Especially in summer time, Romania is affected by deep convections associated with severe weather phenomena (hail, wind gusts, flash floods, frequent thunder).
The interannual variability of the Cb (cumulonimbus) cloud frequency recorded at 20 weather stations from the southern part of Romania in the 1961–2007 interval was investigated in connection with the atmospheric circulation variability.
Using the Canonical Correlation Analysis (CCA) three types of atmospheric circulations have been identified that control about 60% of the interannual variability of the Cb cloud frequency.
It is evident that the spatial distribution of climate elements such as air temperature, precipitation, sunshine duration, etc. is dependent on atmospheric circulation patterns on all scales. The advection of warm/cold and dry/moist air masses determines the large-scale air mass properties and thus the variability of climate elements. These air mass impacts are modified by local topographic features, e.g. luff/lee effects near mountains. Cyclonic or anticyclonic conditions imply dynamical circulation processes, in particular, lifting or subsidence of air masses which is essential, e.g. for precipitation. Special properties of air masses, such as humidity also influence the daily weather, e.g. by formation of clouds or fog.
The aim of this paper is to determine the atmospheric circulations controlling the interannual variability of the cumulonimbus (Cb) clouds for southern Romania (Wallachia area) using analysis of data collected from the weather stations regarding the occurrence of hail-bearing cumulonimbus clouds. The hail phenomenon is particularly local and short-lived and is associated with storms and thunder phenomena, being thereafter associated with the occurrence of the storm Cb-type clouds. The frequency of Cb clouds and of the associated severe weather phenomena is influenced by large-scale atmospheric circulations (Draghici and Militaru, 1985).
The number of days when one or more hail events were recorded is the main indicator of the presence of Cb clouds. That is why a climatology of the hail events is a reliable indicator of the climatology of Cb occurrence frequency in a certain region. Such climatology is useful in the design and operation of the economic units in a manner differentiated by regions, in timely designing the various works over the territory, as well as the means of intervention, in situations of damage or accident.
Cb clouds associated with the hail phenomenon are formed as a result of the dynamic or thermal atmospheric instability (Zinner et al., 2008). A number of theories have been elaborated on the formation of hail, based on observations performed in thunder clouds (Curic et al., 2003). It results from the theories that hail is formed in a convective cloud sufficiently developed above the freezing level to produce ice particles and containing ascending currents strong enough to allow the ice particles grow to the size of hail. For this growth to happen, clouds must expand to heights of several kilometres and either the ascending currents be strong or the amount of water in a liquid form be very large (Potapov et al., 2007). The different shapes of the ice stones depend of the in-cloud temperature where they form. Hail forms in almost any convective cloud undergoing a certain development phase, usually in its mean and high levels (Tuomi and Mäkelä, 2008).
At the former Institute of Meteorology and Hydrology, preoccupation existed for the research of the convection mechanism and of hail occurrence conditions, along with the peculiarities of the hail occurrences distribution over the Romanian territory (Iliescu, 1983).
The recent development of the Doppler radar systems allows the use of certain experimental and operational techniques to diagnose the presence and size of the hail (Doppler radar Meteorological Observations, 2006). Such techniques are a combination between radar reflectivity (Laksmanan et al., 2007) and its derivatives, and information connected to the environment (Absaev et al., 1973).
Radar echoes allow the detection of the thunder phenomena that occur as a result of the development of convective clouds (Francisco, 2001). Most methods for the identification of convective echoes as thunder clouds use as a criterion their height and intensity (Ananova et al., 2007). The process of forecasting the occurrence of convective clouds is not a simple one and the identification of echoes associated to potentially hazardous clouds is not certain (Krauss et al., 2007). Comparing observations from weather stations to synchronic radar ones, it can be observed that in general convective echoes are associated with showers, thunder phenomena and hail (Krauss, Sinkevich, 2007). It is noticeable that clouds with small heights developing within a cold air mass can reach negative temperatures like those necessary for the formation of large-size hail, but that the sub-cooled water amount is usually smaller than during the continental spring–summer storms (Volkova and Uspenskii, 2007).
Development conditions for the cumulonimbus clouds in southern Romania may be influenced by synoptic-scale circulations. For instance, a low-pressure field, centred in the Mediterranean Sea advects warm and moist air and a high-pressure field in eastern Europe advects cold air over southern Wallachia. The confluence of the two air masses will create conditions for Cb cloud occurrence in southern Romania (Ion-Bordei, 2008). Most of the severe weather situations occurring on the Romanian territory are owed to synoptic-scale interactions between an east European pressure system and a complementary cyclonic or anticyclonic field, covering variable portions from the remainder of the continent (Bordei Ion Ecaterina, 2003, 2004). A very small number of cases of severe weather in Romania are associated with a single cyclonic pressure system. Severe weather phenomena have also been associated with the presence of low-pressure systems, either in the central Mediterranean, or over the Balkan Peninsula, over the west of the Black Sea, simultaneously with the presence of an anticyclonic field, either over Central or over Eastern Europe (Bordei Ion Ecaterina, 2003, 2004).
The variation of extreme phenomena frequency has been approached in connection with the long-term frequency of the macrocirculation types (MCPs).
The atmospheric circulation is the principal factor that determines the climate variability in Europe.
A multitude of synoptic classifications exist that can be distinguished by way of the variables used and the classification procedure applied. Circulation pattern classifications typically employ large-scale pressure data while others use local weather elements as input variables (Lambert, 1988).
Besides, a classification can either be subjective, objective, or hybrid. The choice of the appropriate technique depends on the needs of the research, the skills of the investigator, and the nature of the data (Frakes and Yarnal, 1997). Initial classifications were commonly subjective and their application was mainly limited to medium- (at that time, 3–5 days) and long-range forecasting (Baur, 1948).
Some of manual classification of synoptic circulation patterns are based on circulation types over large territories—Europe, the British Isles and Artic seas (Vangengeim 1952; Hess and Brezowsky 1969; Lamb 1972; Dmitriev 2000)—and even over the whole hemisphere (Dzerdzeevskij 1968; Girs, 1971). Atmospheric circulation patterns are rather closely correlated with air temperature fluctuations in Europe. The frequency of circulation forms by Vangengeim-Girs has a higher correlation with the air temperature in northern and eastern Europe, while the circulations groups by Hess-Brezowsky describe better the climate variability in central Europe.
Numerical methods can now perform the classification for the Lamb indices as well as the Grosswetterlagen based on pressure fields. The focus of the classification can then be moved to other areas, e.g. Romania, and can result in a more relevant classification for Romanian conditions.
Changes in the atmospheric circulation over North Atlantic and Europe have been studied recently by several groups of authors employing various methods and various time periods (e.g. Klein W, 1957; Bárdossy and Caspary, 1990; Stefanicki et al., 1998; Mächel et al., 1998; Fu et al., 1999; Slonosky et al., 2000; Maheras et al., 2000; Werner et al., 2000; Plaut and Simonnet, 2001; Jacobeit et al., 2001). Most of them have been confined to analysing changes in frequencies of circulation types and their groups. However, a few recent studies examining the Hess-Brezowsky classification of Grosswetterlagen (Hess and Brezowsky, 1969) have indicated that considerable changes in the persistence (measured by the mean residence time of circulation types) of atmospheric circulation over Europe have occurred in the 1980s and 1990s. The increase in the residence times of Hess-Brezowsky circulation types was first reported by Werner et al. (2000) for zonal circulation in winter, and Kyselý (2000, 2002) for all circulation types in summer, and it may have had pronounced impacts on changes in the occurrence of temperature (and other) climatic extremes, as demonstrated in Kyselý (2002) for the heat wave frequency and intensity in Prague (the Czech Republic).
As regards our country, such classification of the atmospheric circulation was displayed by Topor and Stoica (1965). It was shown that the increase of the blocking circulation frequency is associated with the increase in the extreme precipitation phenomena in Romania (Topor and Stoica, 1965). More recently, a systematic classification of atmospheric circulations for various European regions was performed within the COST Program (Cost Action Proceedings, September 2005). The shift in the frequency of certain circulations is associated with shifts in the frequency of extreme phenomena. Cold frontal precipitation, for instance, has increased in the last 45 years, especially during summer and autumn, while warm front precipitation decreased, especially during winter (Demuzere, M. et al., 2008).
This paper investigates circulations associated to the interannual variation of the Cb clouds through a statistical Cb analysis, from records at 20 weather stations in southern Romania over the 1961–2007 period, combined with observed climatological data fields. The paper is structured thus: Section 2 displays the Data and methods. Circulations associated with the main variation modes of the Cb clouds, identified with the help of the canonical correlation analysis are described in Section 3. Discussions and conclusions follow in section 4.
2. Data and methods
For this study, there were used observation data referring to the annual frequency of the Cb clouds at 20 weather stations from southern Romania (Wallachia area), stations that are covered by the C-band meteorological radar located in Bucharest (Figure 1). Observation data were organized by time series comprising the days with cumulonimbus cloud occurrence and the days with hail for the 1961–2007 period (47 years). We mention that the data series for the above mentioned period was made up by observations performed at the four synoptic terms (00, 06, 12, and 18 UTC). Processing the observation data implied the application of statistical methods for the determination of the spatial and temporal variability of the cumulonimbus hail-bearing clouds (Hodges, 1994). Data referring to the cumulonimbus clouds were obtained from the weather stations within the network of the Romanian National Meteorological Administration (NMA). Those data are regularly archived on an electronic support.
We used data from the reanalysis project from the National Centers for environmental prediction and from the National Center for Atmospheric Research (NCEP/NCAR). Reanalysis Project 1 (NCEP/NCAR) uses an updated review of the literature about the reanalysis/forecasting system for the improvement of data assimilation, using data older than the year 1948 up to the present.
Quasi-observed large-scale data from NCEP-NCAR reanalysis for 1961–2007 over (20°W–80°E; 20°N–75°N) were used to identify the atmospheric circulation patterns related with Cb variability. These reanalysis data were derived through a consistent assimilation and forecast model procedure that incorporated most available weather and satellite information (Kalnay et al.1996). The horizontal resolution of the sea level pressure (SLP) and air temperature (T) fields used in our study is 2.5 grad lat. × 2.5 grad lon. Summer (June/July/August–JJA) monthly anomalies are averaged to obtain JJA SLP and T anomalies. Prior to statistical analysis the data are detrended and normalized by their standard deviation.
The Canonical Correlation Analysis (CCA) is a way to measure the linear relation between two multidimensional variables. Two bases are found, one for each variable, optimum as regards the correlations and at the same time the corresponding correlations are found. In other words, the two bases are found where the correlation matrix between the variables is the diagonal, and correlations on the diagonal are maximized. The size of this new base is equal to or smaller than the smallest size of the two variables. An important property of the canonical correlations us that they are invariant as regards the affine transformation of the variables. This is the most important difference between CCA and the usual correlation analysis, which depends largely on the bases in which the variables are described (von Storch and Zwiers, 1999).
3. Atmospheric circulations associated to the dominant modes of Cb cloud frequency variability
CCA was applied to a set of annual Cb frequencies recorded at 20 weather stations from southern Romania and the sea level pressure field (SLP) within the area delimited by (20°W–80°E, 20°N–75°N) from JJA. Although the Cb frequency is annual, the greatest contribution to the convective developments is that of the summer season, which is why we took into account the summertime SLP within the above mentioned area. Before applying CCA, data were linearly detrended and normalized by the corresponding standard deviation. However, original time series shows strong positive linear trends from over the period 1961–2007 (Table I). It is in accordance with observations from Europe and other parts of the world.
Table I. Linear trend of frequency of Cb over a period (1961–2007) recorded at 20 meteorological stations from southern Romania. Units days/10 year
We identified three types of atmospheric circulations associated with the interannual variability of Cb frequency in southern Romania. Though associated with large-scale atmospheric circulations, the variation modes of the Cb cloud frequency within the studied area have a heterogeneous spatial structure. This is owed to the strong regional effects that influence Cb cloud formation. Despite that, the positive values prevail in these patterns, so that the large (small) values of their amplitude correspond to large (small) Cb frequencies in the considered area.
CCA1 accounts for 25.8% of the Cb variability (Figure 2a)). The corresponding mode in SLP (Figure 2b)) accounts for 14.7% of the variance. The canonical time series associated with these modes (Figure 2c)) are strongly correlated (r = 0.81). The relatively large values in the years 1977 and 1983 (Figure 2c)) show that negative pressure anomalies in eastern Europe area trigger increased values of the Cb cloud frequency in southern and eastern Romania. The relatively small values of the canonical coefficients in the years 1963 and 2007 (Fig. 2c) show that high-pressure values in the mentioned areas are associated with small Cb frequencies in southern and eastern Romania.
The physical mechanism behind these statistical relations may be the following. When the negative phase of the pattern represented in Figure 2b) dominates the atmospheric circulation in the (20°W–80°E; 20°N–75°N) area, an additional advection of cold air from the northern Europe area occurs towards southern Romania. This relation is acknowledged by the positive correlations between the time series associated to the time pattern rendered in Figure 2c) and the surface temperature in most part of that region (Figure 3). However, dynamic processes could be also important in formation of Cb. Also, near the region of studies the correlation is close to zero or positive, which is equivalent to the existence of small positive temperature anomalies. These positive temperature anomalies could favour Cb cloud formation.
Synoptic processes in this type of circulation are characterized by the presence of an altitude thalweg oriented from northeast to southwest and in which the Black Sea region is often shaping a cyclone with closed izohipses. North African anticyclone influence manifests itself as a ridge which can be extended up to the Scandinavian Peninsula. In this type of circulation, polar air penetration affects mainly southeastern Europe. Pressure field's anomaly represented in Figure 2b) suggests a direct northern circulation, quite intense over the geographical space of Romania. Therefore, more important vertical developments could take place only in the western part of the country, but there is increasing pressure here, which induces a component of subsidence, which would inhibit upward vertical movements. On the other hand, in the northwestern basin of the Black Sea, pressure shows a significant drop in the isalobaric nucleus, implying increased cyclonic whirlpools and inherently major cloud systems development likely to cause severe weather events in the eastern extreme of Romania (heavy rainfall).
The second circulation type favouring an increased frequency of the Cb clouds in southern Romania is identified by CCA2 (Figure 4a)). This one accounts for 19.5% of the variance (Figure 4a)). The corresponding mode in SLP (Figure 4b)) accounts for 9.3% of the variance. The canonical time series associated with these modes (Figure 4c)) are correlated (r = 0.66). A positive (negative) pressure anomaly centred over northern half of Europe, accompanied by negative (positive) pressure anomalies over the Black Sea Basin and eastern extreme of Romania favours a high (small) Cb frequency in southern Romania. The temperatures anomalies (Fig. 5) together with local dynamic processes associated with this circulation could explain the increase in the frequency of Cb in southern part of Romania as captured by CCA2 (Figure 4a)).
This circulation is remarkably strong, displaying a positive phase in 1975 and 2006 and a negative one in 1961 and 1987 (Figure 4c)). The synoptic situation was analysed for the above listed years.
This situation reveals the presence of a strong anticyclonic field over the northern half of the continent, while the rest remains under low pressure, with a significant minimum above the Black Sea Basin. This is a typical situation of tightly connected cyclone–anticyclone, often generating serious phenomena at least in eastern and southeastern regions of our country (winter snowstorm). For the Romanian territory, we speak in this situation of intense circulation east-northeast, even if isolines show a slight tendency of diffluence towards the west and the region of very tightly connected cyclone–anticyclone is situated rather over southern Ukraine.
The third circulation type that influences the Cb cloud frequency in southern Romania is identified by CCA3 and accounts for 12.5% of the Cb variability (Figure 6a)). The corresponding mode in SLP (Figure 6b)) accounts for 5.5% of the variance. The canonical time series associated to these modes (Figure 6c)) are correlated (r = 0.63). Positive (negative) pressure anomalies in the area of the Caspian and Black Sea Basin accompanied by negative (positive) pressure anomalies in the northern part of Europe (Figure 6b)) favours cold and dry (warm and moist) air advection in southern Romania, inducing positive (negative) temperature anomalies in this area (Figure 7). This temperature anomalies configuration is related to low frequency of Cb, as negative values are dominant in the CCA3 Cb pattern (Figure 6a)).
This circulation was particularly strong and displayed a positive phase in 1978, 1986, and 2007 and a negative one in 1965, 1983, and 1991 (Figure 6c)). The synoptic situation was analysed for the above listed years.
When negative phase of this pattern occurs (high frequency of Cb) in the north of Europe dominates a field of low atmospheric pressure, with a dominant cyclonic regime, while in the neighbouring territories in the southwest, and especially southeast of Romania, we have a predominant anticyclone regime, and atmospheric pressure is higher. Therefore, the synoptic setup is quite favourable for the development of severe convective phenomena in the hot season, given that over Romania we deal with an uncertain pressure field, without a well developed circulation, a so-called barometric swamp. Within this structure, at ground level, (depending on the structure of vertical upper air) very large-scale convective motions are often favoured, especially that the existence of cyclonic area at the north of the country's periphery involves afferent frontal passages (related to it) which is an additional forcing for the thermal-forced upward during the warm season.
To test the relationship between Cb frequency and hailstorms frequency in southern Romania we have performed EOF analysis of series of clouds frequency and hail frequency recorded at the same stations in the same period. EOF1 of Cb which describes 28% of the Cb variance has a mono-polar structure (not shown) and EOF1 of hail which describes 15% of the hail variance (not shown) has also a mono-polar structure. Associated time series (PC1) represented in Figure 8 are significantly positive correlated (r = 0.50) which indicates that an increased Cb frequency is accompanied by an increased hail frequency in the same region. This relationship is physically consistent and proves that Cb and hail datasets are consistent. The time variation of these modes (Figure 8) is similar with the variation of the amplitude of CCA1 (Figure 2c)).
4. Discussion and conclusions
This article investigated the atmospheric circulations that control the interannual variability of the Cb cloud frequency in the southern part of Romania.
Three atmospheric circulation types were identified as responsible for about 60% of the interannual variability of Cb cloud frequency. The three identified circulation types favour warm and moist air advection from the Mediterranean Sea and cold air advection from northern Europe over southern Romania, which creates in that area conditions for Cb cloud development.
The identified circulation types account for the conditions favouring Cb cloud occurrence, taking into consideration the anomalies found at ground level and the associated dynamical processes. Development conditions for the Cb clouds are influenced by the synoptic-scale circulations.
To get a clearer picture of the atmospheric circulations associated with the interannual variability of the Cb cloud frequency in southern Romania, we analysed several synoptic situations occurred in the years when the CCA-identified patterns where in the positive phase and displayed a wide amplitude.
CCA1 was particularly strong in 1977 (Figure 2c)). We chose as an example the atmospheric circulation of 3 June 1977 (not shown) when a strong Cb developed in southern part of Romania. In this particular day, a western block stops the cloud system originated from the ocean advancing towards the continent. The associated Azores Ridge centred on the British archipelago deviates all cloud systems towards central and eastern Europe. In altitude, there is a nucleus of cold air centred in northern Romania. In particular, this circulation in Romania allows cold air to enter through the anterior part of the anticyclone (in the lower layer of the atmosphere) but also brings the cold air from altitude due to the cold nucleus. Rainfall is due to both dynamic and thermal causes.
Another example is the synoptic situation of 30 May 2006 (not shown) which was related to enhaced Cb activity in southern parts of Romania. That circulation has a spatial structure similar to CCA2 which was in its positive phase and displayed large amplitude in 2006 (Figure 4c)). A synoptic situation typical to this year was on 30 May 2006, when was a meridional circulation over the European continent conditioned by a western block. The Azores Ridge with the centre of maximum pressure in the western British Isles favours normal latitude descent of cold air from north to south, air entrained on the back of a cyclonic nucleus situated in the north of the continent. With regards to Romania, there is a depression field, cold air of polar origin trained from the cyclonic nucleus from the north of the continent seeps in the northern Alps to the Adriatic Sea where it reactivates the cyclogenesis process. Mediterranean cyclone wave approaches our country from the southwestern side.
The synoptic circulation of 5 September 2007 has a spatial structure to CCA3 (not shown), which was notably strong throughout that year (Figure 6c)). At the synoptic scale, there is a blocking circulation, both western and eastern. Meridian circulation is conditioned by the presence of this Azores anticyclone located in the British Isles, very developed in amplitude and spatiality and by an eastern European anticyclone hot and very stable. The presence of these two centres of high baric pressure will force energy and moisture in the depression formed by its passage.
Instability in our area of interest is maintained by the presence of a mobile cyclonic nucleus reduced in size but very active.
The similarity between the spatial structure of CCA modes and the spatial structure of the above described circulation suggests that a large part of interannual variability of the frequency of Cb in southern part of Romania is controlled by interannual variation in the frequency of synoptic-scale circulations. However, a more detailed study using both observational data and atmospheric model experiments are needed to improve our understanding in the relationship between these synoptic-scale circulations and frequency of Cb in southern parts of Romania.
General circulation models probably simulate correctly the occurrence frequencies of those circulations but not necessarily the regional processes also associated to Cb formation in southern Romania. Therefore, investigating the frequency of these circulation types in the experiments with climatic models (scenarios) can give information connected to changes in the frequency of Cb clouds and the associated phenomena (hail, thunder, convective storms) in the coming decades.