Skip to main content
Download PDF

Comparison of extreme precipitation characteristics between the Ore Mountains and the Vosges Mountains (Europe)

Theoretical and Applied Climatology volume 133, pages 1249–1268 (2018)Cite this article

Abstract

Understanding the characteristics of extreme precipitation events (EPEs) not only helps in mitigating the hazards associated with it but will also reduce the risks by improved planning based on the detailed information, and provide basis for better engineering decisions which can withstand the recurring and likely more frequent events predicted in future in the context of global climate change. In this study, extremity, temporal and spatial characteristics, and synoptic situation of the 54 EPEs that occurred during 1960–2013 were compared between two low mountain ranges situated in Central Europe: the Ore Mountains (OM) and Vosges Mountains (VG). The EPEs were defined using the Weather Extremity Index, which quantifies the extremity, duration, and spatial extent of events. Comparative analysis of EPE characteristics showed that in both regions the EPEs were mostly short (lasted 1–2 days) and their seasonal occurrence significantly depended on the synoptic situation and duration of EPEs; the low was related to summer short EPEs, while zonal circulation to winter long EPEs. The EPEs were generally related to lows in OM and to troughs in VG. The lows often moved to OM from the Mediterranean area, i.e. along the Vb track. However, five EPEs in VG occurred during a low with Vb track significantly deflected westwards. The EPEs in VG affected smaller area as compared to that in OM. The comparison of EPEs between the two low mountain ranges is first of its kind and contributes to the understanding of EPE characteristics in the regions.

Introduction

Extreme precipitation has been the focus of atmospheric sciences since 1990s due to its direct impacts, such as the threat posed to the safety of transport, and the indirect impacts such as flooding, erosion, and landsliding which affect large areas even beyond the area of the rainfall occurrence. To be able to reduce these impacts (e.g. loss of lives, large-scale damages to agriculture resources and property, and contamination of clean water), the emphasis on recognition, description, and prediction of precipitation extremes has become more crucial specially in the context of global climate change (Beniston and Stephenson 2004), i.e. an increase in the frequency of weather and climate extremes has been predicted (Pachauri et al. 2014). As documented by simulations of the development in the twenty-first century by Euro-Cortex, almost all European countries might experience an increase in the frequency of extreme precipitation (Söder et al. 2009; Vautard 2013).

Despite the improved prediction of heavy rainfall and enhanced communication with decision makers to issue warnings in Europe (Thieken et al. 2007; Kienzler et al. 2015), considerable causalities and dire financial impacts were induced by the two relatively recent episodes: the heavy rainfall events and related floods in the middle Danube and the Elbe catchments in 2002 and 2013 in Central Europe (Van der Schrier et al. 2013; Thieken et al. 2005; Brazdil et al. 2006; Boucek 2007). It demonstrates the ongoing vulnerability of European societies to weather extremes and demands more detailed insight into the characteristics and conditioning factors of heavy rainfall (e.g. synoptic conditions) in Europe at diverse temporal and spatial scales to make the risk management and warning systems more efficient (Thieken et al. 2007; Socher and Boehme-Korn 2008; Kienzler et al. 2015).

Since the spatial distribution of (mean) precipitation in orographic areas is very complex and not all the processes have satisfactorily been understood (Prudhomme and Reed 1998; Roe et al. 2003; Smith 2006), the spatial distribution of precipitation extremes in orographic areas is even more complicated and needs further attention. Recent papers dealing with heavy rainfall in orographic areas in Europe mostly considered the Alps and the Carpathian Mountains (e.g. Bartholy and Pongracz 2005; Bartholy and Pongrácz 2007; Foresti and Pozdnoukhov 2012; Awan and Formayer 2016). However, in Central Europe, there are many low mountain ranges which are densely populated (especially on their leeward side) as compared to the Alps and the Carpathian Mountains, thus more vulnerable to the damages associated with natural disasters. In addition, the future projections of heavy rainfall in the region are vague (Solomon et al. 2007; Pachauri et al. 2014), which makes the region of Central Europe more appealing for further analyses (Alexander et al. 2006; Solomon et al. 2007; Pachauri et al. 2014).

The current study focuses on several characteristics of extreme precipitation events that are compared between two low mountain ranges situated in Central Europe (Section 2.1): Vosges Mountains (northeastern France) and the Ore Mountains (also named as Krušné hory or Erzgebirge at the Czech-German border). The selection of study areas is related to the orographic effect that is responsible for large difference in precipitation totals between the windward and leeward sides; with the leeward sides considered to be one of the driest regions of the respective countries, i.e. France (Sell 1998) and the Czech Republic (Brádka 1963; DWD DDR and HMÚ ČSSR 1975; Pechala and Böhme 1975; Tolasz et al. 2007).

Concerning the studies about precipitation in the Ore Mountains, past and present variations were analysed separately for the Czech Republic (Tolasz et al. 2007) and Saxony (Franke et al. 2004; Küchler and Sommer 2005) including the Dresden region during reference period 1961–1990 as compared to 1991–2005 in the REGKLAM project (Bernhofer and Surke 2009; Heidenreich and Bernhofer 2011). The changes in extreme precipitation in Saxony during 1901–2100 were also studied by Hänsel et al. (2015). Brázdil (2002) studied the atmospheric extremes and related floods in the (whole) Czech Republic with respect to the global climate change. Only the project INTERKLIM (2014) discussed the variations and projections (until 2100) in precipitation and heavy rainfall (above 95th and 99th percentile) cross-country, i.e. over the Saxon-Bohemian area thus covering the whole Ore Mountains. Nevertheless, the heavy rainfall was paid less attention in the project since it was more focused on variations and trends in precipitation with respect to the climate change.

Individual extreme precipitation events or flood events affecting Czech parts of the Ore Mountains including the Ohře river basin were described in older studies (Hladný and Barbořík 1967; Kakos 1975, 1977; Kynčil and Lůžek 1979; Kynčil 1983; Chamas and Kakos 1988) except Brázdil et al. (2005) who discussed 2 years and longer flood events in the (whole) Czech territory, and Štekl et al. (2001) who analysed heavy rainfall events (daily totals exceeding 150 mm) in the Czech Republic. In Germany, Zolina (2014) studied the changes in wet spells (daily totals exceeding 1 mm). The known extreme precipitation event of August 2002 when the maximum daily precipitation total of 312 mm was measured in the Eastern Ore Mountains at Zinnwald weather station on August 12, 2002 (Munzar et al. 2011) was largely discussed by many authors (Brázdil 2002; Rudolf and Rapp 2002; Ulbrich et al. 2003; Brazdil et al. 2006; Boucek 2007; Socher and Boehme-Korn 2008; Conradt et al. 2013; Kienzler et al. 2015). The June 2013 heavy rainfall event also got wide attention of the authors and was analysed from many perspectives (e.g. Stein and Malitz 2013; Merz et al. 2014; Grams et al. 2014; Schröter et al. 2015). Nevertheless, the studies were mostly country- or Central Europe-delimited and provided event-specific results rather than describing typical characteristics of extreme precipitation. Thus, a regional analysis of a dataset of extreme precipitation events covering the Ore Mountains was needed.

In the Vosges Mountains, only a very local case study has been recently conducted to examine an issue related to heavy rainfall, i.e. the leeward convection under the COPS campaign (Labbouz et al. 2013; Planche et al. 2013). Other recent papers were more focused on expected changes in extreme precipitation and their uncertainties in the Rhine river basin (Bosshard et al. 2013; Pelt et al. 2014) and southern Germany (Söder et al. 2009). The project REKLIP (Parlow 1996) provided a climatological overview of the Upper Rhine area including the weather patterns dominating over the region through a year. Local case studies or studies describing particular extreme precipitation/flood event are given in older literature sources such as Baulig (1950), Hirsch (1972), Maire (1979), Fink et al. (1996), and van Meijgaard and Jilderda (1996). An analysis of extreme precipitation in the Vosges Mountains was therefore missing.

Thus, we performed an analysis of heavy rainfall in the Vosges Mountains and the Ore Mountains separately (Minářová et al. 2017a, d) using the event-adjusted evaluation method for precipitation extremes proposed by Müller and Kaspar (2014). This paper provides a comparative analysis of the characteristics of the extreme precipitation between the two study regions and extends the results mostly concerning the synoptic conditions during the extreme precipitation in the regions. The attributes of the extreme precipitation that are compared in this study have been defined the same way in both areas, which makes their comparison more robust as compared to the works of previous publications that were site/event specific, and used different definitions of heavy rainfall. The results of the comparison in this paper are first of its kind and contribute to understand the patterns of heavy rainfall and its characteristics in the two low mountain ranges in Central Europe, and thus might help in mitigating the natural disasters and subsequent losses associated with extreme precipitation.

Data and methods

Study areas

The study areas generally follow the boundaries of the administrative units comprising the Ore Mountains and the Vosges Mountains. At places, the boundaries were reduced corresponding to the spatial distribution of the weather stations (i.e. the large extra areas in the administrative units beyond the meteorological stations were omitted from the selection) in order to reduce the need of extrapolation of weather data. The two study areas, i.e. Ore and Vosges mountains, have some morphological and relief-related climatological similarities while they differ in the mean annual course of precipitation, as described in the following.

Ore Mountains region

The study area comprising Ore Mountains (OM) and its surrounding area is situated at the Czech-German border (Fig. 1a). The Ore Mountains is a low mountain range, which culminates at Klínovec Mountain (1244 m a.s.l.). The slopes on German side are gentle as compared to the slopes on the Czech side. Typical climate in OM is temperate with the western major airflow from the Atlantic Ocean and is transitional from the oceanic climate that dominates in Western Europe to a continental climate that prevails in Eastern Europe (DWD DDR and HMÚ ČSSR 1975).

Fig. 1
figure1

Study area of (a) the Ore Mountains and (b) the Vosges Mountains, and the spatial distribution of the (a) 167 and (b) 84 analysed rain gauges. The relief is represented in colour scale, i.e. the highest locations are displayed in white. Small schematic maps display the categorization of the two areas as given in Section 2.6.2

Full size image

The main precipitation season is summer, although a secondary winter maximum can be found in mountains. The orographic effect on precipitation, primarily related to the almost perpendicular orientation of the mountain range against the prevailing airflow direction, is mostly responsible for the differences in mean precipitation totals between the (wetter) windward German side including the highest altitudes due to the orographic enhancement of precipitation and the (drier) leeward Czech side due to rain shadow (DWD DDR and HMÚ ČSSR 1975; Pechala and Böhme 1975).

Vosges Mountains region

The study area comprising Vosges Mountains (VG) is situated in northeastern France (Fig. 1b), and constitutes a broader area of the low mountain range, which culminates at Grand Ballon (1424 m a.s.l.). Likewise OM, the VG has gentle windward (western) slopes and steeper leeward (eastern) slopes that dip towards the Upper Rhine Plain (Gley 1867; Alsatia 1932; Ernst 1988; Sell 1998). VG represents a frontier between the temperate oceanic climate in its western part, and continental in the eastern part, mainly Upper Rhine Plain. It also includes microclimatic peculiarities (Sell 1998; Météo-France 2008).

Similar to OM, elevation, prevailing westerlies from the Atlantic Ocean, and the orographic effect related to the nearly perpendicular position of the mountain ridge to the prevailing airflow are among the most important factors responsible for differences in precipitation in the region (Sell 1998; Météo-France 2008). In VG, the differences in mean annual precipitation totals between the wettest and driest stations during 1960–2013 were up to 1730 mm due to the orographic enhancement of precipitation on one side, and the rain shadow on the other (Minářová et al. 2017b). Contrary to OM, the main precipitation season is winter in mountains, although in the Upper Rhine Plain it is summer (Alsatia 1932; Ernst 1988; Sell 1998).

Precipitation time series

In this paper, the daily precipitation totals during 1960–2013, obtained from Météo-France, Deutscher Wetter Dienst (DWD) and Czech Hydrometeorological Institue (CHMI) rain gauging networks, have been analysed. The metadata (e.g. changes in location, measuring instrument) was also acquired with the datasets.

The analysed datasets include data obtained at 168 meteorological stations in VG and 167 meteorological stations in OM. The data from Czech (leeward) side of the OM are available for ten weather stations and span from 1960 to 2005 only. It may affect the results but not significantly, since at regional scale, a higher uniformity of weather patterns is found on the (Czech) leeward side as compared to the (German) windward side (Whiteman 2000; Barry 2008). The assumption is particularly valid for large-scale precipitation; in the case of convection-related precipitation, it might affect the results. However, the methodology of selection of heavy rainfall events (Section 2.5) prioritizes the large-scale precipitation so that a negligible number of events will be influenced. Moreover, in our case, the typical scale of the extreme precipitation events exceeds the density of the stations, i.e. all the events were captured by at least 20 of the stations, which confirms that data from only ten stations on the Czech side do not significantly affect the robustness of the results.

Due to the installation of weather stations in VG with time, and the installation and shutting down of weather stations in OM, not all the stations could record data for the entire study period (54 years). In order to obtain well-fitted Generalized Extreme Value distribution (Section 2.5), only the stations which recorded data for more than half of the study period (i.e. 27 years) were used for identification and characterization of the most extreme precipitation events in the study areas. The 27 years of observations were not bound to the beginning or the end of the 54-year period. In OM, all the 167 stations, while in VG, only 84 out of 168 (half of the stations) measured the daily precipitation totals for more than 27 years, due to the increasing installation of rain gauges increased with time. The criterion of omitting the time series from stations that did not record data for more than half of the study period resulted in VG in an increase in the daily data availability from 35 to 62% in the 1960s, and from 50 to almost 100% since the 1980s, which substantially improved the robustness of the results. Moreover, the criterion of 27 years of observations did not substantially influence the spatial distribution of stations because the omitted stations were randomly distributed and not confined to any specific part of VG. Thus the spatial distribution of stations did not get more uneven, only the number of representatives per spatial unit got reduced.

Relocation of stations and changes in measuring devices or its principles introduce inhomogeneities in the time series. RHtests_dlyPrcp R-package http://etccdi.pacificclimate.org/software.shtml (Wang et al. 2010; Wang and Feng 2013) was conducted to test whether the daily precipitation time series are homogeneous. The test considered the metadata including the changes in measuring devices. No significant relocation and inhomogeneities were noticed for the Czech rain gauges during 1965–2005 (Kyselý 2009), thus a fixed data measurement error of 0.2 mm was used for OM as suggested by the WMO (World Meteorological Organization 2008). In VG, a value of 0.4 mm was used while considering the maximum error estimated for the changes in rain gauges. Minářová et al. (2017b) have stated that lower values (0.2 and 0.3 mm) produce similar results. No major inhomogeneities were noticed in the time series, except for two stations in VG which were homogenized. However, the difference between the raw data and the homogenized data of the two stations is insignificant, i.e. lower than the resolution of the time series (in the order of 10−2 mm). Thus, despite minor inconsistencies in the three national weather networks, the results from the analysed time series can be assumed robust.

Further analysis (Section 2.5) of the time series was based on 1–10 days consecutive non-zero precipitation totals from the individual stations. The threshold of 10 days was assumed to be sufficiently high since longer lasting extreme precipitation events were not awaited to occur in any of the study areas. The length of events shorter than 10 days was not considered adequate based on the study from Pelt et al. (2014), who suggested that mainly the 10-day rainfall events are prone to induce flooding in Upper Rhine river basin, i.e. in VG.

The uneven spatial distribution of stations was considered not to substantially influence the robustness of our results since during the process of definition of heavy rainfall events (Section 2.5) only the common logarithms of return period estimates from stations are interpolated into a regular grid. The common logarithms of return period estimates (and the return period estimates) exhibit flat distribution, which makes their interpolation to the uneven spatial distribution of stations much less sensitive as compared to the interpolation of precipitation totals (e.g. Šercl 2008). Since the resulting Weather Extremity Index (Section 2.5) calculated for individual extreme precipitation events in the study area is defined from the regular grid, it is negligibly influenced by the inhomogeneous spatial distribution of stations. The area affected by individual events is also derived from the regular grid so that the possible uncertainties related to the uneven distribution of the stations are reduced.

Synoptic variables

Synoptic variables (wind velocity, geopotential height, and flux of specific humidity) at 500 and 850 hPa isobaric levels (measured at 12 UTC) were derived from the NCEP/NCAR (National Centers for Environmental Prediction/National Center for Atmospheric Research) data reanalysis (Kalnay et al. 1996) in gridded form at 2.5° horizontal resolution for the period 1960–2010 (Uppala et al. 2005). The gridded form at higher horizontal and temporal resolution of the synoptic data was not used in this study since the focus was on large-scale synoptic situation. Data at 12 UTC (except geopotential height) were averaged from six grid points covering each study area (10–15°E and 50.0°–52.5°N in OM, while 5–10°E and 47.5°–50.0°N in VG). The averaged values were used in the analysis of synoptic conditions occurring during the extreme precipitation events. If an extreme precipitation event lasted longer than 1 day, the value of the day with the highest daily extremity of precipitation E ta (defined in Section 2.5) was assigned to the event. The large-scale synoptic categorization during (and prior to) each analysed heavy rainfall event was done based on detailed check of synoptic maps (including e.g. temperature field) at 500 and 850 hPa levels obtained from http://www.wetterzentrale.de/ at 6-h temporal resolution and from NOAA https://www.esrl.noaa.gov/psd/cgi-bin/data/getpage.pl available at daily scale.

Meridional and zonal airflow components and the components of the flux of specific humidity were computed to know the direction of the airflow and the flux of specific humidity, respectively. The directional fluxes of specific humidity were considered to provide relevant information about the extreme precipitation (Müller et al. 2009). An approach directly based on synoptic (quantitative) data was also suggested to reflect the synoptic conditions during precipitation extremes (Müller and Kašpar 2010; Kašpar and Müller 2014) in a better way than the qualitative approach based on the assignment of weather types over Europe such as the widespread “Grosswetterlagen” concept (Werner and Gerstengarbe 2010).

Digital elevation model, cartographic outputs, and interpolation

For the relief-related information, digital elevation models (DEM) comprising the two study areas were obtained from GeoMapApp (http://www.marine-geo.org/tools/maps_grids.php). The horizontal resolution of the GeoMapApp’s gridded Global Multi-Resolution Topography model is 100 m. The map outputs were produced in Esri’s ArcGIS 10.5 software, where the DEMs were used as base maps, and the synoptic outputs were performed in Golden software Surfer 10.

Ordinary Kriging with raster cell size of 2 km was used for interpolating the common logarithms of return period estimates into a regular grid (procedure given in Section 2.5). The Ordinary Kriging was based on Gaussian semi-variogram model, and the maximum searching radius was set to variable. Co-Kriging or other geostatistical methods with external drifts that could include orography in the interpolation were not considered in this paper since no influence of orography on return period estimates was proved and the return period estimates were thus found not sensitive to orography (Šercl 2008).

Precipitation extremes: event-adjusted evaluation method

Precipitation extremes were defined using the event-adjusted evaluation method proposed by Müller and Kaspar (2014), which allows for quantitative estimation of the extremity of individual heavy rainfall events and their comparison using the variable extremity E ta for a given duration t of an event which affects an area a. At the beginning, return period estimates of precipitation totals (1–10 days in this study) are used to assess the rarity of the totals. The return period estimates are computed at individual rain gauges using the Generalized Extreme Value (GEV) distribution with a maximal value of 1000 years. The approximation of precipitation totals by GEV was found convenient on the basis of the goodness of fit test based on the L-kurtosis τ4 of the fitted distribution and the regional average L-kurtosis τ4R (Hosking and Wallis 1997). This is in good agreement with Kyselý and Picek (2007), who have shown that the GEV approximates the precipitation time series well and is suitable for the estimation of extreme precipitation events in the Czech Republic. Common logarithms of return period estimates calculated at individual gauges are subsequently interpolated into a regular grid (2 km horizontal resolution) using Ordinary Kriging described above (Section 2.4). In the next step (computation of E ta ), the values of return period estimates at resulting grid points are taken one by one in their decreasing order, i.e. irrespective of their position in the study region so that more cells of heavy rain can be detected during one single event.

The E ta corresponds to the multiplication of the radius of a circle R [km] over an area a [km2], that is equal to the area consisting of i number of included grid points, and the common logarithm of the spatial geometric mean G ta of return period estimates N ti [years] for a given duration t [days], i.e. the E ta (Müller and Kaspar 2014):

$$ {E}_{ta}\left[\log \left(\mathrm{year}\right)\mathrm{km}\right]=\log \left({G}_{ta}\right)R=\frac{\sum_{i=1}^n\mathit{\log}\left({N}_{ti}\right)\sqrt{a}}{n\sqrt{\pi }} $$
(1)

Based on the step-by-step inclusion of grid points with lower and lower return period estimate, the E ta stops increasing at one point (maximum E ta ), i.e. the enlarging area does not counterbalance the inclusion of substantially reduced values of return period estimates. This maximal value of E ta is taken as the Weather Extremity Index (WEI) value, and the corresponding area a is the area affected by a heavy rainfall event. However, the WEI varies with duration t of the event (1–10 days considered in this study). The final duration of the event is determined as the first maximal E ta value consecutively calculated for 1 day, 2 days up to 10 days long events, where all the events must overlap, and their 1-day (daily) E ta values must be above zero, i.e. the daily precipitation totals during the event are significantly high or extreme. The given duration of the event determines the final WEI value of the event, and thereby the size of the area that it affected.

The WEI provides quantitative information about the extremity of weather events including the size of the area affected by an event, which is adjusted along with the rarity (return period estimates) and duration based on the two foregoing characteristics (area and rarity) of the event, i.e. the WEI reflects three important characteristics of extreme weather events. Further details can be found in the original work of Müller and Kaspar (2014) about the WEI.

The smooth transition from extreme to non-extreme precipitation events signifies that no critical value of WEI can be suggested to differentiate between the extreme and less extreme events, i.e. the researcher should fix the dataset of further analysed events, e.g. with respect to the length of the study period, climatological features of the study region, and the aim of the study. Either a specific WEI value threshold (e.g. WEI = 30) or an arbitrary number of precipitation events (e.g. 3, 10, 20 events) can be used to fix the dataset. In this paper, 54 extreme precipitation events (EPEs hereafter) from each study area have been compared since it implies on average one EPE per year during the study period.

Comparative methods

Different characteristics of EPEs (duration, affected area, extremity, and synoptic conditions) in OM and VG were expressed as categorical variables (described below) in order to test the in/dependence of the variables and to obtain comparable results between the two regions.

Based on a contingency table between the pairs of variables (e.g. duration and affected area), the Pearson’s chi-squared test of independence (Greenwood and Nikulin 1996) was calculated at 1% significance level. When the test resulted in chi-squared value χ2 exceeding the critical value of χ2 at the 1% confidence level, the null hypothesis (i.e. two variables are independent) was rejected, the chi-squared residuals examined, and the Cramér’s V (Cramér 1946) calculated. The Cramér’s V is a measure of the association between the two variables and it varies from 0 (i.e. no association between the two variables) to +1 (i.e. the two variables are identical). Cramér’s V shows the percentage of the maximum possible variation of the two variables, and its square is considered the mean square correlation between the two variables. Since the Cramér’s V tend to be 1 without meaningful evidence of correlation with increasing difference between the number of rows and number of columns, and the χ2 values tend to increase with the number of cells, the derived categorical variables of the EPEs characteristics were defined to maximum four categories.

Temporal characteristics of EPEs

Two categories of EPEs were defined on the basis of the frequency of durations of EPEs: short EPEs (lasting 1–2 days) and long EPEs (3–10 days). The distinction corresponds to the frequency distribution of 1–10 days EPEs in the dataset of EPEs, with 1–2 days (short) EPEs occurring much more frequent as compared to the 3–10 days (long) EPEs in both OM and VG (Minářová et al. 2017a, d). Two and four categories of EPEs were defined based on their occurrence in halves of the year (summer half-year SHY from April to September/winter half-year WHY from October to March) and meteorological seasons (e.g. spring covering calendar days from March 01 to May 31), respectively. The occurrence of EPEs in SHY/WHY and seasons was assigned according to the calendar date of the first day of the EPE. A sensitivity analysis proved that the selection of the first day as compared to second and up to 10th day of the EPE has no significant influence on the seasonal distribution of EPEs (only up to 2–3 EPEs from 54 EPEs in OM and VG were influenced by the change of the assigned date).

Spatial characteristics of EPEs

For easy comparison between the OM and VG, the area affected by the EPEs was expressed as the percentage of the total of each study area. Four categories of EPEs were defined based on the percentage of the area that the EPEs affected, as follows: local EPEs (affecting less than 20% of the study area), district EPEs (affecting 20–49% of the area), regional EPEs (50–79%), and large EPEs (≥ 80% of the study area was affected by the EPEs).

The location of the centre of gravity of return period estimates of precipitation during EPEs in the two areas enabled a division of EPEs in perspective of relief into three categories (represented in the schemes in Fig. 1): EPEs affecting mountains MT (> 450 m a.s.l. in OM, and > 400 m a.s.l. on the windward side, and 300 m a.s.l. on the leeward side of the Vosges Mountains, starting from the mountain ridges in both regions), foreland F (west-northwestwards of the ridges), and lee L (covering Podkrušnohorské pánve basins in OM and the Upper Rhine river Plain in VG). The fixed elevation limits for EPEs affecting MT in OM cannot be similar as in VG since the mean altitude of OM is greater than that of VG, where the mean altitude is lowered by low situated Upper Rhine Plain. Nevertheless, based on the obvious elevation characteristics of the individual study area in DEM (Fig. 1), the selected elevation limits are considered convenient in both areas because the delimitation accurately captures the mountain ranges and separates them from their surroundings. An extra (fourth) category called “total” T was added to the three categories of the relief to ensure the case when very long return period estimates were scattered in MT, L, and F without any specific predominance. In this case, the calculated coordinates of the centre of gravity would not be meaningful.

Geographical location (latitude, longitude) of the centre of gravity of return period estimates of precipitation during EPEs allows for a categorization of EPEs with respect to cardinal points as follows: EPEs affecting southern part of the study area S and northern part N in VG, and western part W and eastern part E in OM. OM and VG were divided into two parts based on the mean perpendicular line to the main mountain ridge as displayed in Fig. 1a and Fig. 1b, respectively. The division was motivated considering the prevailing direction of airflow (Section 2.1) and ensured that the division differed from the MT/L/F/T division described above. A third category C was used for the case when the longest return period estimates were scattered covering the whole study area, similar to the category T for relief.

Extremity and synoptic conditions of EPEs

The categorization of EPEs according to their extremity was based on the WEI values. Since the WEI values vary non-linearly with the size of the study area, as demonstrated by Schiller (2016), and the OM and VG differ in size, the WEI values from one study area have to be transformed to be comparable with those of the second area. The conversion is possible through computation of maximum theoretical WEI value in the two regions, i.e. 1000 years is the return period estimate of precipitation in all grid points and the area affected is equal to the size of the study area. In our case, the WEI values from OM remained the same, while the WEI values from VG were converted as if the VG area was of the same size as that of the OM. The converted (i.e. comparable) WEI values corresponded to the multiplication of the previous WEI values in VG by the ratio of the maximum theoretical WEI value in OM to that in VG. This study thus provides the first example on real data of how to compare the WEI values between various regions. Based on the extremity (converted WEI values), the EPEs were arbitrarily classified into four categories: E1 (WEI < 35), E2 (WEI from 35 to 49), E3 (50–99), and E4 (WEI ≥ 100).

Synoptic variables (Section 2.3) enabled the categorization of EPEs into those with airflow/flux of specific humidity from Southeast SE, Southwest SW, Northwest NW, and Northeast NE at 500 and 850 hPa isobaric levels. Detailed visual inspection of the synoptic maps (pressure and temperature fields) at a 6-h temporal resolution over Europe enabled to categorize the prevailing large-scale synoptic situation during EPEs into four categories per study area. The study region situated under low pressure with closed isobars (cyclone) was classified as “low”, whereas the region under low pressure with unclosed isobars was classified as “trough”. The low was further investigated in order to assess the origin of the low (e.g., cut-off low) and its track based on the classification of tracks of cyclones over Europe proposed by van Bebber (1891). “Zonal” category was assigned to the airflow over the study region which was parallel to the line of latitude (in OM and VG from the West), while that parallel to the longitudinal line was categorized as “meridional” airflow (in OM from the North). In VG, since no purely meridional airflow was observed during the EPEs, the airflow similar to the meridional airflow but Northwesterly was classified as “NW”. The categorization was personally discussed with and was approved by the specialist on the synoptic situations over Europe, Dr. Hoy, author of, e.g. Hoy et al. (2012a, b).

Results and discussion

Comparison of the ten strongest EPEs in OM and VG

The ten strongest EPEs (highest WEI) from both the OM and the VG are presented in Table 1 and Table 2, respectively. None of the ten strongest EPEs from OM overlaps any of the ten strongest in VG. Comparison of the WEI values from OM with the WEI values converted from VG shows that four of the five overall strongest EPEs (i.e. highest WEI values) occurred in VG, which suggests that the EPEs are stronger (have higher extremity) in VG as compared to that in OM. In OM, 60% of the ten strongest EPEs were long and seven of the ten EPEs occurred in SHY (Table 1), whereas in VG, 80% of the ten strongest EPEs were short and five of the ten EPEs occurred in WHY (Table 2). No local EPE was identified among the ten strongest in both OM and VG, although comparatively smaller area was affected by EPEs in VG than in OM.

Table 1 Ten strongest EPEs from the OM arranged in the decreasing order of their extremity (WEI). The first column corresponds to the starting day of EPEs. “CardP” stands for the categorization of EPEs based on cardinal points, “x” for no available data, and FQUV for the flux of specific humidity. The categorized variables are described in Section 2.6. Winter half-year (October–March) EPEs are depicted in italics and long EPEs (3–10 days) are displayed in bold
Full size table
Table 2 As Table 1, but from VG; WEI values were converted to be comparable with those from OM (Section 2.6.3)
Full size table

Table 1 and Table 2 also show that the majority of the strongest EPEs were associated with a low and a trough in OM and VG, respectively. Lows were associated with nine of the ten strongest EPEs in OM and two of the ten strongest EPEs in VG, including the EPE (starting from May 23, 1983) that affected the largest area of VG.

It is worth noticing that since three EPEs in OM and two EPEs in VG occurred during 2011–2013, they were not included in the analysis of synoptic variables that were available until 2010. Nevertheless, it was less than 6 and 4% of events in the dataset of EPEs in OM and VG, respectively, which according to Zolina et al. (2013) does not influence the accuracy of the results.

Dependent characteristics of EPEs in OM and VG

Table 3 and Table 4 summarize the significantly dependent characteristics of EPEs at 1% p value in OM and VG, respectively. Note that the flux of specific humidity at 850 hPa level was shown in OM because of the strongest dependence on half-year; other synoptic variables (moisture flux at 500 hPa level and wind direction at both 500 and 850 hPa levels) were also significantly dependent on half-year and resulted in the same (positive/negative) associations. In VG, the associations are depicted for the wind at 500 and 850 hPa levels, although similar associations were found for the flux of specific humidity at 500 and 850 hPa levels, respectively. The comparison between the two datasets of 54 EPEs also revealed that 3 EPEs from VG overlapped those in OM, most significantly the EPE from the end of October 1998 (6th strongest EPE in VG, Table 2) which was related to strong zonal circulation and also affected OM as 40th strongest EPE in the dataset (Minářová et al. 2017a).

Table 3 Chi-squared residuals of the significantly dependent variables at 1% p value in OM. The variables and their categories are described in Section 2.6. FQUV stands for the flux of specific humidity and “Card P” for cardinal points
Full size table
Table 4 As Table 3, but in VG
Full size table

In OM, the characteristics were significantly dependent on the occurrence in half-year and meteorological season, although more dependent on half-year, whereas in VG they were significantly dependent on the occurrence during seasons and less on half-year. The insignificant dependence of characteristics on half-year in VG as compared to that on the season may correspond to the April–September definition of SHY, since significant dependence was observed when the SHY was defined as spanning from March–August. It is related to the differences (shift) in annual course of precipitation between OM and VG with monthly precipitation maximum in VG in December (in mountains) and June (in the Upper Rhine Plain) as compared to July in OM (Minářová et al. 2017c).

Significant dependencies of the EPE characteristics are depicted in OM in Fig. 2a and Fig. 3, and in VG in Fig. 2b. Figure 2a shows that in OM, the short EPEs occurred mostly in SHY and long EPEs in WHY. The NW flux of specific humidity at 850 hPa isobaric level prevailed during WHY and long EPEs, while the other directions of the flux were related to SHY and short EPEs. The SHY EPEs were mostly of smaller spatial extent (district to local), though some large EPEs (including the strongest) were also identified in SHY. Low (and meridional circulation) was the dominant synoptic situation related to EPEs in OM in SHY, while in WHY it was the zonal circulation and trough. The strongest EPEs (E3–E4) occurred mainly in the second half of the calendar year (especially in summer months) and the E4 EPEs affected the largest area, up to 100% (Fig. 3). The WHY EPEs affected large to regional area of OM (not less than 50% of OM), and were severe in the mountains or affected the total area. The largest SHY EPEs occurred heavily in foreland or mountains and the least spatially extended EPEs affected mostly the lee of the mountains. The eastern part of OM was the most affected by EPEs in SHY, except the EPEs affecting the foreland that were more associated with western part of the region. The results are in conformity with fragmentary information about heavy rainfall in smaller or broader part of the region (e.g. SMUL 2008; INTERKLIM 2014).

Fig. 2
figure2

Significantly dependent characteristics of EPEs in a OM and b VG. Ten strongest EPEs from OM (Table 1) and VG (Table 2) are numbered starting from the strongest (Nr. 1), except one in OM where it was beyond the available synoptic dataset. Note that the reversed values of the components of flux of specific humidity are displayed in a to match the cardinal points

Full size image
Fig. 3
figure3

Annual course of EPEs and dependent spatial characteristics of EPEs in OM (abbreviations described in Section 2.6)

Full size image

In VG, the seasonal occurrence of EPEs was significantly dependent on other characteristics of EPEs (Fig. 2b); the NW wind direction at 500 hPa occurred during winter EPEs, SW wind direction mostly during autumn or summer EPEs, and spring EPEs were rather related to northern wind direction. The SW wind direction was typical for EPEs associated with trough and affecting mostly foreland or mountains, while the NW wind direction in winter corresponded with EPEs from zonal to NW circulation affecting also foreland and mountains of VG. All the five EPEs associated with low occurred under the SE to eastern airflow, none of them occurred in winter. Long EPEs corresponded with western airflow and affected the foreland or mountains the most. These findings provide new insights into the topic about heavy rainfall in VG, very limitedly dealt in literature except Minářová et al. (2017d, b).

Temporal characteristics of EPEs

The datasets of 54 EPEs showed that in both OM and VG, the EPEs were mostly short, i.e. lasted 1–2 days, which matches the expectations because much longer events are rare over one specific location, and a sequence of similar atmospheric patterns is of limited duration. The EPEs instead of occurring only in the main humid season occurred in all seasons in both OM and VG (Minářová et al. 2017a, d). A strong dependence was found between the duration of EPEs (short/long) and half-year (HY) and/or season of their occurrence. In OM, the Crámer’s V was 0.5 for the dependence HY-duration and season-duration, and positive association was found between long EPEs and EPEs in WHY, and winter and autumn, while negative for long EPEs and EPEs in SHY, and summer and spring, and vice versa for the short EPEs (Table 3). In VG, the duration and season were significantly dependent, with a positive association between the long EPEs and its occurrence in spring and winter, and between the short EPEs and summer and autumn (Table 4). It is in a good agreement with the expectation since the long events occur mostly in WHY and winter season due to the larger circulation patterns over Europe in winter which hence favour a stronger coherence between regions and thus longer duration of events at specific regions at this time of the year (Barry 2008; Oliver 2008). The dependence for the HY and duration in VG was insignificant due to the substantially fewer representatives of the long EPEs (9) as compared to the short EPEs (45) in the dataset of EPEs in VG.

Synoptic situation during the EPEs

Synoptic conditions of EPEs in OM

The analysis of synoptic maps obtained from http://www.wetterzentrale.de/ and https://www.esrl.noaa.gov/psd/cgi-bin/data/getpage.pl showed that in OM the 850-hPa isobaric level is necessary for appropriate identification of the synoptic causes related to the EPEs which correspond to strong fluxes of specific humidity (Fig. 4). It is in good agreement with Müller and Kašpar (2010) and Kašpar and Müller (2014), who found the 850 hPa level to be important in the analysis of synoptic conditions over the region and intense flux of specific humidity at 850 hPa level as a predictor of EPEs in East Bohemia (Czech Republic).

Fig. 4
figure4

Meridional (a) and zonal (b) component of flux of specific humidity (colour scale) and geopotential height (contour) at (left) 500 hPa level and (right) 850 hPa level for the two most frequent synoptic patterns during EPEs (the day with highest E ta ) in OM: a low over Central Europe (August 8, 1978), and b strong zonal circulation (December 12, 1986). More information about the EPEs is given in Table 1

Full size image

The low was the most frequent synoptic situation during the 54 EPEs (one exemplary EPE in Fig. 4a). The low occurred during 61% of 54 EPEs and during nine of the ten strongest EPEs (Table 1), and was often produced as cut-off low (in 29% of cases). The cut-off low was also the synoptic cause of the severe heavy rainfall event in May/June 2013 in Central Europe (Grams et al. 2014) as in Table 1. Besides, the lows frequently moved along the Vb track (35% of cases), i.e. from the Mediterranean area towards the northeast to Poland/Ukraine (van Bebber 1891). It was the case of the widespread precipitation in August 2002 (Table 1) which was related to the low with Vb track (Rudolf and Rapp 2002). A list of track of cyclones (including the Vb) related to heavy precipitation events during 1961–2002 over Central Europe was provided by Hofstätter et al. (2016). The cyclones with Vb track during the EPEs in OM were in good agreement with those listed in their study despite minor shift in the date of occurrence (e.g. August 12 of August 11, 2002 in Table 1) which corresponds to the difference in position of the study region (OM is situated westwards from the region Czech Republic-Slovakia-Poland). Nevertheless, all the EPEs in OM related to lows with Vb track could not be checked for their validity with the existing literature, since many of the EPEs in OM were not identified as extreme/heavy at larger spatial scale.

Strong zonal (western) circulation was the second most frequent synoptic pattern related to EPEs, and sixth strongest EPE (Table 1) in OM (16%). The wind and the depicted flux of specific humidity (Fig. 4b) was mainly from northwest (NW), which corresponds with the direction perpendicular to the mountains and thus is particularly prone to the orographic enhancement of precipitation on the western windward side and in the mountains of the region (Pechala and Böhme 1975; INTERKLIM 2014). The trough and strong meridional circulation were the last synoptic situations during the EPEs in OM. Since they were identified during less than 25% of EPEs, they were not detailed and depicted in the paper.

Synoptic conditions of EPEs in VG

As in OM, anomalies in the flux of specific humidity represented the EPEs in VG. Fifty percent of EPEs and half of the ten strongest EPEs occurred when a trough was situated over the region (Table 2). The trough (Fig. 5a), generally related to stationary cold front (Minářová et al. 2017d), produced in most of the cases southwestern airflow and flux of specific humidity to VG, which induced important orographic enhancement of precipitation on the southwestern slopes of the mountains which are higher than those in the North of the area (Fig. 1b). The southwestern airflow direction was related to precipitation totals exceeding 100 mm in Alsace in REKLIP (1995).

Fig. 5
figure5

Meridional (a, c) and zonal (b) component of flux of specific humidity in colour scale and geopotential height in contour at (left) 500 hPa level and (right) 850 hPa level for the three frequent synoptic patterns during EPEs (the day with highest E ta ) in VG: a trough and related southwestern airflow to VG (strongest EPE, November 12, 1996); b strong zonal circulation (October 18, 1998); and c low over Central Europe (May 24, 1983). More information about the EPEs is given in Table 2

Full size image

One third of the EPEs occurred within the strong western zonal circulation (Fig. 5b). The northwestern zonal circulation together with the western zonal circulation influenced VG during 44% of EPEs. Although the 500 hPa level better represents EPEs associated with the two most dominant patterns (i.e. trough and zonal circulation), the 850 hPa level is needed for the identification of the lows, as in OM.

The lows (Fig. 5c), seldom represented during the EPEs in VG, frequently moved along the Vb track and were characterized by inducing extreme precipitation within the eastern (northeast–southeast) direction of airflow and flux of specific humidity to VG. The fifth strongest EPE in VG (Table 2) occurred under cyclone with Vb track, whose track was strongly deviated westwards. The validity of cyclones with Vb track affecting VG is supported by the findings of Paul and Roussel (1985) who stated that the heavy rainfall event on May 22–26, 1983 in Alsace and Lorraine occurred due to reversal airflow from east of air masses originated from Mediterranean area, i.e. the Vb track of cyclone. The lows over the Bay of Biscay were more typical for southwestern airflow in the region and led to the enhanced precipitation totals due to significant orographic lifting.

Dependence of synoptic conditions on other EPE characteristics in OM and VG

The synoptic situation was significantly dependent on HY and season in both OM and VG (Crámer’s V from 0.4 to 0.6). The chi-squared residuals showed a positive association of lows and meridional circulation in SHY, and of zonal circulation and troughs in WHY in OM (Table 3). It is in good agreement with the literature, when, e.g. the cyclones related to summer heavy rainfall events often induce strong northern (meridional) airflow to the region (Pechala and Böhme 1975; SMUL 2008; INTERKLIM 2014), whereas during winter when the circulation in mid-latitudes is more pronounced and the zonal circulation more frequent (Oliver 2008; Houze 2014), the heavy rainfall is more often associated with zonal circulation. In VG, the positive associations were found between spring EPEs and troughs and lows, summer EPEs and troughs and lows, autumn EPEs and zonal circulation, and winter EPEs and NW and zonal circulation (Table 4). The zonal circulation related to autumn and winter EPEs fulfils the expectations, as in OM. The troughs related to spring and summer EPEs in VG might correspond with an increased potential thermal difference between warm air (near the ground or from southern latitudes) and cold air (aloft or from Arctic) during the seasons (REKLIP 1995; Oliver 2008).

The direction of the flux of specific humidity and wind were significantly dependent on HY and season in both OM and VG (Crámer’s V from 0.3 to 0.4), which correspond with the seasonal circulation patterns in Europe. Figure 6 shows that over the year, most of the EPEs in VG occurred within western airflow at 500 hPa level (Fig. 6a) and southwestern airflow at 850 hPa level (Fig. 6b), which agrees with general circulation over the area found in REKLIP (1995), while in OM, the EPEs occurred mostly within northeastern to southern airflow at 500 hPa level (Fig. 6a) and northern airflow at 850 hPa level (Fig. 6b). The northern airflow corresponds with the usual position of the lows (mostly over Poland) responsible for almost two thirds of the EPEs in the region. The strongest EPEs in VG were mostly related to southwestern airflow, whereas in OM to northeastern and northwestern airflow at 500 and 850 hPa level, respectively.

Fig. 6
figure6

Zonal and meridional airflow components during EPEs in OM and VG a at 500 hPa and b at 850 hPa isobaric levels; the reversed values of the components are displayed to match the cardinal points

Full size image

Contrary to VG, the duration of EPEs was significantly dependent on all synoptic variables of EPEs in OM (Crámer’s V 0.3–0.4). The long EPEs were positively associated with northwestern and in some cases with southwestern airflow, while the short EPEs were positively dependent on the northeastern and southeastern direction of airflow (not depicted). The dependence is robust since the northwestern wind direction is typical for winter events in OM when also long EPEs were found. The short EPEs occurring in summer were often related to eastern wind direction following the expectations.

Spatial characteristics and extremity of EPEs

The area affected by the 54 EPEs in VG was comparatively smaller to that in OM; no large EPEs could be identified in VG, while in OM these were the strongest (Table 1). The affected area in OM was significantly dependent on HY (Crámer’s V 0.4) with positive association of WHY EPEs having regional to large affected area, and positive association of SHY EPEs with local to district area affected (Table 3). In SHY, it is due to less frequent stationary cold fronts that might affect large areas as compared to WHY (Houze 2014). In VG, the area affected by EPEs was significantly dependent on the duration of EPEs (Crámer’s V 0.4) with positive association between the long EPEs and regional (i.e. largest in VG) spatial extent (Table 4). It agrees with the expectations, since longer events have a higher potential to affect larger areas as the systems move. It also suggests that in VG, the actual precipitation fields are rather smaller as compared to those in OM, although they might be more unstable.

The extremity of EPEs in VG showed a significant dependence on the wind direction at 500 hPa level (Crámer’s V 0.3) with the strongest E4 EPEs positively associated with NE wind direction (Table 4). The expected significant dependence of extremity on the size of the area affected by EPEs (from the definition of WEI) was found only in OM (Crámer’s V 0.3) and not in VG, which suggests that the WEI does not need to substantially favour EPEs affecting larger areas. Stronger events (E3, E4) in OM significantly tended to affect large areas, i.e. ≥ 80% of the study area, which might be due to the most frequent association of EPEs in OM with stationary lows (western sector) inducing longer lasting precipitation that can affect a larger area. Contrary to OM, in VG the extremity of EPEs may increase with the duration rather than with the area affected by EPEs.

The characteristic relief was significantly dependent on the size of the affected area, extremity, and cardinal points of EPEs in OM (Crámer’s V 0.4, 0.3, and 0.6, respectively). The EPEs that affected the mountains the most were of district to regional extent, and positively associated with E1 EPEs (i.e. least strong). The EPEs that affected the leeward side were positively associated with the expected local EPEs and E3 to E4 EPEs with total area T (Table 3). In VG (Table 4), the relief and season were significantly dependent (Crámer’s V 0.3)—the EPEs affecting the leeward side of the Vosges Mountains were positively associated with summer, which is in conformity with mixed patterns and leeward convection in summer in the region (Sell 1998; Labbouz et al. 2013). The winter EPEs were positively associated with those affecting the mountains, which fits in stronger orographic enhancement of precipitation in winter (Barry 2008).

The characteristic relief was also significantly dependent on the characteristic cardinal points in OM and VG (Crámer’s V 0.6 and 0.8, respectively). As expected, the EPEs affecting the W part of OM were positively associated with foreland and those affecting E with the mountains, despite the higher elevation of the Western OM than the Eastern OM (Fig. 1a), and the leeward side (Table 3). However, the Eastern OM were also associated with heavy rainfall due to the cyclones with Vb track (van Bebber 1891), and August 2002 event in particular (Munzar et al. 2011). In VG, the EPEs that affected the southern part were positively associated with those strongest in mountains, whereas the EPEs that affected the northern part were related to those affecting the foreland or the lee (Table 4). It might be related to lower potential orographic effect on precipitation in the northern part of the area due to the lower elevation of mountains in that part as compared to the highest elevated southern part, where the orographic effect can be more efficient (Fig. 1b).

The spatial distribution of the superimposed and averaged return period estimates of SHY and WHY EPEs for OM (10 WHY EPEs out of 54 EPEs) is displayed in Fig. 7 and for VG (24 WHY EPEs out of 54 EPEs) in Fig. 8. The EPEs with the longest return period estimates are not found in mountains where the highest totals are mostly recorded, but often on the windward side (in SHY in OM and in SHY and WHY in VG). In OM, longer return period estimates are typical in SHY (Fig. 7a) in comparison with WHY (Fig. 7b), whereas in VG they are of similar length (up to around 50 years) in SHY (Fig. 8a) and WHY (Fig. 8b). It might be related to the differences in the mean annual course of precipitation between OM and VG with more seasonal differences in the annual course in various parts of VG (Section 2.1). However, Fig. 7 and Fig. 8 show that the EPEs are spatially rather inhomogeneous in OM as compared to the EPEs in VG, where they are more concentrated in specific regions, i.e. northwestern windward and northeastern lee side in SHY and northern and southwestern windward side in WHY. In WHY, the spatial distribution in VG might be related to the extratropical cyclonic zone shifted southwards during winter (Oliver 2008), and the troughs in southwest-northeast direction influencing mainly the southwestern part of the region, where the orographic enhancement of precipitation plays a crucial role in producing EPEs of high return period levels.

Fig. 7
figure7

Superimposed and averaged return period estimates during EPEs in a SHY and b WHY in OM

Full size image
Fig. 8
figure8

Superimposed and averaged return period estimates during EPEs in a SHY and b WHY in VG

Full size image

Despite rather inhomogeneous spatial distribution of averaged return period levels in OM, it can be observed that in SHY (Fig. 7a) the highest return period estimates affected mostly the area northwards of the main mountain ridge; its central and eastern part in particular. This is in good agreement with literature attributing the record daily precipitation total (i.e. 312 mm on August 12, 2002, at Zinnwald weather station) in the Eastern Ore Mountains (Munzar et al. 2011). In WHY (Fig. 7b), the highest average return period estimates of the EPEs are more concentrated to a north-south oriented belt that is situated in the middle of the study area. The belt comprises also the lee (Czech) side of the mountains since the orographic enhancement of precipitation can take place at any side of the mountains depending on the actual position of the synoptic system (Barry 2008).

Conclusions

Several characteristics (temporal and spatial characteristics, extremity, and synoptic conditions) of EPEs were compared between two low mountain ranges situated in Central Europe, i.e. the OM and the VG. Based on the daily precipitation data from rain gauges during 54 years, the EPEs were defined using WEI, which provided a quantitative assessment of extremity of events, including rarity, and variables duration and spatial extent. Contrary to the previous studies that were mostly based on one study region and generally used different definitions of heavy rainfall, in this study, the EPEs and their characteristics have been defined the same way, thus it provides a robust comparison between the two regions.

Comparative analysis of dependence between 12 pairs of characteristics of EPEs in OM and VG shows that the duration of EPEs and synoptic situation during EPEs are significantly dependent on the seasonal occurrence of EPEs in both OM and VG. The low was related to SHY/summer EPEs, and zonal circulation to WHY/winter EPEs. The NW airflow and moisture flux prevailed during WHY/winter EPEs as well. The long EPEs (3–10 days) were positively associated with WHY/winter and short EPEs with SHY/summer. Short EPEs dominated in both the datasets, occurred in all seasons, and were not confined to only the main precipitation season in both regions. The higher extremity of EPEs was found in VG as compared to OM and the area affected by EPEs in OM was generally greater than that in VG, where no large EPE was identified. The long EPEs tended to affect larger area as compared to that affected by short EPEs in both OM and VG. The spatial distribution of rarity showed that the windward side of the VG is the most affected by EPEs in both SHY and WHY, while in OM it is more heterogeneous with longer return periods in central and Eastern OM in SHY. The most frequent synoptic situation was low in OM and trough related to the stationary cold front in VG. Five EPEs in VG were also related to lows moving along the Vb track, i.e. from the Mediterranean towards the Northeast that strongly deviated westwards from the usual direction.

To the best of our knowledge, the paper provides first objective comparison of a greater dataset of EPEs between two orographic regions and contributes to broadening the understanding of heavy rainfall characteristics in OM and VG which is useful for improving the regional urban planning, mitigating the hazards, and reducing the risks associated with extreme precipitation by, e.g. climate change withstanding engineering decisions. It might motivate for analogous analyses in similar areas in Central Europe which are still not studied in detail, in order to provide a whole and precise picture of EPEs in low mountain ranges in Central Europe. Thus, the future research will be dedicated to further investigation and comparison of the EPE characteristics in similar regions based on WEI, and gain insight into the occurrence of cyclones with Vb track during EPEs in Central Europe.

References

  1. Alexander LV, Zhang X, Peterson TC et al (2006) Global observed changes in daily climate extremes of temperature and precipitation. J Geophys Res Atmospheres 111:D05109. doi:10.1029/2005JD006290

  2. Alsatia (1932) L’Alsace: précis de la géographie régionale des départements Haut-Rhin et Bas-Rhin. Alsatia, Colmar

  3. Awan NK, Formayer H (2016) Cutoff low systems and their relevance to large-scale extreme precipitation in the European Alps. Theor Appl Climatol:1–10. doi:10.1007/s00704-016-1767-0

  4. Barry RG (2008) Mountain weather and climate third edition, 3rd edn. Cambridge University Press, Cambridge

    Book  Google Scholar 

  5. Bartholy J, Pongracz R (2005) Tendencies of extreme climate indices based on daily precipitation in the Carpathian Basin for the 20th century. Idojárás 109:1–20

    Google Scholar 

  6. Bartholy J, Pongrácz R (2007) Regional analysis of extreme temperature and precipitation indices for the Carpathian Basin from 1946 to 2001. Glob Planet Change 57:83–95. doi:10.1016/j.gloplacha.2006.11.002

    Article  Google Scholar 

  7. Baulig H (1950) Les inondations de décembre 1947

  8. Beniston M, Stephenson DB (2004) Extreme climatic events and their evolution under changing climatic conditions. Glob Planet Change 44:1–9. doi:10.1016/j.gloplacha.2004.06.001

    Article  Google Scholar 

  9. Bernhofer C, Surke M (2009) Das Klima in der REGKLAM-Modellregion Dresden. Leibniz-Institut für Ökologische Raumentwicklung (eds). Rhombos-Verl, Berlin

  10. Bosshard T, Kotlarski S, Zappa M, Schär C (2013) Hydrological climate-impact projections for the Rhine River: GCM–RCM uncertainty and separate temperature and precipitation effects. J Hydrometeorol 15:697–713. doi:10.1175/JHM-D-12-098.1

    Article  Google Scholar 

  11. Boucek J (2007) August 2002 catastrophic flood in the Czech Republic. In: Vasiliev OF, VanGelder P, Plate EJ, Bolgov MV (eds) Extreme hydrological events: new concepts for security. Springer, Dordrecht, pp 59–68

    Chapter  Google Scholar 

  12. Brádka J (1963) O srážkovém stínu za Krušnými horami. Meteorol Zprávy 16:26–28

    Google Scholar 

  13. Brázdil R (2002) Meteorologické extrémy a povodně v České republice - přirozený trend nebo následek globálního oteplování?

  14. Brázdil R, Dobrovolný P, Elleder L, et al (2005) Historické a současné povodně v České republice. Masarykova univerzita v Brně, Český hydrometeorologický ústav v Praze

  15. Brazdil R, Kotyza O, Dobrovolny P (2006) July 1432 and August 2002—two millennial floods in Bohemia? Hydrol Sci J-J Sci Hydrol 51:848–863. doi:10.1623/hysj.51.5.848

    Article  Google Scholar 

  16. Chamas V, Kakos V (1988) Mimořádná průtrž mračen a povodeň na Jílovském potoce dne 1. 7. 1987. Sborník Českoslov Geogr Spol 93:265–278

    Google Scholar 

  17. Conradt T, Roers M, Schröter K et al (2013) Comparison of the extreme floods of 2002 and 2013 in the German part of the Elbe River basin and their runoff simulation by SWIM-live. Hydrol Wasserbewirtsch 57:241–245. doi:10.5675/HyWa-2013,5-4

    Google Scholar 

  18. Cramér H (1946) Mathematical methods of statistics. Princeton University Press, Princeton

    Google Scholar 

  19. DWD DDR, HMÚ ČSSR (1975) Podnebí a počasí v Krušných horách. SNTL - Nakladatelství technické literatury, Praha

    Google Scholar 

  20. Ernst F (1988) Panorama de la géographie physique de l’Alsace; et Les régions naturelles de l’Alsace

  21. Fink A, Ulbrich U, Engel H (1996) Aspects of the January 1995 flood in Germany. Weather 51:34–39. doi:10.1002/j.1477-8696.1996.tb06182.x

    Article  Google Scholar 

  22. Foresti L, Pozdnoukhov A (2012) Exploration of alpine orographic precipitation patterns with radar image processing and clustering techniques. Meteorol Appl 19:407–419. doi:10.1002/met.272

    Article  Google Scholar 

  23. Franke J, Goldberg V, Eichelmann U et al (2004) Statistical analysis of regional climate trends in Saxony, Germany. Clim Res 27:145–150. doi:10.3354/cr027145

    Article  Google Scholar 

  24. Gley G (1867) Géographie physique, industrielle, administrative et historique des Vosges, 3rd edn. V.e Gley Impr. V.e & Durand Libraire, Épinal

    Google Scholar 

  25. Grams CM, Binder H, Pfahl S et al (2014) Atmospheric processes triggering the central European floods in June 2013. Nat Hazards Earth Syst Sci 14:1691–1702. doi:10.5194/nhess-14-1691-2014

    Article  Google Scholar 

  26. Greenwood PE, Nikulin MS (1996) A guide to chi-squared testing. Wiley, New York

    Google Scholar 

  27. Hänsel S, Schucknecht A, Böttcher F, et al (2015) Niederschlagsveränderungen in Sachsen von 1901 bis 2100 Starkniederschlags- und Trockenheitstrends. Selbstverlag des Deutchen Wetterdienstes, Offenbach am Main

  28. Heidenreich M, Bernhofer C (eds) (2011) Klimaprojektionen für die REGKLAM-Modellregion Dresden. Rhombos Verl, Berlin

    Google Scholar 

  29. Hirsch F (1972) Bassin représentatif de la Bruche: Intensité des pluies dans le bassin, une méthode d’analyse. Société Météorologique Fr 443–456

  30. Hladný J, Barbořík J (1967) Studie krátkodobých hydrologických předpovědí v povodí Ohře. Sborník HMÚ 1:1–38

    Google Scholar 

  31. Hofstätter M, Chimani B, Lexer A, Blöschl G (2016) A new classification scheme of European cyclone tracks with relevance to precipitation. Water Resour Res n/a-n/a doi: 10.1002/2016WR019146

  32. Hosking JRM, Wallis JR (1997) Regional frequency analysis: an approach based on L-moments. Cambridge University Press

  33. Houze RA (2014) Cloud dynamics. Academic Press

  34. Hoy A, Jaagus J, Sepp M, Matschullat J (2012a) Spatial response of two European atmospheric circulation classifications (data 1901–2010). Theor Appl Climatol 112:73–88. doi:10.1007/s00704-012-0707-x

    Article  Google Scholar 

  35. Hoy A, Sepp M, Matschullat J (2012b) Atmospheric circulation variability in Europe and northern Asia (1901 to 2010). Theor Appl Climatol 113:105–126. doi:10.1007/s00704-012-0770-3

    Article  Google Scholar 

  36. INTERKLIM (2014) Der Klimawandel im böhmisch-sächsischen Grenzraum. Změna klimatu v česko-saském pohraničí. Sächsisches Landesamt für Umwelt, Dresden

    Google Scholar 

  37. Kakos V (1975) Meteorologické příčiny povodní v první polovině prosince 1974

  38. Kakos V (1977) Meteorologické příčiny povodní v oblasti Krušných hor

  39. Kalnay E, Kanamitsu M, Kistler R et al (1996) The NCEP/NCAR 40-year reanalysis project. Bull Am Meteorol Soc 77:437–471. doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2

    Article  Google Scholar 

  40. Kašpar M, Müller M (2014) Combinations of large-scale circulation anomalies conducive to precipitation extremes in the Czech Republic. Atmospheric Res 138:205–212. doi:10.1016/j.atmosres.2013.11.014

    Article  Google Scholar 

  41. Kienzler S, Pech I, Kreibich H et al (2015) After the extreme flood in 2002: changes in preparedness, response and recovery of flood-affected residents in Germany between 2005 and 2011. Nat Hazards Earth Syst Sci 15:505–526. doi:10.5194/nhess-15-505-2015

    Article  Google Scholar 

  42. Küchler W, Sommer W (2005) Klimawandel in Sachsen: Sachstand und Ausblick. Sächsisches Staatsministerium für Umwelt und Landwirtschaft, Dresden

    Google Scholar 

  43. Kynčil J (1983) Povodně v Krušných horách a jejich podhůří v letech 1784-1981: Příspěvek k dějinám čes. hydrologie. Povodí Ohře, podnik pro provoz a využití vodních toků

  44. Kynčil J, Lůžek B (1979) Historické povodně v povodí Bíliny a Ohře. Povodí Ohře

  45. Kyselý J (2009) Trends in heavy precipitation in the Czech Republic over 1961–2005. Int J Climatol 29:1745–1758. doi:10.1002/joc.1784

    Article  Google Scholar 

  46. Kyselý J, Picek J (2007) Regional growth curves and improved design value estimates of extreme precipitation events in the Czech Republic. Clim Res 33:243–255. doi:10.3354/cr033243

    Article  Google Scholar 

  47. Labbouz L, Van Baelen J, Tridon F et al (2013) Precipitation on the lee side of the Vosges Mountains: multi-instrumental study of one case from the COPS campaign. Meteorol Z 22:413–432. doi:10.1127/0941-2948/2013/0413

    Article  Google Scholar 

  48. Maire G (1979) Analyse des fortes pluies de 1h à 48h: Bassin de l’Ill, région Alsace. Ministère de l’agriculture, Université Louis Pasteur, Strasbourg

    Google Scholar 

  49. Merz B, Elmer F, Kunz M et al (2014) The extreme flood in June 2013 in Germany. Houille Blanche:5–10. doi:10.1051/lhb/2014001

  50. Météo-France (2008) Climatologie des Vosges. Météo-France au service des Vosges: le centre dépertemental d’Épinal, Épinal

  51. Minářová J, Müller M, Clappier A, et al (2017b) Duration, rarity, affected area, and weather types associated with extreme precipitation in the Ore Mountains (Erzgebirge) region, Central Europe Press: doi:10.1002/joc.5100

  52. Minářová J, Müller M, Clappier A (2017c) Seasonality of mean and heavy precipitation in the area of the Vosges Mountains: dependence on the selection criterion. Int J Climatol 37:2654–2666. doi:10.1002/joc.4871

    Article  Google Scholar 

  53. Minářová J, Müller M, Clappier A (2017d) Seasonality of mean and heavy precipitation in the area of the Vosges Mountains: dependence on the selection criterion. Int J Climatol 37:2654–2666. doi:10.1002/joc.4871

    Article  Google Scholar 

  54. Minářová J, Müller M, Clappier A, Kašpar M (2017a) Characteristics of extreme precipitation in the Vosges Mountains region (north-eastern France). Presstime doi: 10.1002/joc.5102

  55. Müller M, Kaspar M (2014) Event-adjusted evaluation of weather and climate extremes. Nat Hazards Earth Syst Sci 14:473–483. doi:10.5194/nhess-14-473-2014

    Article  Google Scholar 

  56. Müller M, Kašpar M (2010) Quantitative aspect in circulation type classifications—an example based on evaluation of moisture flux anomalies. Phys Chem Earth Parts ABC 35:484–490. doi:10.1016/j.pce.2009.09.004

    Article  Google Scholar 

  57. Müller M, Kašpar M, Řezáčová D, Sokol Z (2009) Extremeness of meteorological variables as an indicator of extreme precipitation events. Atmospheric Res 92:308–317. doi:10.1016/j.atmosres.2009.01.010

    Article  Google Scholar 

  58. Munzar J, Auer I, Ondráček S (2011) Central European one-day precipitation record. Moravian Geographical Reports 64:107–112

    Google Scholar 

  59. Oliver JE (2008) Encyclopedia of world climatology. Springer Science & Business Media

  60. Pachauri RK, Allen MR, Barros VR et al (2014) Climate change 2014: synthesis report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. IPCC, Geneva

    Google Scholar 

  61. Parlow E (1996) The regional climate project REKLIP—an overview. Theor Appl Climatol 53:3–7. doi:10.1007/BF00866406

    Article  Google Scholar 

  62. Paul P, Roussel I (1985) Les précipitations exceptionnelles d’avril et mai 1983 à l’origine des fortes crues en Alsace et en Lorraine

  63. Pechala F, Böhme W (eds) (1975) Podnebí a počasí v Krušných horách, 1. vyd. SNTL, Praha

  64. Pelt SC van, Beersma JJ, Buishand TA et al (2014) Uncertainty in the future change of extreme precipitation over the Rhine basin: the role of internal climate variability. Clim Dyn 44:1789–1800. doi:10.1007/s00382-014-2312-4

  65. Planche C, Wobrock W, Flossmann AI et al (2013) Small scale topography influence on the formation of three convective systems observed during COPS over the Vosges Mountains. Meteorol Z 22:395–411. doi:10.1127/0941-2948/2013/0402

    Article  Google Scholar 

  66. Prudhomme C, Reed DW (1998) Relationships between extreme daily precipitation and topography in a mountainous region: a case study in Scotland. Int J Climatol 18:1439–1453. doi:10.1002/(SICI)1097-0088(19981115)18:13<1439::AID-JOC320>3.0.CO;2-7

    Article  Google Scholar 

  67. REKLIP (1995) Klimaatlas Oberhein Mitte-Süd: REKLIP. Regio-Klima-Projeckt. Vdf Hochschulverl, Zürich

    Google Scholar 

  68. Roe GH, Montgomery DR, Hallet B (2003) Orographic precipitation and the relief of mountain ranges. J Geophys Res Solid Earth 108:n/a–n/a. doi: 10.1029/2001JB001521

  69. Rudolf B, Rapp J (2002) Das Jahrhunderthochwasser der Elbe: Synoptische Wetterentwicklung und klimatologische Aspekte. DWD Klimastatusbericht 172–187

  70. Schiller J (2016) Eine Sensitivitätsanalyse des Weather Extremity Index (WEI) nach Müller und Kaspar zur Beschreibung extremer Niederschläge unter Verwendung radarbasierter Niederschlagsmessungen des Deutschen Wetterdienstes. University of Cologne

  71. Schröter K, Kunz M, Elmer F et al (2015) What made the June 2013 flood in Germany an exceptional event? A hydro-meteorological evaluation. Hydrol Earth Syst Sci 19:309–327. doi:10.5194/hess-19-309-2015

    Article  Google Scholar 

  72. Sell Y (1998) L’Alsace et les Vosges. Delachaux et Niestlé, Lausanne

    Google Scholar 

  73. Šercl P (2008) Hodnocení metod odhadu plošných srážek (Assessment of methods for area precipitation estimates). Meteorol Zprávy Meteorol Bull 61:33–43

    Google Scholar 

  74. Smith RB (2006) Progress on the theory of orographic precipitation. Spec Pap 398:1–16

    Google Scholar 

  75. SMUL (2008) Sachsen im Klimawandel - Eine Analyse. Sächsisches Staatsministerium für Umwelt und Landwirtschaft, Dresden

    Google Scholar 

  76. Socher M, Boehme-Korn G (2008) Central European floods 2002: lessons learned in Saxony. J Flood Risk Manag 1:123–129. doi:10.1111/j.1753-318X.2008.00014.x

    Article  Google Scholar 

  77. Söder M, Conrad M, Gönner T, Kusch W (2009) Les changements climatiques en Allemagne du Sud: Ampleur – Conséquences – Stratégies. Klimaveränderung und Konsequenzen für die Wasserwirtschaft (KLIWA), Mainz

  78. Solomon S, Quin D, Manning M, et al (2007) Climate change 2007—the physical science basis: Working Group I contribution to the Fourth Assessment Report of the IPCC, IPCC. Cambridge University Press, Cambridge, UK and New York, NY, USA

  79. Stein C, Malitz G (2013) Das Hochwasser an Elbe und Donau im Juni 2013

  80. Štekl J, Brázdil R, Kakos V et al (2001) Extrémní denní srážkové úhrny na území ČR v období 1879–2000 a jejich synoptické příčiny, 1st edn. Národní klimatický program České republiky, Praha

    Google Scholar 

  81. Thieken AH, Kreibich H, Mueller M, Merz B (2007) Coping with floods: preparedness, response and recovery of flood-affected residents in Germany in 2002. Hydrol Sci J-J Sci Hydrol 52:1016–1037. doi:10.1623/hysj.52.5.1016

    Article  Google Scholar 

  82. Thieken AH, Muller M, Kreibich H, Merz B (2005) Flood damage and influencing factors: new insights from the August 2002 flood in Germany. Water Resour Res 41:W12430. doi:10.1029/2005WR004177

    Article  Google Scholar 

  83. Tolasz R, Brázdil R, Bulíř O, et al (2007) Altas podnebí Česka/Climate atlas of Czechia, 1st edn. Český hydrometeorologický ústav, Universita Palackého

  84. Ulbrich U, Brücher T, Fink AH et al (2003) The central European floods of August 2002: part 1—rainfall periods and flood development. Weather 58:371–377. doi:10.1256/wea.61.03A

    Article  Google Scholar 

  85. Uppala SM, KÅllberg PW, Simmons AJ et al (2005) The ERA-40 re-analysis. Q J R Meteorol Soc 131:2961–3012. doi:10.1256/qj.04.176

    Article  Google Scholar 

  86. van Bebber WJ (1891) Die Zugstrassen der barometrischen Minima nach den Bahnenkarten der deutschen Seewarte für den Zeitraum 1875–1890

  87. Van der Schrier, G, van den Besselaar E, Leander R, et al (2013) Central European flooding 2013—Euro4m CIB

  88. van Meijgaard E, Jilderda R (1996) The Meuse flood in January 1995. Weather 51:39–45. doi:10.1002/j.1477-8696.1996.tb06183.x

    Article  Google Scholar 

  89. Vautard R (2013) Des projections climatiques d’une précision inégalée sur toute l’Europe. http://www.insu.cnrs.fr/node/4634. Accessed 10 Feb 2014

  90. Wang XL, Chen H, Wu Y et al (2010) New techniques for the detection and adjustment of shifts in daily precipitation data series. J Appl Meteorol Climatol 49:2416–2436. doi:10.1175/2010JAMC2376.1

    Article  Google Scholar 

  91. Wang XL, Feng Y (2013) RHtests_dlyPrcp user manual. Clim Res Div Atmospheric Sci Technol Dir Sci Technol Branch Environ Can Tor Ont Can Retrieved Febr 25:2014

    Google Scholar 

  92. Werner PC, Gerstengarbe F-W (2010) PIK Report No. 119—Katalog Der Grosswetterlagen Europas nach Paul Hess und Helmut Brezowsky 7., verbesserte und ergänzte Auflage

  93. Whiteman CD (2000) Mountain meteorology: fundamentals and applications. Oxford University Press

  94. World Meteorological Organization (2008) Guide to meteorological instruments and methods of observation. World Meteorological Organization, Geneva

    Google Scholar 

  95. Zolina O (2014) Multidecadal trends in the duration of wet spells and associated intensity of precipitation as revealed by a very dense observational German network. Environ Res Lett 9:025003. doi:10.1088/1748-9326/9/2/025003

    Article  Google Scholar 

  96. Zolina O, Simmer C, Belyaev K et al (2013) Changes in the duration of European wet and dry spells during the last 60 years. J Clim 26:2022–2047. doi:10.1175/JCLI-D-11-00498.1

    Article  Google Scholar 

Download references

Acknowledgements

We thank Météo-France, DWD (Deutscher Wetterdienst), and CHMI (Czech Hydrometeorological Survey) for provided precipitation data, and NCEP/NCAR re-analysed gridded data of synoptic variables. We extend great thanks to the BGF (French Government scholarship) and DBU (Deutsche Bundesstiftung Umwelt), and project CRREAT (reg. number: CZ.02.1.01/0.0/0.0/15_003/0000481) call number 02_15_003 of the Operational Programme Research, Development and Education for financially supporting the research for 15 and 6 months, respectively. We also thank M.Phil. Syed Muntazir Abbas for his valuable remarks during the revision of the manuscript and the language corrections.

Author information

Affiliations

  1. Laboratory Image, City, Environment, National Centre for Scientific Research & University of Strasbourg (3 rue de l’Argonne, F-67000, Strasbourg), Strasbourg, France

    Jana Minářová & Alain Clappier

  2. Department of Physical Geography and Geoecology, Faculty of Science, Charles University in Prague (Albertov 6, 128 43 Praha 2), Prague, Czech Republic

    Jana Minářová & Miloslav Müller

  3. Institute of Atmospheric Physics, Academy of Sciences of the Czech Republic (Boční II 1401, 141 31 Praha 4), Prague, Czech Republic

    Jana Minářová, Miloslav Müller & Marek Kašpar

Authors
  1. Jana Minářová
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Miloslav Müller
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Alain Clappier
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Marek Kašpar
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Jana Minářová.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Minářová, J., Müller, M., Clappier, A. et al. Comparison of extreme precipitation characteristics between the Ore Mountains and the Vosges Mountains (Europe). Theor Appl Climatol 133, 1249–1268 (2018). https://doi.org/10.1007/s00704-017-2247-x

Download citation