The applicability of the Hess–Brezowsky synoptic classification to the description of climate elements in Europe

This study deals with the applicability of the Hess–Brezowsky synoptic classification to the description of surface climate elements, specifically minimum and maximum temperature and precipitation, in Europe. The suitability of the Hess–Brezowsky classification for this purpose in the European domain is analyzed using the two-sample Kolmogorov–Smirnov test. The test was performed on European Climate Assessment and Dataset data from 113 stations for the years 1961–2000. The suitability of the classification for describing climate elements at a given station was assessed according to the percentage of the circulation types, under which the probability distribution functions of these elements differed from the rest of the values. The classification is deemed the most suitable for describing climate elements in Germany and its neighboring countries and least suitable in the Mediterranean, southeastern and Eastern Europe. The classification is more applicable to describing minimum and maximum temperature than precipitation, and its overall synoptic-climatological applicability is better in winter than in summer.

One of the most important factors influencing surface climate is the atmospheric circulation. By being observed for longer periods of time, the atmospheric circulation can be divided in several of its typical states—circulation patterns (or types). Tools for recognizing and catalogizing such patterns are referred to as circulation type classifications.

Since the atmospheric circulation can be described in a lot of different ways, there are several different approaches to these classifications. For Europe only, dozens of atmospheric circulation classifications have been developed, varying in spatial and temporal resolution, number of circulation types, or the classified variable over the area of interest, such as the air trajectories (e.g., Seibert et al. 2007), cyclone paths (e.g., Sickmöller et al. 2000; Gaffney et al. 2007), and the wind field (e.g., Beaver and Palazoglu 2006; Burlando et al. 2008), but most often the sea level pressure and height fields. They also differ in the process of defining circulation types and assigning individual synoptic situations to them, which can be carried out manually (subjective classifications) or by applying numerical methods (objective classifications). For a more detailed reading about circulation classifications and their applications, see Huth et al. (2008).

Classifications developed for various European regions were collected in a database created within the COST Action 733 (Philipp et al. 2010; Philipp et al. 2016) in order to enable their systematic evaluation and comparison based on various criteria, such as their ability to describe climate variables (e. g. Beck and Philipp 2010; Casado et al. 2010; Huth 2010; Tveito 2010; Ustrnul et al. 2010) . More recently, the database has been used to analyze effect of the properties of classifications on their ability to describe surface climatic conditions (Beck et al. 2016; Huth et al. 2016) and to investigate changes of frequency and persistence of circulation types in Europe (Cahynová and Huth 2016; Kučerová et al. 2017).

This study deals with one of the oldest and most widely used circulation classifications in Europe, the subjective Hess–Brezowsky classification, also known as Grosswetterlagen, GWL (Hess and Brezowsky 1952; Werner and Gerstengarbe 2010). It has been praised for its ability to capture the large-scale characteristics of weather regimes, while still focusing on local detail (James 2007) and for its intuitive naming convention (James 2007; Wapler and James 2015; Minářová et al. 2017a).

It is based on the catalog developed by Baur et al. (1944), which contained 21 synoptic types defined first by the geographical distribution of the pressure field and the position of the frontal zone above the European continent and the adjacent ocean, secondly by the weather patterns over Central Europe which may be cyclonic or anticyclonic according to the prevailing influence of surrounding pressure systems. It was reworked and later further revised by Hess and Brezowsky (1952, 1969, 1977); the number of types has been increased to 29, and their definition has begun to also take into account geopotential levels of 500 hPa and sea level pressure. In 1952, a calendar was published, assigning a circulation type for each day from 1 January 1881 to 31 December 1950. Since then, the catalog has been updated and published several times by Gerstengarbe and Werner (most recently in Werner and Gerstengarbe 2010). Since 1999 to present, the catalog has been administered by the German Weather Service.

The Hess–Brezowsky (henceforth HB) classification consists of three groups of circulation forms—zonal, meridional, and mixed U (Werner and Gerstengarbe 2010). Those are further divided into ten major types (known as Grosswettertypen, GWT) and lastly into 29 circulation types (Grosswetterlagen, GWL) and one transitional type—type U, the “undefined” type. The circulation types are defined by the location of pressure systems and frontal zones over Europe, the prevailing direction of air masses and character (anticyclonic or cyclonic) of the weather in Central Europe. In the catalog, each circulation type must last at least three consecutive days, with the exception of type U. If the transition from one circulation type to another is not clear, 1 or 2 days can be classified under the transitional type. Days during longer transitions are assigned to the preceding or following type.

Although originally developed for the purpose of weather forecasting in Germany, nowadays, the HB classification is used for other purposes—mainly synoptic-climatological studies—and also in other parts of Europe. For example, it was used to investigate the link between atmospheric circulation and floods on the Vltava river, Czech Republic (Kyselý et al. 2003), and air temperature in Poland (Ustrnul 2006). Together with its automated version made by Paul James, the SynopVis Grosswetterlagen (SVG, described in Hoy et al. 2013a), the classification was used to analyze thunderstorm occurrence (Wapler and James 2015) and the variability of the frost-free season (Wypych et al. 2017) in Central Europe and the characteristics of extreme precipitation in the Ore and Vosges Mountains regions (Minářová et al. 2017a, 2017b).

The frequency of HB major types and their relationship to the variability of air temperature in Karkonosze Mountains (Migała et al. 2016) was investigated, as well as the persistence of HB circulation types (Kyselý and Domonkos 2006) and their connection to extreme temperatures in Europe (Kyselý 2008; Kyselý and Huth 2008) and the Czech Republic (Kyselý 2007), although it was later confirmed that the HB catalog contains inhomogeneities affecting persistence (Cahynová and Huth 2009; Kučerová et al. 2017). As of late, the automated SVG catalog is frequently used for the analysis of trends instead (Fleig et al. 2015; Hoy et al. 2013b; Nilsen et al. 2017).

Outside of Central Europe, the HB classification was utilized to analyze circulation types leading to extremely hot days in Europe (Cony et al. 2010), the longest heat waves in Serbia (Unkašević and Tošić 2009), and the cold spells in Northeastern Romania and Moldova during the winter of 2012 (Planchon et al. 2015); to investigate the influence of atmospheric circulation on aerosol cloud–mediated processes in the Baltic region (Krüger 2014) and the variability of snow cover in Lithuania (Rimkus et al. 2014); and to develop a method for short-term temperature and precipitation prediction in Spain (Ortiz-García et al. 2012; Ortiz-García et al. 2014). The objective computational version of the classification, Objective Grosswetterlagen (OGWL, James 2007), was used for the identification of weather patterns causing heavy winter rainfall in Brittany, France (Planchon et al. 2009) and to investigate the variability of winter precipitation in Israel and Jordan (Black 2012).

Recently, the HB catalog was used (together with others) to analyze atmospheric circulation associated with the occurrence of tornadoes in the Czech Republic (Brázdil et al. 2020) and to evaluate circulation conditions that led to the occurrence of extremely hot days in June 2019 in central Europe and Iberia (Sulikowska and Wypych 2020) and to heat episodes in 2018 in Europe (Hoy et al. 2020).

It is evident that the HB classification has been used outside of the region it was originally developed for, and indeed, previous studies have confirmed the possibility of utilizing the classification in areas even relatively far from Central Europe, for example in Greece (Anagnostopoulou et al. 2004) or Russia (Hoy et al. 2013a). However, previous studies aimed at comparing various circulation classifications according to their synoptic-climatological applicability, i.e., their ability to describe variability of surface climate elements in different European regions, show that the applicability of the HB classification varies across Europe. For instance, according to a preliminary assessment by Huth (2010), the classification provides a relatively good stratification of maximum temperature over the majority of Europe, the main exceptions being the Iberian Peninsula, north of Scandinavian Peninsula, and parts of southeastern Europe. Casado et al. (2010) proved the classification to be insufficient to describe variability of winter precipitation over Spain. Thus, the aim of this study is to analyze the applicability of the HB classification to describing variability of climate elements, namely, minimum and maximum air temperature and precipitation, across the European continent. Moreover, attention will be paid to whether the applicability varies during the year, specifically between summer and winter.

Climate data for minimum and maximum air temperature and precipitation were obtained from the European Climate Assessment and Dataset (ECA&D) (Klein Tank et al. 2002; Klok and Klein Tank 2009). Since the best data coverage for all selected variables is achieved between 1960 and 2000 (Klok and Klein Tank 2009), the period from 1961 to 2000 was chosen for the analysis. The authors considered extending the analysis period, as it would guarantee more robust results. However, some stations would have to be excluded from the analysis, such as all Greek stations (available only until 2004).

Winter (December, January, February) and summer (June, July, August) were analyzed as they represent two opposite phases of the annual cycle, with spring and autumn generally illustrating the transfer between these two seasons. We assume that the synoptic-climatological applicability of the classification would be worse in the transitional seasons than in winter but better than in summer, as was demonstrated by Schiemann and Frei (2010) for the Alpine region. Thus, for the sake of brevity, spring and autumn are not included in the analysis.

A total of 113 stations across the European continent and the surrounding area have been selected for the analysis (Table 1; Fig. 1). Station data from some countries (e.g., Poland) were not publicly available from the ECA&D website at the time off the analysis but became available later thanks to changes in the data policies of the project participants (https://www.ecad.eu).

Map of stations used for the analysis

Full size image

During the initial inspection of the data, some missing values were detected. In most cases, they spanned one to two consecutive days, and in some cases several weeks. Time series with missing values spanning more than three consecutive days were omitted from the analysis in the given season; four stations had to be omitted completely (the initial dataset contained 117 stations).

Otherwise, the missing values (less than 1% at a given station) have been either filled in by linear interpolation or, in the case of precipitation data, taken from E-OBS—daily gridded observational dataset for precipitation, temperature, and sea level pressure in Europe, created by interpolation of the daily data from the ECA&D project database (Haylock et al. 2008).

After completing the data, each day in the analyzed period (winter and summer months from 1961 to 2000) was assigned a circulation type found in the synoptic calendar by Werner and Gerstengarbe (2010). If the synoptic type occurred in less than 1% of days in the given season, it was omitted from the analysis for this season. This applied to the NA, NEA, and NEZ types in winter and the SEA, SEZ, SA, and SS types in summer (Table 2).

The ability of the HB classification to describe variability of the surface climate elements in Europe was analyzed using the two-sample Kolmogorov–Smirnov test, following the approach described by Huth (2010). The Kolmogorov–Smirnov test compares the probability distribution of two random data samples by constructing and comparing their empirical cumulative distribution functions (cdfs). The null hypothesis asserts that these distribution functions are the same.

In this case, for each station, the cdf of the first sample X is constructed from n₁ values of the given element during the days with the given circulation type and selected season, and the cdf of the second sample Y is constructed from n₂ values of the same element for the rest of days.

Subsequently, the difference between the cdfs of X and Y is calculated. The test statistic is the maximum difference D_max, which is then compared with critical value D_c. At the significance level of α = 0.05, D_c is calculated as

where N is the total number of days in the given season. If D_max > D_c, the null hypothesis is rejected, implying that the tested samples have a different probability distribution with a 95% confidence, and in this case, that the values of the chosen climate element during the particular circulation type are well separated from the rest of the data. The test was performed for each station, all climate elements, for both seasons, and for each circulation type, except for the types that were excluded owing to a low occurrence (4 types excluded in summer, 3 in winter).

The synoptic-climatological applicability of the HB classification, i.e., its suitability for describing variability of climate elements at a given station, was quantified according to the percentage of the circulation types under which the probability distribution functions of the element differed from the rest of the values (referred to as differentiated types in this paper); the higher the percentage, the higher the applicability. Following Huth (2010), the applicability was considered “good” at stations with at least 70% of differentiated circulation types.

An example of probability distributions and cdfs of a differentiated (WZ) and non-differentiated (WS) circulation type can be seen in Fig. 2. The maximum winter temperature during days under the WZ and WS types was tested at the Praha-Klementinum station (No. 10), which was chosen due to the authors’ affiliation. It is obvious that the winter maximum temperature was higher under the WZ type (\( \overline{Tx_{WZ}}=7.6{}^{\circ}\mathrm{C}\Big) \) when compared with the rest of the days (\( \overline{Tx}=2.5{}^{\circ}\mathrm{C} \)). The test statistics were D_maxWZ = 0.040; D_cWZ = 0.102 for WZ, which led to the rejection of the null hypothesis. This corresponds to the advection of maritime air masses from the west leading to warming in winter, which is characteristic for this circulation type over much of the European continent. On the other hand, both the probability distribution and cdf of maximum temperature under the WS circulation type (\( \overline{Tx_{WS}}=3.6{}^{\circ}\mathrm{C}\Big) \) are very similar to those constructed for the rest of the types (\( \overline{Tx}=3.5{}^{\circ}\mathrm{C}\Big) \), which means temperatures do not significantly differ under WS when compared with the other types. The test statistics were D_maxWS = 0.396; D_cWS = 0.054 for WS, which lead to the acceptance of the null hypothesis.

Probability distributions and cdfs of maximum temperature in winter at the Praha-Klementinum station (No. 10) for a non-differentiated (WS) and differentiated (WZ) circulation type

Full size image

The percentage of differentiated circulation types at individual stations for each analyzed element and both seasons are displayed in Fig. 4. Stations with a good synoptic-climatological applicability are highlighted in Fig. 5. They tend to be clustered together, defining areas where the classification is a suitable tool to describe variability of the analyzed climate elements.

As is shown in Figs. 3, 4, and 5, the ability of the classification to describe variability of maximum temperature (with average of 67% differentiated types for summer, 77% for winter) is generally better than of minimum temperature (52% for summer, 74% for winter), and both are better differentiated than precipitation (41% for summer, 58% for winter). The synoptic-climatological applicability of the classification is better in winter than in summer for all the three climate elements.

Dependence of the percentage of stations at which circulation types was differentiated on the frequency of their occurrence for minimum (TN) and maximum (TX) temperature and precipitation (RR) in winter (DJF, left) and summer (JJA, right); dashed lines represent logarithmic fit

Full size image

The percentage of differentiated circulation types at individual stations for a summer (RR JJA) and b winter precipitation (RR DJF), c summer (TN JJA) and d winter minimum temperature (TN DJF), and e summer (TX JJA) and f winter maximum temperature (TX DJF)

Full size image

Stations with more = black dot / less = cross than 70% of differentiated circulation types for a summer (RR JJA) and b winter precipitation (RR DJF), c summer (TN JJA) and d winter minimum temperature (TN DJF), and e summer (TX JJA) and f winter maximum temperature (TX DJF)

Full size image

In winter, the HB classification is suitable for describing the minimum and maximum temperature in most of western, central, northern, and northeastern Europe. The lowest percentage of differentiated types was recorded at stations in the Mediterranean, the Balkans, and western Russia; nevertheless, at most of these, more than 60% of the circulation types were differentiated, which is still a fairly large number. Interestingly, in winter months, stations with the absolute highest percentage of differentiated circulation types were mostly found outside of Germany (for example station No. 53, Valkenburg, Netherlands, and No. 58, Lindesnes Fyr, Norway, with 96% differentiated types for both minimum and maximum temperature) which means that the classification is more successful in describing winter minimum and maximum temperature in some fairly distant areas than in the region it was created for.

In summer, the area of good applicability to describing temperature is smaller than in winter and shifted to the west. In the case of minimum temperatures, it spreads from Central Europe to France, Denmark, and southern Norway; for maximum temperatures, it also extends to most of the Scandinavian Peninsula, the Baltics, and the British Isles. On the other hand, the applicability is limited in the Mediterranean, southeastern and Eastern Europe, and Iceland. For summer minimum temperature, the area with the limited applicability also includes most of the Scandinavian Peninsula.

In describing the variability of precipitation, the classification is less successful than for temperature. Summer precipitation, in particular, is difficult to be stratified outside of Germany and its close vicinity. The lowest percentage of differentiated circulation types was recorded at stations in the Mediterranean and in eastern and southeastern Europe. At eight stations, none of the circulation types was differentiated: Spanish stations Alicante (No. 90), Malaga (No. 92), and Madrid (No. 91) and all Greek stations except for Larissa (No. 38). A rather low percentage of differentiated types was also recorded in the north of the Scandinavian Peninsula and in Iceland.

The applicability of the classification to describe winter precipitation is considerably better. Stations with more than 70% of differentiated circulation types were located largely in western and northwestern Europe and Norway. This category also includes three Spanish stations where in summer, less than 10% circulation types were differentiated. Overall, the ability of the HB classification to describe variability of precipitation is best in Germany, the Netherlands, and southern Norway. On the other hand, for both seasons, it is limited in the Mediterranean, eastern and southeastern Europe, and north of the Scandinavian Peninsula.

Results of the K–S test for individual circulation types are summarized in Table 3. On average, the most differentiated circulation types are TRM (69%), WA (66%), and BM (66%) in summer and WZ (90%), WA (83%), and HM (83%) in winter. The least differentiated types are HNZ (31%), HNFZ (38%), and WW (39%) in summer and TRW (36%), TB (41%), and NWA (47%) in winter.

For most of the circulation types, maximum temperature is better differentiated than minimum temperature, and both are better differentiated than precipitation, which can be seen in Fig. 3. Furthermore, there is a connection between the frequency of occurrence of a circulation type and the percentage of stations where it was differentiated (see Table 3; Fig. 3), as more common types seem to have better results. This might be to some extent related to the method used for the analysis as the K–S test favors larger-sized samples: Eq. (2) indicates that the critical value of the test decreases with the growing size of the type; that is, the more often a circulation type occurs, the greater the chance that the null hypothesis will be rejected for the same D_max. This means that more common types could be differentiated because they occur more frequently and not because of their physical and meteorological properties. However, mapping the results of the K–S test for the individual types shows that areas with the acceptance or rejection are geographically coherent, which indicates that the results do reflect actual properties of the types.

The highest result, 96%, was achieved by two circulation types, BM and WZ, for winter maximum temperature. Both circulation types were differentiated at all but four stations, two of which are in Turkey, well outside of the “home” domain of the HB classification (Fig. 6).

The results of the K–S test for the BM and WZ circulation types for winter maximum temperature; circulation type is differentiated = large dot / not differentiated = small dot

Full size image

The absolute lowest result, 23%, was achieved for the TRW circulation type for winter precipitation and the HNZ type for summer precipitation. Whereas the result for HNZ does not differ substantially from types with a similar frequency in that season (2.2%), the TRW circulation type scores rather low when compared with other types with a similar frequency of occurrence in winter (1.7%), especially the SA circulation type (Fig. 7). Interestingly, the type is differentiated at some stations in areas where the applicability of the HB classification tends to be limited, such as the Iberian Peninsula (Fig. 7). We briefly discuss these results in the next section.

The results of the K–S test for the TRW and SA circulation types for winter precipitation; circulation type is differentiated = large dot / not differentiated = small dot

Full size image

The SA type was omitted from the analysis for summer due to its very low occurrence in the season (only 0.5%). In winter, it scored exceptionally well (RR 79%, TN 79%, TX 69%), especially when compared with similarly common types. SA is characterized by a warm southerly airflow into Central Europe directed by an extensive blocking anticyclone over Eastern Europe and a low air pressure area over the eastern Atlantic and parts of Western Europe. The Atlantic frontal zone runs from the sea area north of the Azores to Southwestern Europe, where it turns north. Storm tracks affect Southwestern and Western Europe, while mostly dry conditions prevail elsewhere. Over the Iberian Peninsula, the advection of maritime air masses from the Atlantic leads to precipitation distinct enough to be differentiated from the other types, which is not the case for stations in France, northern Italy, Great Britain, and southern Norway (Fig. 7). In winter, the warm advection during the SA type sometimes manifests only in higher altitudes, while a layer of cold air stays near the surface, which may explain the better results for minimum (79%) than maximum (69%) temperature.

TRW is defined by a trough extending from the Arctic Ocean to the Western Europe and the Iberian Peninsula flanked by areas of high air pressure over the middle Atlantic and Western Russia. Frontal zone runs from the middle Atlantic to Spain and from there towards the northeast via western Central Europe to Scandinavia. On the western side of the trough, relatively warmer (in winter) or colder (in summer) maritime air is advected from the Atlantic into western and southwestern Europe. TRW is one of the least differentiated circulation types in winter (RR 23%, TN 40%, TX 46%). In Fig. 7, we can see that for winter precipitation, it is differentiated mainly in the areas most affected by the low air pressure trough. The results are also relatively poor for winter temperatures, which are differentiated on all stations in Iceland, and most of the stations in Norway or Turkey, while in most parts of Europe, including Central Europe, they remain undifferentiated (not shown). In summer, its results are distinctly better (RR 50% TN 54%, TX 75%), and it is differentiated mainly in western (for all elements) and southwestern Europe (for temperatures), i.e., again in places most affected by the trough and the advection of maritime air masses connected therewith. Interestingly, the differentiation is also very good in the eastern half of Europe for maximum temperature, which is related to the warm, southerly advection characteristic for areas on the eastern side of the trough. This agrees well with the fact that the TRW type is often related to the occurrence of hot days (Sulikowska and Wypych 2020) and heat waves (Kyselý 2007; Unkašević and Tošić 2009) in the area, which is also true for the SA type.

The HB classification was originally created for the purpose of weather forecasting in Central Europe; however, previous studies have confirmed that it can be used relatively far outside this area, for example in Greece (Anagnostopoulou et al. 2004) or Russia (Hoy et al. 2013a). The aim of this study was to quantify the usability of the HB classification over the European continent. The K–S test was used on data from 113 meteorological stations located throughout Europe to determine whether air temperature (minimum, maximum) and precipitation at these stations behave differently under a particular circulation type when compared with other types. According to the percentage of types that were differentiated at a station for a given element, we derived the applicability of the classification to describe the element at said station. The applicability is considered “good” at stations with at least 70% of differentiated circulation types.

The results show that the HB classification is applicable to describing minimum and maximum temperature across much of the European continent, usually well outside the region it was originally constructed for. This agrees with the study by Hoy et al. (2013c), which investigated the impacts of atmospheric circulation on air temperature in Europe and northern Asia using HB types rearranged into four new circulation forms (westerlies, easterlies, southerlies, northerlies) according to similar air mass inflow into Central Europe (for further reading, see Hoy et al. 2013a). The resulting anomaly maps of air temperature for these main circulation forms illustrated the spatial extent of their influence and signal strength. Similarly to our results, both were more pronounced in winter and less visible in summer. In winter, all circulation forms affected air temperatures over the most of the European continent, and the signal spread as far as eastern Russia for the easterlies (types such as HFA/Z and HNFA/Z). On the other hand, the signal tended to be weakest over the Mediterranean, particularly in Greece and Turkey. In summer, the anomaly patterns were less pronounced, with the most prominent ones being in Central Europe or its vicinity.

For precipitation, good applicability tends to be limited to Germany and its vicinity, especially in summer. This is in agreement with other studies comparing the HB classification and different circulation classifications according to their synoptic-climatological applicability in various European regions (Casado et al. 2010; Huth 2010; Ustrnul et al. 2010). The worst performance was observed in the Mediterranean, with some stations in Spain and Greece scoring 0% differentiated types for summer precipitation. The limited performance of the HB classification in the Mediterranean is related to the different climate conditions found there. During summer, the atmospheric circulation is rather stable, and weather conditions are influenced by the South Asian thermal low in eastern and the Azores anticyclone in western and central Mediterranean (Kostopoulou and Jones 2007). Precipitation occurs mostly during the winter half-year and is more sensitive to the complex orographic features of the Mediterranean and local circulation systems. The applicability of the classification was higher for winter precipitation, with three stations in Spain scoring over 70% (Madrid, No. 91; Salamanca, No. 93; and Badajoz, No. 113). A closer look at the results revealed that most of the differentiated types in this area belonged to southerlies, northerlies, and easterlies circulation forms (as defined by Hoy et al. 2013a). These results are in good agreement with a study carried out by Hoy et al. (2014), in which the aforementioned forms were connected to positive precipitation anomalies in this area in the winter half-year.

The HB classification is more successful in describing air temperature than precipitation. This is due to the greater dependence of the occurrence of precipitation and rainfall amounts on influences, such as orography and local circulation, which cannot be as well distinguished by circulation classifications designed for large geographic areas (Kostopoulou and Jones 2007; Tveito 2010). This is particularly true for convective precipitation.

Of course, even air temperature can be influenced by local factors, such as elevation, orography, and distance from the sea. In the case of summer minimum temperature, some stations relatively close to each other differed by more than 20% of differentiated circulation types, for example stations Nimes (No. 18; 60%) and Mont-Aigoual (No. 17, 84%) in France. We hypothesize that this is caused by different orography, as the station with fewer differentiated types is situated in a valley where the climate elements recorded would be less influenced by the large-scale atmospheric circulation than those recorded on the mountain station, where conditions are much closer to free atmosphere, especially in summer.

The fact that the applicability of the classification is better in winter than in summer can be attributed to a weaker influence of the atmospheric circulation on surface climate elements in the summer season and a stronger influence of solar radiation and local factors such as elevation, orography, and distance from the sea. This applies in particular to summer precipitation, which tends to be convective, heterogeneously distributed and, as was already stated above, harder to capture by large-scale circulation classifications. The limited applicability to describing precipitation in summer is characteristic for other classification methods as well; it was demonstrated, e.g., for Belgium (Brisson et al. 2011), the Iberian Peninsula (Cortesi et al. 2014), and Italy (Vallorani et al. 2018).

Table 3 shows the percentage of stations at which a specific circulation type was differentiated. This information is useful because it shows how much (over how large an area) each type affects climatic elements within Europe. But based on these statistics alone, it is not possible to say whether a type is good or bad. A type that is poorly differentiated for precipitation might be more successful for another element. Types differentiated at a small number of stations in summer may have a greater effect during winter. Rejection or acceptance of the K–S test was mapped for all individual circulation types, selected examples are displayed in Figs. 6 and 7. Areas with rejection or acceptance of the test were relatively geographically coherent; for well-differentiated types, we can see clusters of stations at which the type was not differentiated, and vice versa. This means that even circulation types that show poor results in Table 3 may contain important information. There was no room left in this study to examine the spatial differentiation of all individual types and its synoptic-climatological causes; therefore, we only limited ourselves to the description of two examples in the previous section: SA and TRW. These two types were chosen mainly because of their similar frequency of occurrence yet distinctly different performance in winter, with TRW being much less differentiated. The reason for this was not further investigated in this study and would require additional research; however, we wonder if wrong assignment of a circulation type could play a role in this case, as James (2007) stated that TRW is one of the circulation types that are assigned more often in the HB catalog than it would be according to its objective version (OGWL).

In this study, the values of the selected climate elements during the days with a particular circulation type were compared with their values under the rest of circulation types. A question arises whether it is possible for two (or more) circulation types to be differentiated from the rest of the data but be similar to each other. As Tveito (2010) demonstrated in his study comparing the applicability of classifications gathered in the COST733 catalog to describing precipitation in Norway, there were indeed cases where the distribution function of a circulation type was well differentiated from the rest of the data but not from every other type. However, this seemed to concern mainly classifications with fewer (less than 10) types and was much less likely to occur with classifications with a higher number of types (particularly 26 or more).

The HB circulation classification was originally invented for the purpose of weather forecasting in Germany; however, nowadays, it is often used for different purposes, mainly synoptic-climatological studies, and in other parts of Europe as well. The aim of this study was to analyze the applicability of the HB circulation classification to describing variability of minimum and maximum temperature and precipitation across the European continent.

The ability of the HB classification to describe variability of the surface climate elements in Europe was analyzed using the two-sample Kolmogorov–Smirnov test which compares the probability distribution of two random data samples by constructing and comparing their empirical cumulative distribution functions (cdfs). The synoptic-climatological applicability of the classification was quantified according to the percentage of the circulation types under which the probability distribution functions of the element differed from the rest of the values (i.e., the percentage of the circulation types for which the null hypothesis was rejected). The applicability was considered good at stations with at least 70% of differentiated circulation types.

The classification proves to be effective at describing the variability of the maximum and minimum temperature across most of the European continent, especially in winter. As expected, stations with the highest percentage of differentiated circulation types are usually located in Germany and surrounding countries. In describing the variability of precipitation, the classification was less successful than for minimum and maximum temperature. This is due to the greater dependence of the occurrence of precipitation and rainfall amounts on local influences, such as orography and local circulation, which cannot be as well distinguished by circulation classifications designed for large geographic areas.

The applicability of the classification is better in winter than in summer for all analyzed climate elements—summer precipitation, in particular, is difficult to be stratified outside of Germany and its close vicinity. This can be attributed to a weaker influence of the atmospheric circulation on surface climate elements in the summer season and a stronger influence of solar radiation and local factors such as elevation, orography, and distance from the sea.

It should be noted that the limited synoptic-climatological applicability of the HB classification in the abovementioned regions does in no way mean it cannot be used for synoptic-climatological research in these areas. For example, it could still serve to identify circulation patterns responsible for events such as extreme rainfall and cold or heat waves, although there are probably more suitable regional classifications available.

Affiliations

1. Department of Physical Geography and Geoecology, Faculty of Science, Charles University, Prague, Czechia

   Petra Sýkorová & Radan Huth

2. Central Forecasting Office, Czech Hydrometeorological Institute, Prague, Czechia

   Petra Sýkorová

3. Institute of Atmospheric Physics, Czech Academy of Sciences, Prague, Czechia

   Radan Huth

Corresponding author

Correspondence to Petra Sýkorová.

Conflict of interest

The authors declare that they have no conflict of interest.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Reprints and Permissions