperature gradient has been found to exhibit a marked SAO in the Southern
Ocean, the maxima of which occurs during the transition seasons. In the eastern hemisphere this maximum
is near 55°S, while in the Pacific and Atlantic Oceans the maximum amplitude occurs between 60 and 65°S.
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
499
Figure 12 (Continued)
The study of the time variation of the SAO has revealed very considerable variability of amplitude and,
to a lesser extent, phase. The variations of this feature have been found to be often large scale and
coherent, which together with the comparison with the station data suggests that the findings of this study
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
500
I. SIMMONDS AND D.A. JONES
Figure 13. Time series of the amplitude of the second harmonic of the annual and five-year mean annual cycles of the pressure at
Casey (a), Faraday (b), Halley Bay (c), and Scott Base (d). The units are hPa
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
501
Figure 13 (Continued)
based on the ASH data are reliable. In both the mid- and high latitudes the SAO of the zonal mean
pressure has, on occasions, dominated the mean annual cycle explaining more than 75% of the annual
variance, while during individual 12 month periods the component in one or both latitude zones has
virtually disappeared. In the long-term the SAO in the ASH data was strongest during the 1970s,
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
502
I. SIMMONDS AND D.A. JONES
Figure 14. Time series of the amplitude of the second harmonic of the 5-year mean annual cycle of the zonally averaged 500 hPa
temperature at 50, 55, 60, and 65°S. The units are K
weakened towards the early 1980s and has remained relatively weak, particularly in the mid-latitudes,
during the early 1990s. It is noteworthy that these changes are consistent with simultaneous changes in
extratropical cyclone activity (see Jones, 1994), with the half-yearly waves of cyclone frequency and
central pressure having also weakened since the 1970s, particularly at mid-latitudes.
A similar temporal analysis of the thermal component of the SAO has revealed much variability.
During individual 12 monthly periods the second harmonic at 500 hPa has been responsible for in excess
of 75% of the annual variance in the vicinity of its maximum while, like that in the pressure, it has, on
occasions, been very weak and almost disappeared. At the lower frequencies, the second harmonic of the
temperature gradient has shown a general decrease during the study period, with this decrease being most
marked at 62.5°S where the contribution from the Pacific Ocean in which the SAO of the MSLP has
commensurately weakened most is greatest (see Hurrell and van Loon, 1994).
The temporal variations of the SAO of the temperature gradient in the high latitudes and those of the
MSLP half-yearly wave in the mid- and high latitudes have been found to be statistically related, with the
strength of these relationships greatest for the lower frequencies. For the series of the yearly amplitudes,
the correlations between each of the considered thermal descriptors of the SAO and those of the pressure
at 50 and 70°S have been found to be greater than 0.55. Notably, the correlations between the amplitudes
of the half-yearly waves of the temperature gradient at 62.5°S and the pressure at 50 and 70°S in the
5-year mean data have been found to be greater than 0.8.
As to what variations in the half-yearly cycles of the temperatures in the mid- and high latitudes have
given rise to the variability of the SAO of the meridional temperature gradient and, presumably, that of
the MSLP is an important question. The variations of the half-yearly cycle of the meridional temperature
gradient at high latitudes are due to variations which have occurred in both the phase and amplitude of
the half-yearly waves of the temperature in the mid-latitude ocean and Antarctic regions. Time series have
indicated that the observed variability has been largely due to differing variations in the amplitude of the
half-yearly wave of the temperatures at mid- and high latitudes, rather than due to variations of the
phase.
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
SEMIANNUAL OSCILLATION IN THE SOUTHERN EXTRATROPICS
503
In attempting to identify the causes of the variability in the temperature gradient, it makes sense to ask
which temperature components have had most impact on variations of these gradients. To this end, in
Figure 14 we show the time series of the amplitude of the SAO of the 5-yearly mean temperature at 50,
55, 60, and 65°S. These reveal that the changes in the amplitude are more marked at the higher latitudes,
and, in particular, the weakening of the SAO of the meridional temperature gradient during the late 1970s
is largely due to a reduction in the amplitude at the higher latitudes (60 and 65°S) which has not occurred
at the lower latitudes (50 and 55°S). The tendency for the greater variability to occur at higher latitudes
in this figure indicates that the variability of the SAO of the meridional temperature gradient is largely
due to variations at the higher latitudes (over the Antarctic continent and the sea ice regions). Whether
such changes in the atmospheric dynamics are due to variations in the local boundary conditions or in
response to remote changes is not clear, and demands further investigation.
Finally, the finding of statistically significant relationships between the thermal and pressure components of the large scale semiannual wave suggests that the dynamical mechanisms which give rise to the
wave in the pressure in association with that in the baroclinicity in the mean, are active in linking the
temporal variations of the thermal and pressure components of the SAO. Given the prominence of the
SAO, this coupling suggests that a key to understanding the variability of the mass and wind fields across
the SH lies in the understanding and documentation of the variability of the large scale temperature and
temperature gradients.
REFERENCES
Carleton, A.M. 1981. ‘Monthly variability of satellite-derived cyclonic activity for the Southern Hemisphere winter’, J. Climatol., 1,
21 – 38.
Guymer, L.B. 1986. ‘Procedures and concepts used in Southern Hemisphere Analyses at WMC, Melbourne’, Proceedings of the
Second International Conference on Southern Hemisphere Meteorology, Wellington, New Zealand, American Meteorology Society,
Boston, pp. 10 – 16.
Howarth, D.A. 1983. ‘An analysis of the variability of cyclones around Antarctica and their relationship to sea-ice extent’, Ann.
Assoc. Am. Geogr., 73, 519–537.
Hsu, C.P.F. and Wallace, J.M. 1976. ‘The global distribution of the annual and semiannual cycles in sea level pressure’, Mon. Wea.
Re6., 104, 1597 – 1601.
Hurrell, J.W. and van Loon, H. 1994. ‘A modulation of the atmospheric annual cycle in the Southern Hemisphere’, Tellus, 46,
325 – 338.
Jones, P.D. and Limbert, D.W.S. 1989. Antarctic Surface Temperature and Pressure Data, Oak Ridge National Laboratory, Report
No. 3215, Oak Ridge, TN, 29 pp.
Jones, D.A. and Simmonds, I. 1993: A climatology of Southern Hemisphere extratropical cyclones. Climate Dynam., 9, 131 – 145.
Jones, D.A. and Simmonds I. 1994. ‘A climatology of Southern Hemisphere anticyclones’, Climate Dynam., 10, 333 – 348.
Jones, D.A. 1994. ‘An objective study of Southern Hemisphere synoptic activity’, PhD Thesis, School of Earth Sciences, University
of Melbourne, Parkville, Australia, 392 pp.
Le Marshall, J.F., Kelly, G.A.M. and Karoly, D.J. 1985. ‘An atmospheric climatology of the Southern Hemisphere based on ten
years of daily numerical analyses (1972–82): I Overview’, Aust. Meteorol. Mag., 33, 65 – 85.
Meehl, G.A. 1991. ‘A reexamination of the mechanism of the semiannual oscillation in the Southern Hemisphere’, J. Climate, 4,
911 – 925.
Russell-Head, D. and Simmonds, I. 1993. Temporal structure of surface weather parameters at Casey, Da6is, Mawson and Macquarie
Island, Pub. No. 35, Meteorology Section, School of Earth Sciences, University of Melbourne, Parkville, Australia, 181 pp.
Sinclair, M.R. 1994. ‘An objective cyclone climatology for the Southern Hemisphere’, Mon. Wea. Re6., 122, 2239 – 2256.
Trenberth, K.E. 1979. ‘Interannual variability of the 500 mb zonal mean flow in the Southern Hemisphere’, Mon. Wea. Re6., 107,
1515 – 1524.
Trenberth, K.E. 1981. ‘Observed Southern Hemisphere eddy statistics at 500 mb: Frequency and spatial dependence’, J. Atmos. Sci.,
38, 2585 – 2605.
Trenberth, K.E. 1991. ‘Storm tracks in the Southern Hemisphere’, J. Atmos. Sci., 48, 2159 – 2178.
Turner, J., Colwell, S.R. and Harangozo, S. 1997. ‘Variability of precipitation over the coastal western Antarctic Peninsula from
synoptic observations’, J. Geophys. Res., 102, 13999–14007.
Tzeng, R.Y., Bromwich, D.H. and Parish, T.R. 1993. ‘Present-day Antarctic climatology of the NCAR community climate model
version 1’, J. Climate, 6, 205–226.
van Loon, H. and Rogers, J.C. 1984. ‘Interannual variations in the half-yearly cycle of pressure gradients and zonal wind at sea level
on the Southern Hemisphere’, Tellus, 36, 76–86.
van Loon, H., Kidson, J.W. and Mullan, A.B. 1993. ‘Decadal variation of the annual cycle in the Australian dataset’, J. Climate,
6, 1227 – 1231.
van Loon, H. 1967. ‘The half-yearly oscillations in the middle and high southern latitudes and the coreless winter’, J. Atmos. Sci.,
24, 472 – 486.
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
504
I. SIMMONDS AND D.A. JONES
van Loon, H. 1972. ‘Temperature, pressure, and wind in the Southern Hemisphere’, Meteorology of the Southern Hemisphere,
Meteorol. Monogr., Vol. 13, American Meteorology Society, Boston., pp. 25 – 100.
Vose, R.S., Heim, R., Schmoyer, R.L., Karl, T.R., Steurer, R.L., Eischeid, J.K. and Peterson, T.C. 1992. The global historical
climatology Network: Long-term Monthly Temperature, Precipitation, Sea Le6el Pressure, and Station Pressure Data, Report No.
3912, Oak Ridge National Laboratory, TN, 99 pp.
White, G.H. and Wallace, J.M. 1978. ‘The global distribution of the annual and semiannual cycles in surface temperature’, Mon.
Wea. Re6., 106, 901–906.
Xu, J.S., von Storch, H. and van Loon, H. 1990. ‘The performance of four spectral GCMs in the Southern Hemisphere: The January
and July climatology and the semiannual wave’, J. Climate, 3, 53 – 70.
© 1998 Royal Meteorological Society
Int. J. Climatol. 18: 473 – 504 (1998)
Документ
Категория
Без категории
Просмотров
2
Размер файла
880 Кб
Теги
857
1/--страниц
Пожаловаться на содержимое документа