Content uploaded by Piia Post
Author content
All content in this area was uploaded by Piia Post on Sep 16, 2014
Content may be subject to copyright.
BOREAL ENVIRONMENT RESEARCH 7: 281–289 ISSN 1239-6095
Helsinki 2 October 2002 © 2002
Circulation weather types and their inß uence
on temperature and precipitation in Estonia
Piia Post, Valdur Truija and Janno Tuulik
Department of Environmental Physics, University of Tartu, Ülikooli 18, 51014
Tartu, Estonia
Post, P., Truija, V. & Tuulik, J. 2002: Circulation weather types and their inß u-
ence on temperature and precipitation in Estonia. — Boreal Env. Res. 7: 281–
289. ISSN 1239-6095
An existing objective classiÞ cation scheme of the atmospheric circulation, where
daily circulation is determined through the strength, direction and vorticity of the
geostrophic ß ow has been applied over the Baltic Sea region for the time period of
1968–1997. The results at sea level and the higher isobaric levels of 500 hPa, 700 hPa
and 850 hPa are presented here. The analysis revealed that the most common circula-
tion types are anticyclonic and cyclonic. The mean-square-error skill scores are used to
investigate classiÞ cationʼs suitability for describing the variability of the local (Pärnu)
daily weather elements. The skill scores of the objective classiÞ cation are essentially
higher than those for the German Weather Serviceʼs “Grosswetterlagen” scheme, but
the scores are still low due to the high variability of daily temperature and precipitation
within the weather types. Temperature is best described by the classiÞ cations at higher
levels of pressure (500 hPa and 700 hPa), but precipitation is best described by those at
the lower levels (sea level and 850 hPa). Developing one good classiÞ cation for both
variables is non-trivial.
Introduction
The large-scale atmospheric circulation is an
important surface climate factor. It determines
the dispositions of baric systems and the domi-
nating airß ows. Therefore, several attempts have
been made to relate the large-scale atmospheric
circulation to local weather conditions. Inside
general circulation models this is achieved by
proceeding from physical laws and deriving
parameterisations for unresolved processes.
Traditionally there are two basic methods for
describing large-scale atmospheric circulation:
the use of circulation indices (NAO index etc) or
weather patterns, the latter is based on similarly
situated baric systems on synoptic maps. Nowa-
days, these simpliÞ ed circulation classiÞ cations
are developed mostly for two purposes: investi-
gation of the temporal variability of atmospheric
circulation with the further aim of forecasting
trends in it, or downscaling of weather elements
from climate models output.
Several atmospheric circulation classiÞ cations
have been developed for the area of Europe.
Two manual ones, applying a daily resolution,
have very long time series: the German Weather
Serviceʼs Grosswetterlagen (Gerstengarbe et al.
1993), and Lambʼs classiÞ cation for the British
282 Post et al. • BOREAL ENV. RES. Vol. 7
Isles (Lamb 1972). The other group of clas-
siÞ cations, called “automatic” or “objective”
classiÞ cations, have increased in popularity with
the development of computer technology (Yarnal
1993).
Grosswetterlagen (GWL) is a planetary scale
classiÞ cation that is especially applicable for
Central Europe (Bissoli and Dittmann 2001).
Post and Tuulik (1999), and Keevallik et al.
(1999) investigated the suitability of the GWL
to describe the weather elements variability in
Estonia. Their results show that the dispersion of
meteorological elements within the weather types
is very large, and several weather patterns could
be interpreted for Estonia independently from
the interpretations for Central Europe (Post and
Tuulik 1999). The horizontal scale of the classi-
Þ cation is much larger than the scale of cyclones
and anticyclones, (which actually determine the
circulation at higher midlatitudes), and since the
process of classifying begins in the middle of the
area, it does not work for peripheries (the Baltic
Sea area already belongs to periphery).
Therefore, our goal was to introduce a syn-
optic scale, automatic classiÞ cation for the Baltic
Sea region atmospheric circulation, and to prove
that the connections with meteorological param-
eters are stronger compared to GWL.
The chosen scheme for atmospheric circula-
tion was developed by Jenkinson and Collison
(1977) (JC) and was initially used for the Brit-
ish Isles region. It was designed as an automatic
version of Lambʼs classiÞ cation. The circulation
pattern for a given day is described using the
locations of the centers of high and low pressure
that determine the direction of the geostrophic
airß ow. It uses coarsely gridded pressure data
and is therefore easily applicable in any area
with available data. Besides the British Isles this
method has already been exploited in several
other European regions: the Netherlands (Buis-
hand and Brandsma 1997), Sweden (Linderson
2001) and Portugal (Trigo 2000).
The classifi cation of weather
types
We applied the JC classiÞ cation for the Baltic
Sea area, centered at 60°N, 22.5°E. The scheme
was used independently for the data at sea level
and at 850 hPa, 700 hPa and 500 hPa isobar
levels. The daily air pressure and geopotentional
height data used in this study originated from
NCEP/NCAR reanalysis (Kalnay et al. 1996).
The data values from 16 points (shown in
Fig. 1) were used to calculate the following
geostrophic airß ow indices: the zonal or west-
erly ß ow, and the meridional or southerly ß ow.
Combining of these two gives the resultant ß ow
(F) and the direction of the ß ow. The westerly
shear vorticity and the southerly shear vorticity
were calculated analogically, the sum of these
gives the total vorticity (Z). The latter describes
Fig. 1. Locations of clas-
siÞ cations. Numbers mark
the 16 points, applying
Jenkinson and Collison
(1977) scheme for the
Baltic Sea area; stars
mark the original location.
BOREAL ENV. RES. Vol. 7 • Weather types’ infl uence on weather in Estonia 283
the rotation of the atmosphere, positive values
correspond to cyclonic circulation and negative
to anticyclonic. The equations for these indices
can be found in Jenkinson and Collison (1977).
We have adjusted the constants that account for
relative differences between the grid point spac-
ing in the east–west and north–south direction to
account for the more northerly location of our
classiÞ cation compared to the original applica-
tion. The geostrophic resultant ß ow F units are
expressed as hPa per 10° latitude at 60°N (each
unit is equivalent to 0.56 m s–1). The (geos-
trophic) vorticity Z units are expressed as hPa
per 10° latitude at 60°N, per 10° latitude. 100
units are equivalent to 0.40 times the Coriolis
parameter at 60°N.
The weather types are deÞ ned by comparing
the numeric values of F and |Z|:
— if , the airß ow is straight and the atmo-
spheric circulation is classiÞ ed into eight
directional weather types according to the
direction of the airß ow (N, NO, O, SO, S,
SW, W, NW).
— If , the airß ow is strongly cyclonic
(Z > 0) or anticyclonic (Z < 0), and the atmo-
spheric circulation is classiÞ ed into synoptic
C (cyclonic) or A (anticyclonic) type, respec-
tively.
— If , the airß ow is partly cyclonic or
anticyclonic, and the atmospheric circulation
is classiÞ ed into 16 hybrid types according
to the direction of atmospheric rotation and
the direction of the airß ow (CN, CNO, CO,
CSO, CS, CSW, CW, CNW, AN, ANO, AO,
ASO, AS, ASW, AW, ANW).
— If the airß ow is weak, an unclassiÞ ed
weather type (U) occurs: and
, where sZ, sWF and sSF
are the standard deviations of total vorticity,
westerly and southerly ß ow, respectively.
In the original scheme one numeric thresh-
old was set to distinguish the unclassiÞ ed type
from the others. We adapted the scheme also
for higher levels, where the strenght of the ß ow
is greater than at the sea level. Therefore, the
threshold values were related to the standard
deviation of airß ow indices.
The Baltic Sea region atmospheric
circulation description using weather
types
The Baltic Sea and its adjacent coasts form
a region where the inß uences of the climatic
zones of Northwestern, Central and Northeast-
ern Europe meet and mingle (Mietus 1998).
The climate of this region is controlled, perhaps
more than that in other parts of Europe, by the
main great pressure systems that govern the air
ß ow over the continent: the Islandic low, and the
Azores high. Additionally, in winter, the branch
of the Asian maximum that extends to the South-
ern Europe and the polar high pressure region
are important. In winter both the Islandic low
and the Azores maximum strengthen, and this
results in westerly and southwesterly ß ow over
the Baltic and a relatively mild climate over the
region. But when the Islandic minimum is weak-
ened and shifted westwards towards the Ameri-
can coast, the inß uence of Arctic is prevailing. In
this case northerly and northeasterly winds blow
and this causes cold and severe conditions. The
mean ß ow is especially intensive in January. In
February and March the intensity of the mean
ß ow over the Baltic region decreases, becoming
at its weakest in April. Also in May the pressure
gradient is very weak.
In summer, the atmospheric pressure distribu-
tion is different. As the northern hemisphere heats
more than the southern one, subequatorial zones
of low pressure shift to north and with this the
subtropic zone of high pressure shifts to north.
High pressure regions form over the oceans, the
ocean low pressure centers weaken and the Azores
maximum strengthens. Over the Asian continent
pressure is much lower in comparison to winter.
In June and July, the mean ß ow could be speciÞ ed
as northwesterly to westerly. In August, the mean
pressure gradient starts to increase again.
All these features of the Baltic Sea region
atmospheric circulation are present in our data
and can be seen in the compounded results pre-
284 Post et al. • BOREAL ENV. RES. Vol. 7
sented in Table 1 and Fig. 2. In Fig. 2 the Þ elds
of normalized sea level air pressure for all SLP
classiÞ ed types are presented. To reduce the
inß uence of very high and low pressures, the
Þ elds were normalized: for the highest pres-
sure of the region it was assigned a value of 1,
and for the lowest region a value of –1, there-
fore the units are relative. The highest and the
lowest pressure areas are marked with a “+” and
“–” respectively and the isolines are drawn after
every 0.15. The pressure patterns are easily inter-
preted. In case of synoptic types, the anticyclone
(A) or cyclone (C) sits in the middle of the area.
In case of directional types, the pair consisting of
cyclone and anticyclone determines the airß ow.
At the time of hybrid types the air pressure dis-
tribution is similar to the respective directional
one, only the centre of the region is more inß u-
enced by the anticyclone or cyclone.
Seasonal pressure distributions are also pre-
sented. During the cold half of the year stronger
pressure gradients were observed, but mostly
the pressure pattern of different weather types
appeared similar in all seasons. There were some
exceptions. The N type in winter: because of the
well pronounced low, the air ß ow to the Baltic
proper was from the Atlantic, in summer vice
versa from the Arctic. The O type in winter: the
strong Siberian high caused the airß ow from
southeast. In summer during this type the main
ß ow is from northeast, from the Arctic.
The dominating weather types for all seasons
were anticyclonic (A) and cyclonic (C). This con-
tradicts the well-known fact that the annual mean
number of cyclones (132) passing over Estonia
(and the whole Baltic region) is much larger than
the number of anticyclones (65) (Prilipko 1982).
The contradiction could be explained by the fact
that the cyclones move faster and have smaller
dimensions. Linderson (2001), who introduced
Table 1. Occurrences of weather types (%) for the 500 hPa isobar level (500GPH) and the sea level (SLP) clas-
siÞ cation for the Baltic Sea region (1968–1997) and for Southern Scandinavia (1881–1995) for comparison from
Linderson (2001) = LL.
Weather type 500GPH Year SLP Dec Jan Feb SLP Jun Jul Aug SLP Year LL Year
A 17.5 18.7 19.0 19.3 17.0
C 15.1 11.7 18.7 14.6 10.7
N 4.4 3.0 6.5 4.5 3.3
NO 1.1 1.7 3.6 2.7 1.9
O 0.6 2.1 2.0 2.3 2.5
SO 0.9 3.3 1.4 2.4 3.8
S 3.6 7.1 3.4 5.5 4.4
SW 9.5 9.6 7.7 8.7 7.7
W 11.3 10.0 5.4 7.9 11.3
NW 9.5 6.9 4.4 5.6 7.7
AN 1.4 0.9 2.2 1.6 1.1
ANO 0.4 0.7 1.6 1.3 0.8
AO 0.3 0.6 1.3 1.1 1.1
ASO 0.3 1.1 1.0 1.3 1.4
AS 1.3 2.4 1.3 1.9 1.4
ASW 2.7 3.4 1.9 2.7 2.2
AW 3.8 3.4 1.8 2.8 3.3
ANW 3.6 2.6 1.6 1.9 2.5
CN 1.3 1.0 2.3 1.5 1.1
CNO 0.4 1.0 1.3 1.0 0.5
CO 0.3 0.7 0.6 0.8 0.8
CSO 0.3 0.7 0.5 0.6 0.8
CS 1.3 0.9 1.3 1.4 1.1
CSW 2.8 1.8 3.0 2.4 2.2
CW 3.3 2.5 2.9 2.5 2.5
CNW 2.9 1.8 1.5 1.4 1.9
U 0.4 0.3 1.8 0.7 4.9
BOREAL ENV. RES. Vol. 7 • Weather types’ infl uence on weather in Estonia 285
the JC classiÞ cation for the area, with the centre
at 55°N, 15°E, also supports the dominance of
anticyclonic weather. Her results are presented
in the last column of Table 1. Summer was the
season of the largest contribution of the cyclonic
vorticity (if to take the C type and cyclonic
hybrids together then 32.2% of days).
The Baltic region lies in a zone of highly
variable westerlies. The main ß ow directions
came out from the occurence distribution of the
weather types in different seasons (Table 1): in
winter the W, SW and S types were the prevail-
ing directional types, and out of four seasons the
directional types had the largest contribution
(48.3%) (because of the large pressure gradient).
The low values of pressure gradients in spring
and summer were observed also from the lower
load of directional types in these seasons: 38.8%
and 34.5% respectively. The classiÞ cation agrees
with the fact that at the 500 hPa level the west-
erly ß ow is stronger than at sea level, and the
easterly weather types vanish.
Fig. 2. Normalized annual
mean sea level pres-
sures (in relative units)
of SLP weather types for
the period 1968–1997. In
parenthesis are shown
the relative occurrences
of weather types. N, NO,
O, SO, S, SW, W, NW =
directional types for differ-
ent directions; AN, ANO,
etc = anticyclonic hybrid
types; CN, CNO, etc. =
cyclonic hybrid types; A=
anticyclonic; C = cyclonic
and U = unclassiÞ ed type.
286 Post et al. • BOREAL ENV. RES. Vol. 7
Classifi cation defi ciencies
The JC scheme is a simple synoptic scale atmos-
pheric circulation classiÞ cation. We have taken
the scheme as it is without making any substan-
tial modiÞ cations, but there are some approxima-
tions and deÞ ciencies what should be considered
when interpreting the results.
The scheme uses geostrophic approxima-
tion that is physically justiÞ ed only for the free
atmosphere. The geostrophic wind speed overes-
timates the real wind speed in cyclonic motion
and underestimates it in anticyclonic motion.
For large-scale motions in midlatitudes the dis-
crepancy is about 10%–20%, and inside intense
cyclones even larger discrepancies can occure.
In the JC classiÞ cation scheme for calculating
the ß ow F, the means of pressures in two or three
points are used. In case of a high curvature of
isobars, this averaging causes a pressure gradient
and consequently an underestimation of the ß ow
strength. This averaging ampliÞ es the difference
Fig. 3. Mean squared
error skill scores (%) for
the predicted daily values
of (a) temperature, (b)
precipitation sums and (c)
precipitation occurrence
at Pärnu in the period
1968–1997. The abbrevia-
tions are the same as in
Table 2.
0
10
20
30
40
50
60
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Skill score (%)
a
15
20
25
30
35
40
Months
Skill score (%)
SLP 850GPH 700GPH 500GPH
c
5
10
15
20
25
30
35
Skill score (%)
b
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
BOREAL ENV. RES. Vol. 7 • Weather types’ infl uence on weather in Estonia 287
Table 2. Annually averaged mean squared error skill scores (%) for the predicted daily values of weather elements
at Pärnu (1968–1997) and in parenthesis at De Bilt (1949–1993) (Buishand and Brandsma 1997). SLP-classiÞ ca-
tion for the sea level, 850GPH for the 850 hPa level, 700GPH for the 700 hPa level, 500GPH for the 500 hPa level
and GWL = German Weather Serviceʼs Grosswetterlagen.
SLP 850GPH 700GPH 500GPH GWL
T 21.0 (28.7) 25.7 29.2 29.5 22.8 (39.6)
Psums 23.9 (19.3) 23.7 20.0 16.5 10.8 (16.1)
Pocc 28.8 (24.8) 30.9 30.1 26.8 14.5 (25.8)
of the geostrophic wind from the gradient wind
in case of anticyclones, and decreases the differ-
ence in case of cyclonic motion.
The weather types were deÞ ned using numer-
ical thresholds for circulation indicies. The only
ground for concrete numerical values of thresh-
olds is that the resulting mean pressure Þ elds
show different circulation patterns. But, the pro-
cedure of comparing Z to F (or dividing Z to F)
kicks out many combined cases of high cyclonic
vorticity, high wind speeds and powerful weather
events from cases of cyclonic weather type. And
from the mean pressure Þ elds (Fig. 2) it is clearly
seen that the classes NO, O, S and SW show a
clear cyclonic cuvature. This also explains in
part the anticyclonic prevailing: some cases of
cyclonity are included into directional types.
The infl uence of weather types on
the daily temperature and
precipitation
One of our purposes was to investigate the inß u-
ence of atmospheric circulation on the mete-
orological regime of the Baltic region. Under
research were data from several meteorological
stations in Estonia, but the results coincided to
such extent that only results from Pärnu mete-
orological station (58.37°N, 24.50°E) are pre-
sented here. The location of Pärnu on the eastern
coast of the Baltic Sea is shown in Fig. 1. Daily
deviations from the long-term monthly mean
temperature, daily precipitation sums and the
proportion of rainy days (precipitation occur-
rence) for the distinct weather types over the
period of 1968–1997 were calculated.
To evaluate the capability of the classiÞ cation
to describe the variability of the local (Pärnu)
temperature and precipitation, the dispersion
analysis was used. We calculated the mean-
squared-error skill scores as done by Buishand
and Brandsma (1997) for the predicted daily
values that show to how large extent the variabil-
ity of the temperature and precipitation is deter-
mined by the changing of weather types. The
skill score gives the proportion of the explained
variance and is the square of the multiple cor-
relation coefÞ cient. The magnitude of the skill
score is determined by the squared deviations of
the individual values from monthly averages.
The skill scores reveal which pressure levels
were the most suitable for explaining the tem-
poral variability of surface weather elements
(Fig. 3). The skill scores had a strong annual
cycle for temperature. In summer the skill scores
were the largest for the 500 hPa classiÞ cation
and in winter for the sea level pressure classiÞ ca-
tion. This shows that in summer the daily tem-
perature of the local area was more connected
with movements in the free atmosphere, but in
winter the processes in the boundary layer were
more important. From here it follows that the
classiÞ cation describes temperature variations
better if several levels in the atmosphere are
taken into account. Precipitation processes were
more related to surface conditions: the largest
skill scores were for the sea level and 850 hPa
classiÞ cations.
For the objective classiÞ cation at sea level,
the precipitation skill scores were higher in
Pärnu than in De Bilt (in the Netherlands), but
skill scores for temperatures showed the oppo-
site (Table 2). We also included the results of
Grosswetterlagen to obtain comparison with
the previous works (Keevallik et al. 1999, Post
and Tuulik 1999). For GWL the skill scores in
Pärnu were lower than in De Bilt for all weather
288 Post et al. • BOREAL ENV. RES. Vol. 7
elements. That proves that the Grosswetterlagen
classiÞ cation is less suitable for Estonia than for
the Netherlands. Concerning temperature, the
Baltic Sea area is a problem also for the objec-
tive scheme (skill scores are lower than for the
Netherlands). It could be related to the fact that
for this spatial scale of classiÞ cation, the Atlantic
Ocean as a homogeneous boundary surface is
too far from the Baltic area, but close enough to
the Netherlands. As our results for the precipita-
tion were better than those for Netherlands, the
determining scale maybe more local in this case
and the vicinity of Atlantic may not play a role.
Nevertheless, the skill scores for the objective
weather types at Pärnu were higher than for
Grosswetterlagen. It should therefore be possible
to develop a classiÞ cation that better describes
the variability of weather elements. The investi-
gation shows that classiÞ cations cannot describe
the variability of temperature and precipitation
equally well. The solution would be to Þ nd rela-
tions between the circulation parameters (direc-
tion and strenght of the dominant airß ow, and
vorticity) and any weather element separately, as
it has been done in many downscaling studies,
and based on the results, to establish respective
classiÞ cations. This would eliminate also the
subjectivity of thresholds.
Discussion and conclusion
From the present work, it follows that for
investigating the temporal variability of the
atmospheric circulation in the Baltic Sea region
the Jenkinson and Collison (1977) scheme is
certainly good enough. It distinguishes certain
different weather patterns; it is physically easily
interpretable and uses only sea level pressures
(that have the longest available time series). Its
simple applicability to the Baltic Sea area pres-
sure Þ elds on different levels has also given
better relations with weather elements than
GWL. Temperature is best described by the clas-
siÞ cations at higher levels of pressure (500 hPa
and 700 hPa) and precipitation at the lower ones
(sea level and 850 hPa). But the skill scores stay
still low, which means that the scheme should be
improved for downscaling purposes.
The results suggest that if we want the clas-
siÞ cation to describe well the variability of
regional weather elements, then several circula-
tion indices on different height levels should be
taken into account. The use of simple physical
models, like geostrophic wind approximation,
gives a direct physical interpretation of the clas-
siÞ cation and a reasonable number of classes.
The approximation of geostrophic wind is physi-
cally justiÞ ed only for the free atmosphere. If we
have only sea level pressure Þ elds, some statisti-
cal classiÞ cation could be used. If the task is to
conserve also the physical meaning, the sea level
winds in combination with empirical relations
from the boundary layer dynamics could be used
to get upper air motions.
Acknowledgements: This work was Þ nancially supported
by the Estonian Science Foundation (grant No 4347). Two
anonymous reviewers gave many useful comments to the
Þ rst version of the paper. All this is gratefully acknowledged.
References
Bissoli P. & Dittmann E. 2001. The objective weather type
classiÞ cation of the German Weather Service and its
possibilities of application to environmental and mete-
orological investigations. Meteorol. Z. 10: 253–260.
Buishand T.A. & Brandsma T. 1997. Comparison of circula-
tion classiÞ cation schemes for predicting temperature
and precipitation is the Netherlands. Int. J. Climatol.
17: 875–889.
Gerstengarbe F.W., Werner P.C., Busold W., Rüge U. &
Wegener K.-O. 1993. Katalog der Grosswetterlagen
Europas nach Paul Hess und Helmuth Brezowsky. Ber.
Dtsch. Wetterdienstes, 113, 4th ed., Selbstverlag des
Deutschen Wetterdienstes, Offenbach am Main, Ger-
many, 249 pp.
Jenkinson A.F. & Collison F.P. 1977. An initial climatol-
ogy of gales over the North Sea Synoptic Climatology
Branch Memorandum No. 62, Meteorological OfÞ ce,
Bracknell, 18 pp.
Kalnay E., Kanamitsu M., Kistler R., Collins W., Deaven
D., Gandin L., Iredell M., Saha S., White G., Woollen
J., Zhu Y., Leetmaa A., Reynolds R., Chelliah M., Ebisu-
zaki W., Higgins W., Janowiak J., Mo K.C., Ropelewski
C., Wang J., Jenne R. & Joseph D. 1996. The NCEP/
NCAR 40-year reanalysis project. Bull. Am. Meteorol.
Soc. 77: 437–472.
Keevallik S., Post P. & Tuulik J. 1999. European circulation
patterns and meteorological situation in Estonia. Theor.
Appl. Climatol. 63: 107–116.
Lamb H.H. 1972. British Isles weather types and a register of
daily sequence of circulation patterns, 1861–1971. Geo-
phys. Mem. 110, Meteorol. OfÞ ce, London, 85 pp.
Linderson M.L. 2001. Objective classiÞ cation of atmospheric
BOREAL ENV. RES. Vol. 7 • Weather types’ infl uence on weather in Estonia 289
circulation over Southern Scandinavia. Int. J. Climatol.
21: 155–169.
Mietus M. 1998. The climate of the Baltic Sea Basin. Marine
Meteorology and Related Oceanographic Activities,
Report No. 41, WMO/TD-No. 933, 180 pp.
Post P. & Tuulik J. 1999. About the relationships between
Estonian weather elements and European circulation pat-
terns. Phys. Chem. Earth (B) 24: 97–102.
Received 23 January 2002, accepted 25 June 2002
Prilipko G.I. [Prilipko G.I.] (ed.) 1982. [The climate of
Tallinn], Gidrometeoizdat, Leningrad, 267 pp. [In Rus-
sian].
Trigo R.M. & DaCamara C.C. 2000. Circulation weather
types and their inß uence on the precipitation regime in
Portugal. Int. J. Climatol. 20: 1559–1581.
Yarnal B. 1993. Synoptic Climatology in Environmental
Analysis, Belhaven Press, London, and Florida, 195 pp.
