Nat. Hazards Earth Syst. Sci., 12, 687–697, 2012 www.nat-hazards-earth-syst-sci.net/12/687/2012/ Natural Hazards doi:10.5194/nhess-12-687-2012 and Earth © Author(s) 2012. CC Attribution 3.0 License. System Sciences

Precipitation extremes in the wettest Mediterranean region (Krivosije)ˇ and associated atmospheric circulation types

V. Ducic´1, J. Lukovic´1, D. Buric´2, G. Stanojevic´3, and S. Mustafic´1 1University of Belgrade, Faculty of Geography, Belgrade, Serbia 2Hydrometeorological Service of , Podgorica, Montenegro 3Geographic Institute “Jovan Cvijic”´ of the Serbian Academy of Sciences and Arts, Belgrade, Serbia

Correspondence to: V. Ducic´ ([email protected]) Received: 5 May 2010 – Revised: 15 August 2011 – Accepted: 12 February 2012 – Published: 22 March 2012

Abstract. The aim of this paper is to analyse indices of eas are sparsely distributed. To set up appropriate land use extreme in Krivosije,ˇ Montenegro, the wettest management strategies, it is important to estimate changes in Mediterranean region, from the period 1951–2007 and their the frequency and the intensity of precipitation extremes over relationships with atmospheric circulation using “SynopVis time. Such estimates are even more important in the wettest Grosswetterlagen” (SVG) series. Data from two stations Mediterranean areas (such as Krivosije,ˇ Montenegro) where were analysed, namely (42◦340N and 18◦390E) and heavy rainfall may have serious implications. (42◦270N and 18◦310E). Four indices of pre- Extreme precipitation events have been analysed for cipitation extremes (SDII, R75p, R95p, R95pTOT) were different Mediterranean countries and weather stations assessed including number of dry days. The results sug- (Gonzalez-Hidalgo et al., 2001; Goodess and Jones, 2002; gest that the number of days with precipitation decreased. Brunet et al., 2007; Durao et al., 2009; Kioutsioukis et al., To analyse the relationship between extreme precipitation 2010; Toreti et al., 2010). Tolika et al. (2007) analysed events and circulation types we have used an efficiency co- extreme precipitation conditions over the Eastern Mediter- efficient (Ec). Regarding relation to atmospheric circulation, ranean and their association with circulation types. westerly, southwesterly and northwesterly circulation types In their analyses of climate variability for the eastern re- with anticyclonic features over Central Europe are more fre- gion of the Mediterranean, Kostopoulou and Jones (2007a, quent for dry days (days with R < 1.0 mm) and northerly, b) found signifcant increases in the anticyclonic circulation easterly and southerly types for wet and very wet days (R75p types for the winter season. They found statistically signif- and R95p indices). The types with cyclonic condition over cant negative trends for many cyclonic types in all seasons, Central Europe show a large proportion of wet and very especially for those types associated with activity of depres- wet days. Also, activity of Genoa cyclogenesis and oro- sion in the central region of the Mediterranean during the graphic influence over a small area are the main reasons for period 1958–2000. They also suggest that anticyclonic cir- the high precipitation amounts recorded in the Krivosijeˇ re- culation types lead to dry conditions, while cyclonic circu- gion (Crkvice). lation brings southern maritime airflows into an area and is thus associated with increases in rainfall. Dunkeloh¨ and Jacobeit (2003) investigated circulation- rainfall relationships for the whole Mediterranean area and 1 Introduction for all seasonally different types of coupled variability. Trends towards drier conditions in large parts of the Mediter- There has been considerable interest in the study of precip- ranean region have also been observed. Sousa et al. (2011) itation and temperature extremes (Klein Tank and Konnen,¨ investigated patterns and extremes of drought indices in the 2003; Zhang, 2000, 2005, 2009; Moberg et al., 2006; Zolina, 20th century in the Mediterranean region and found a clear 2008; Durao et al., 2010; Ceasar et al., 2011) because of their trend towards drier conditions in most western and central great impact on the environment and society. Precipitation Mediterranean regions. Exceptions were northwestern Iberia extremes, in particular, often have major impacts on ero- and most of Turkey; these areas showed an increase in mois- sion that can result in soil degradation. This risk is especially ture availability. present in the Mediterranean region, where agricultural ar-

Published by Copernicus Publications on behalf of the European Geosciences Union. 688 V. Ducic´ et al.: Precipitation extremes in the wettest Mediterranean region

Table 1. List of stations.

Country Station name WMO Period Latitude (N) Longitude (E) Elevation (m) Montenegro Crkvice – 1951–2007 42◦340 18◦390 937 Herceg Novi 13455 1951–2007 42◦270 18◦310 40

Table 2. Definitions and abbreviations of the ETCCDI indices of precipitation extremes used in this study.

Index Definition Units SDII Simple Daily Precipitation Intensity Index (Annual to- mm day−1 tal/Number of days with precipitation ≥1 mm day−1) R75p Number of days with precipitation amount above a site Days specific threshold value for moderate days, calculated as the 75th (R75 %) percentile of the distribution of daily precipitation amounts at days with Rd ≥1 mm in the 1961–1990 baseline period. R95p Number of days with precipitation amount above a site Days specific threshold value for very wet days, calculated as the 95th (R95 %) percentile of the distribution of daily precipitation mounts at days with Rd ≥1 mm in the 1961–1990 baseline period R95pTOT Fraction of annual total precipitation due to events % exceeding the 1961–1990 95th percentile

Definitions including formulas are available from ETCCDI website http://cccma.seos.uvic.ca/ETCCDI/.

In contrast, analyses of six daily rainfall categories for between these indices of precipitation extremes and “Syn- 265 stations in Mediterranean Spain, Italy, Cyprus and Israel opVis Grosswetterlagen” (SVG) series circulation types for for the period 1951–1995 have shown increases in extreme European land areas. daily rainfall in spite of the fact that total rainfall has gen- erally decreased. This pattern is explained by an increase in the frequency and persistence of sub-tropical anticyclones, 2 Data and methodology particularly over the Mediterranean (Alpert et al., 2002). In this paper, daily and annual precipitation time series Unkaseviˇ c´ et al. (2004) studied annual precipitation vari- from the neighbouring meteorological stations Crkvice and ability over Serbia and Montenegro. They examined the de- Herceg Novi were used. Both of the stations are located in pendence of three selected absolute measures of variabil- the southwestern part of Montenegro (Fig. 1). Daily precip- ity of mean annual precipitation (standard deviation, abso- itation data were obtained from the Hydrological and Me- lute mean deviation and mean absolute interannual variabil- teorological Service of Montenegro. The Crkvice meteoro- ity). In addition, two cases of extreme precipitation in Serbia logical station (Table 1) is located in the Krivosijeˇ region on were analysed using the gamma probability density function. the southeastern slope of Orijen Mountain. It presents the Their results show no significant changes for either case of highest average annual precipitation extreme found in Eu- extreme precipitation. rope (http://www.ncdc.noaa.gov/oa/climate/globalextremes. The region of Krivosijeˇ is poorly documented in terms of html). The high annual precipitation is mostly due to oro- precipitation extreme analysis and their relationships with at- graphic uplift (Radovanovic´ et al., 2008). It is located in mospheric circulation types even though it is Medeterranean a rural settlement with 76 inhabitants (census 2002) and in precipitation extreme. This highlights intersts for our study an area of typical Mediterranean , with little surface which is the first one over the investigated area. water and an abundance of groundwater. The average an- nual precipitation during the 1961–1990 baseline period was The aims of this paper are (a) to create a statistical analy- 4593 mm, with the wettest season being winter (1703 mm). sis of indices of precipitation extremes in the wettest place in the Mediterranean region and (b) analysis of the relationship

Nat. Hazards Earth Syst. Sci., 12, 687–697, 2012 www.nat-hazards-earth-syst-sci.net/12/687/2012/ Table 4. Trend analysis of number of dry days for the Crkvice and Herceg Novi station (with statistical significance values) Station Threshold Mann–Kendall test value Crkvice Rd<1mm 3.59 ***

Rd<1mm 2.97 ** Herceg Novi *** Significant at 0.1% level. ** Significant at 1% level.

V. Ducic´ et al.: Precipitation extremes in the wettest Mediterranean region 689 Table 5. Trend analysis of precipitation extremes (with statistical significance values) for the Crkvice and Herceg Novi station SDII 55 Station Index Threshold Mann–Kendall test value 50 Crkvice SDII / 0.24 45 SDII ** 55 R75p Rd>48 mm 2.79 40 + 50

R95p Rd>143 mm 1.70 mm 35 + 45 R95pTOT 143 mm 1.81 30 40 Herceg Novi SDII / 0.49 mm R75p Rd>23.4 mm 1.94 + 25 35 R95p Rd>59.9 mm 0.32 20 30 R95pTOT 59.9 mm 0.54 195125 1956 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006 20 ** Significant at 1% level. Fig. 3. Average daily precipitation intensity1951 1956 1961per day 1966 1971with 1976 precipitation 1981 1986 1991 (SDII) 1996 2001 for 2006 the Crkvice station for the period + Significant at 10% level. 19512007. Dashed line indicatesFig. 3. AverageSen’s estimator daily precipitation of the slope. intensity per day with precip- Fig. 3. Average daily itationprecipitation (SDII) intensity for the per Crkvice day with station precipitation for the (SDII) period for 1951–2007. the Crkvice station for the period It is also important to determine to what extent annual19512007. precipitation, Dashed line Dashed decreases indicates line Sen’s indicates in theestimator Sen’s of estimator the slope. of the slope. number of precipitation days and decreases in daily precipitation intensity could be explained by R75p, R95p 60 the changes in the number of wet and very wet days. Therefore, R75p and R95p were analysed. R75p, R95p 60 There is statistically significant decrease in R75p (1% level of significance)50 and R95p (10% level of significance) for Crkvice (Table 5). Herceg Novi also shows significant50 decrease 40 in R75p index (10% levelFig. 1. ofLocation significance). map. There is an decrease trend in R95p index40 but this Fig. 1. Location map change is nonsignificant. 30

Days 30 Days Dry days 20 Daily precipitation time series used for the analysis cover the period20 from 1951 until 300 2007. Although data quality 290 control was provided by the Hydrological 10 and10 Meteorological Service of Montenegro, data280 were additionally quality controlled following the standards 270 0 0 1951 1956 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006 recommended by WMO (2002).260 Metadata for the stations were considered in 1951terms 1956 of 1961 relocation 1966 1971 1976 1981 1986 1991 1996 2001 2006 or maloperation at the station.250 There were missing data for the Crkvice station (8.12%). They Days Fig.4. Number of days with Rd>75th (R75p in black) and Rd>95th (R95p in gray) percentile for the Crkvice station Fig.4. Number of days withFig. Rd>75th 4. Number (R75p ofdays in black) with Rand> Rd75th>95th (R75p (R95p in black) in gray) and percentileR > for the Crkvice station were calculated using the complete240 series from the Herceg Novi station.for the Misprints period 19512007. are Dashed also lined indicates Sen’s estimator of thed slope. 230 for the95th period (R95p 19512007. in gray) percentile Dashed line for the indicates Crkvice Sen station’s estimator for the period of the slope. identified and corrected. 220 1951–2007. Dashed line indicates Sen’s estimator of the slope. R95pTOT 210 In assessing changes in precipitation extremes or analysing climate variability70 time seriesR95pTOT homogenity is a basic requirement.200 Unfortunately, an advanced correction70 method60 for daily 1951 1956 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006 ity60 control50 was provided by the Hydrological and Mete- 40 precipitation time series is not developed yet (Toreti et al, 2010). In this paper50 data were first Fig. 2 . Number of dry days for the Crkvice station for the period 19512007. Dashed line indicatesorological Sen’s30 estimator Service of Montenegro, data were additionally 40 visuallyof the examined slope. and comparedFig. 2. Number with of dry those days for from the Crkvice neig stationhbouring for the period stations.quality The20 controlled data arefollowing also the standards recommended by 1951–2007. Dashed line indicates Sen’s estimator of the slope. 30 10 WMOPercent (2002). Metadata for the stations were considered in carefully evaluated applying a Multiple Analysis of Series for Homogenization20 (MASH)0 method. terms10 10 of relocation or mal-operation at the station. There The MASH method was developed in the Hungarian Meteorological ServicewerePercent (Szentimrey,0 missing20 data for the1994, Crkvice station (8.12 %); they were 1999, 2003); it is a relative Thehomogeneity second station test used that in this does paper not is Herceg presume Novi (Ta- that thecalculated 10reference30 using series the complete are series from the Herceg Novi 20 40 ble 1), which is located in Boka Kotorska bay in the lower station. Misprints1951 1956 1961 are 1966 also 1971 identified 1976 1981 and 1986 corrected. 1991 1996 2001 2006 homogeneous (Costa and Soares, 2009). In this study a version MASHv3.0230 extended for part of the Krivosijeˇ region and at the base of the south- In assessing changes in precipitation extremes or homogenization of daily precipitation data, has been applied.Fig. 5. Fraction Portions of precipitation of40 time amount series due to very with wet days (R95pTOT) for the Crkvice station for the period ern slopes of Orijen Mountain. It is situated in a small analysing1951 1956 climate 1961 variability, 1966 1971 1976 time 1981 series 1986 1991 homogenity 1996 2001 is 2006 a 19512007. Dashed line indicates Sen’s estimator of the slope. inhomogeneities were excludedtourist resortfrom withthe 3754analysis. people (census 2002). The aver- basic requirement. Unfortunately, an advanced correction Expert team on Climateage annual Change precipitation Detection during theand 1961–1990 IndiceFig. 5. Fractions baseline (ETCCDI)Kostopoulou of pe- precipitation whichmethod and amount foris Jones supported daily due (2005) to precipitation very noticedwet by d ays time contrasting(R95pTOT) series is notfor trends the developed Crkvice of heavy station rainfall for the events period riod was 1922 mm and the maximum was19512007. duringbetween the Dashed winter an increaseline indicatesyet in (Toreti the Sen’s central et estimator al., Mediterranean 2010). of the In slope. this (Italy) paper and data a decrease were first over vi- the Balkans. World Meteorological Organization (WMO) Commision for Climatology, the Joint Commision (662 mm). These two stations were selected because CrkviceThe index suallyR95pTOT examined is a measure and compared of very with extreme those fromprecipitation neighbour- events. It should show for Oceanography and Marineis the Meteorology wettest place in the(JCOMM) Mediterranean and and thewhether HercegKostopoulou Research Novithe obtained Programme anding stations.results Jones are The (2005)on in dataagreementClimate are noticed also with carefully contrasting the amount evaluated of tr endsprecipitation applying of heavy during rainfall wet days. events Variability and Predictabilityhas a(CLIVAR) complete data developed series. There islist another betweenof temperature rain gaugean increase sta- and ina precipitationthe Multiple central Analysis Mediterranean indices. of Series for (I Homogenizationtaly) and a decrease (MASH) over the Balkans. In this paper we selected tiona set at Vrbanjaof four located indices in the of Krivo precipitationsijeˇ region; however,The extremes index we R95pTOT (Tablemethod. 2).is The a The MASHmeasure chosen method of very was developedextreme prec in theipitation Hungar- events. It should show precipitation indices describedid notfour include wet thisextremes station because (R75p, more whetherR95p, than 50 SDIIthe % ofobtained and the R95pTOT). ianresults Meteorological are inThe agreement Serviceindices (Szentimrey, with the amount 1994, 1999, of precipitation 2003); during wet days. data were missing. it is a relative homogeneity test that does not presume that are defined in terms of numbersDaily precipitationof days that time exceed series used either for the absolute analysis cover or percentilethe reference thresholds. series are The homogeneous (Costa and Soares, index R95pTOT is the fractionthe period of annual from 1951 precipitati until 2007.on that Although is due data to qual-very wet2009). days. In It this was study chosen a version MASHv3.02 extended for to explore the supposedly amplified response of extreme precipitation events to the change in the total amount of precipitationwww.nat-hazards-earth-syst-sci.net/12/687/2012/ (Klein Tank and Können, 2003). This is particularly Nat. important Hazards Earth in Syst. Sci., 12, 687–697, 2012 690 V. Ducic´ et al.: Precipitation extremes in the wettest Mediterranean region

Table 3. GWT and GWL with definitions (Gerstengarbe, Werner, 2005).

GWT GWL Definition of GWL 01 WA West nticyclonic

West (W) 02 WZ West cyclonic 03 WS Southern West 04 WW West angular 05 SWA Southwest anticyclonic Southwest (SW) 06 SWZ Southwest cyclonic 07 NWA Northwest anticyclonic Northwest (NW) 08 NWZ Northwest cyclonic 09 HM Central European high Cen. Euro. High (HM) 10 BM Central European ridge Cen. Euro. Low (TM) 11 TM Central European low 12 NA North anticyclonic 13 NZ North cyclonic 14 HNA North, Iceland high, anticyclonic North (N) 15 HNZ North, Iceland high, cyclonic 16 HB British Isles high 17 TRM Central European trough 18 NEA Northeast anticyclonic Northeast (NE) 19 NEZ Northeast cyclonic 20 HFA Fennoscandian high anticyclonic 21 HFZ Fennoscandian high cyclonic East (E) 22 HNFA Norwegian Sea/Fennoscandian high anticyclonic 23 HFNZ Norwegian Sea/ Fennoscandian high cyclonic 24 SEA Southeast anticyclonic Southeast (SE) 25 SEZ Southeast cyclonic 26 SA South anticyclonic 27 SZ South cyclonic South (S) 28 TB British Isles low 29 TRW Western Europe trough Ue Transitional days

Table 4. Trend analysis of number of dry days for the Crkvice and Expert Team on Climate Change Detection and In- Herceg Novi station (with statistical significance values) . dices (ETCCDI), which is supported by World Meteorolog- ical Organization (WMO) Commision for Climatology, the Station Threshold Mann–Kendall test value Joint Commision for Oceanography and Marine Meteorol- ogy (JCOMM) and the Research Programme on Climate ∗∗∗ Crkvice Rd < 1mm 3.59 Variability and Predictability (CLIVAR), developed a list of Herceg Novi Rd<1mm 2.97∗∗ temperature and precipitation indices. In this paper we se- lected a set of four indices of precipitation extremes (Ta- ∗∗∗Significant at 0.1 % level. ble 2). The chosen precipitation indices describe four wet ∗∗Significant at 1 % level. extremes (R75p, R95p, SDII and R95pTOT). The indices are defined in terms of numbers of days that exceed either abso- lute or percentile thresholds. The index R95pTOT is the frac- homogenization of daily precipitation data has been applied. tion of annual precipitation that is due to very wet days. It Portions of time series with inhomogeneities were excluded was chosen to explore the supposedly amplified response of from the analysis.

Nat. Hazards Earth Syst. Sci., 12, 687–697, 2012 www.nat-hazards-earth-syst-sci.net/12/687/2012/ SDII 55

50

45

40

mm 35

30

25

20 1951 1956 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006

Fig. 3. Average daily precipitation intensity per day with precipitation (SDII) for the Crkvice station for the period 19512007. Dashed line indicates Sen’s estimator of the slope.

R75p, R95p 60

50

40

30 Days 20

10

0 1951 1956 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006

Fig.4. Number of days with Rd>75th (R75p in black) and Rd>95th (R95p in gray) percentile for the Crkvice station for theV. period Duci c´19512007. et al.: Precipitation Dashed line extremesindicates Sen in’s the estimator wettest of Mediterranean the slope. region 691

R95pTOT Table 6. The linear trends for GWLs with the highest Ec for dry 70 days in Crkvice and Herceg Novi. 60 50 40 HM WA SA SWA NWA 30 20 −0.015 0.037 0.085 0.050 0.044 10 Percent 0 10 20 30 Table 7. The linear trends for GWLs with the highest Ec for R75p 40 nad R95p indices in Crkvice and Herceg Novi. 1951 1956 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006 TM TB HFZ HNFZ TRM NEZ SEZ Fig. 5. Fraction of precipitation amount due to very wet days (R95pTOT) for the Crkvice station for the period Fig. 5. Fraction of precipitation amount due to very wet days 19512007. Dashed line indicates Sen’s estimator of the slope. −0.004 0.028 −0.063 −0.105 −0.046 −0.006 −0.039 (R95pTOT) for the Crkvice station for the period 1951–2007.

Dashed line indicates Sen’s estimator of the slope. Kostopoulou and Jones (2005) noticed contrasting trends of heavy rainfall events between an increase in the central Mediterranean (Italy) and a decrease over the Balkans. The index R95pTOTTable 5. Trendis a measure analysis of of precipitation very extreme extremes prec (withipitation statistical events.land It areasshould were show used to explore the relationship between pre- whether the obtained significanceresults are values) in agreement for the Crkvice with and the Herceg amount Novi of station. precipitation cipitationduring wet in Krivodays.sijeˇ region and atmospheric circulation. The SVG system is an objective computational version of Station Index Threshold Mann–Kendall the 29-type Hess and Brezowsky Grosswetterlagen (GWL) test value system of classifying European synoptic regimes (Table 3). The GWLs are defined according to the direction of move- Crkvice SDII −0.24 ∗∗ ment air flow and frontal zones and anticyclonic and cyclonic R75p Rd >48 mm −2.79 + features in relation to Central Europe. According to the di- R95p Rd >143 mm −1.70 R95pTOT 143 mm 1.81+ rection of movement, air flow GWLs are grouped into ten supertypes (Grosswettertypen, GWT) (Table 3). The SVG Herceg Novi SDII / −0.49 + method is based on MSLP, Z500 and Z500-Z1000-Thickness R75p Rd >23.4 mm −1.94 and uses the “20th Century Reanalysis” dataset from 1871 R95p Rd >59.9 mm −0.32 to 2008, extended to the present day with NCEP Reanalysis. R95pTOT 59.9 mm 0.54 The domain for the pattern correlations has been more tightly focussed on the European land areas and the 29 types have ∗∗ Significant at 1 % level. been optimized by separating them out more effectively in + Significant at 10 % level. phase-space. The SVG series have the same meanings and nomenclature as the original types and are also filtered so that the minimum allowed event duration is three days (as in extreme precipitation events to the change in the total amount Hess and Brezowsky). Also, SVG system has “transitional of precipitation (Klein Tank and Konnen,¨ 2003). This is par- days”, denoted “Ue” (for Uebergangstag) – days with poor ticularly important in areas that are at extreme risk of soil correspondence to any of the 29 types. . Number of dry days with precipitation less than −1 To analyse the relationship between extreme precipitation 1 mm day has also been analysed. events and circulation types, we have used an efficiency co- The statistical significance of the calculated trends of efficient (Ec). This coefficient presents ratio between a rel- the indices was tested using a Mann-Kendall test. This ative frequency of a given circulation type in extreme pre- method has been applied because it is more suitable for cipitation events to its mean frequency in the investigated non-parametric distributions. Mann-Kendall test is used for period. If the efficiency coefficient is higher than 1.0, this trend analysis in ETCCDI workshops (Zhang et al., 2005). means that the given circulation type tends to be condutive to Sen’s (1968) slope estimator has been used for trend esti- extreme dry days and wet (R75p and R95p indices) precipi- mation of indices. This is a more robust approach for trend tation events. A linear trend of frequency of circulation types estimation of indices based on daily data. was calculated for GWLs which have the highest value of Ec A new and improved version of Objective Grosswetter- in Crkvice and Herceg Novi. lagen catalogue (James, 2007) is “SynopVis Grosswetterla- Cony et al. (2006) used the effectiveness coefficient in gen” (SVG)1. The SVG series circulation types for European order to determine which Hess-Brezowsky Grosswetterla- gen have occured more/less frequently during extreme cold 1 The SVG catalogue was kindly sent by the author himself events than others for 135 stations distributed around Eu- P. M. James rope. Also in their study, Kysely´ and Huth (2008) use

www.nat-hazards-earth-syst-sci.net/12/687/2012/ Nat. Hazards Earth Syst. Sci., 12, 687–697, 2012 692 V. Ducic´ et al.: Precipitation extremes in the wettest Mediterranean region

Fig. 6. The efficiency coefficient (Ec) of GWLs for dry days in station Crkvice, period 1951–2007.

Fig. 7. The efficiency coefficient (Ec) of GWLs for R75p and R95 in station Crkvice, period 1951–2007.

Nat. Hazards Earth Syst. Sci., 12, 687–697, 2012 www.nat-hazards-earth-syst-sci.net/12/687/2012/ V. Ducic´ et al.: Precipitation extremes in the wettest Mediterranean region 693

Fig. 8. The efficiency coefficient (Ec) of GWLs for dry days, station Herceg Novi, period 1951–2007.

Fig. 9. The efficiency coefficient (Ec) of GWLs for R75p and R95p in station Herceg Novi, period 1951–2007.

www.nat-hazards-earth-syst-sci.net/12/687/2012/ Nat. Hazards Earth Syst. Sci., 12, 687–697, 2012 Italy (the western Mediterranean centre) and the other is over the Aegean Sea and Cyprus (the eastern Mediterranean centre). In the Gulf of Genoa, the frequency of cyclones (especially of intense cyclones) is particularly high in winter (Maheras et al., 2001). The Adriatic Sea also appears to be an area of highfrequency cyclone occurrence (Radinović, 1987). During the winter, anticyclones over eastern and northern Europe (Siberian anticyclone) and Genoa cyclones cause a depression that moves from the Adriatic Sea towards the southeast and east and brings in humid weather conditions. This is the main reason why Crkvice and Herceg Novi have their maximum precipitation rates during the winter. Brzović and Jurčec (1997) found that the majority of Adriatic cyclones do not show significant surfacepressure changes in their centres during their movement along the Adriatic. They suggested that the orographic influence that occurs in such a small area is of great importance for cyclone development and that cyclone development is too rapid to allow for a latent heat release to contribute significantly to cyclone intensification. 694 V. Ducic´ et al.: Precipitation extremes in the wettest Mediterranean region

Fig. 10. Composite mapFig. of 10. seaComposite level map pressure of sea level pressure(hPa) (hPa) and and 500 500 hPa hPa geopotential geopotential fields for the North fields Atlantic for Ocean the and North Europe, Atlantic Ocean and 28 December 2000 (http://www.wetterzentrale.de/). Europe, December 28, 2000 (http://www.wetterzentrale.de/). this coefficient for examining the relationship between Hess- decreases in daily precipitation intensity could be explained Brezowsky circulation types which are condutive to sum- by the changes in the number of wet and very wet days. The cyclonesmer generated heat waves at stations over over most the of Europe. Gulf Planchon of GenoaTherefore, R75pare and important R95p were analysed. for annual precipitation as et al. (2009) calculated precipitation amounts and percent- There is statistically significant decrease in R75p (1 % well as for extremeages for precipitation objective GWL types and identified events. weather Thereforepatterns level of, significance) a case and R95p studies (10% level ofhas significance) been separated and causing heavy winter rainfall in Britanny (France). for Crkvice (Table 5). Herceg Novi also shows significant analysed for the day when the maximum daily precipidecrease intation R75p index was (10 % levelrecorded of significance). in There Crkvice for 1951 is a decrease trend in R95p index but this change is non- 2007. The analysis 3for Results the and maximum discussion singleday precisignificant.pitation amount (recorded on December 28, Kostopoulou and Jones (2005) noticed contrasting trends 2000, at 452.5 mm)3.1 Statistical shows analysis strong of extreme precipitation activity in Genoaof heavy cycl rainfallogenesis. events between an increase On in that the central day there was a the Krivosijeˇ region Mediterranean (Italy) and a decrease over the Balkans. prevailing Grosswetterlagen Cental European Low The(GW index R95pTOTL TM). is a measure This of veryweather extreme precip- pattern shows the There is an incresing trend in number of dry days in Crkvice itation events. It should show whether the obtained results highest value of Ecand for Herceg wet Novi (Table and 4), indicatingvery thatwet the number events. of dry areTh ine agreement HNZ with thetype amount was of precipitation preceded during wet this events. The days has incresed significantly (0.1 % level of significance). days. This index shows an increasing statistically significant analysis of compositeAn increse sealevel at both stations wouldpressure reflect drier climate.(hPa) This andtrend 500 for Crkvice hPa according geopotential to the Mann-Kendall test.fields The (Fig.10) shows may have effects on vegetation and ecosystems (Peterson et intensity of precipitation during very wet days increased at a al., 2001). rate of 1.8 % annualy (Fig. 5). synoptic conditions Aduring simple measure which of precipitation the intensity areas (SDII) of does low pressure and the direction of frontal zones not indicate much change over the 57 yr period. A negative were over the Northand Atlantic nonsignificant trend and was foundWestern at both stations. and Central Europe towards the Mediterranean. Genoa cyclogenesis and frontalIt is also important zones to determine slowly to what extentmoving annual pre- eastwards have caused humid and unstable air cipitation, decreases in the number of precipitation days and mass from southsouthwestern direction bringing heavy precipitation at the slopes of Orijen Mountain. Nat. Hazards Earth Syst. Sci., 12, 687–697, 2012 www.nat-hazards-earth-syst-sci.net/12/687/2012/

V. Ducic´ et al.: Precipitation extremes in the wettest Mediterranean region 695

3.2 Relationships between the indices of precipitation The Mediterranean area is well known as a region of fre- extremes in the Krivosijeˇ region and circulation quent cyclone formation. There are two main centres of types maximum cyclone frequency: one is in the Gulf of Genoa and Southern Italy (the western Mediterranean centre) and The effectiveness coefficient (Ec) of the circulation types the other is over the Aegean Sea and Cyprus (the eastern was calculated in days with R < 1.0 (dry days), wet (R75p) Mediterranean centre). In the Gulf of Genoa, the frequency and very wet (R95p) days. The value of Ec coefficients of cyclones (especially of intense cyclones) is particularly higher than 1.0 show that certain circulation type is more high in winter (Maheras et al., 2001). The Adriatic Sea also frequent in these days, thus the Ec less than 1.0 shows that appears to be an area of high-frequency cyclone occurrence circulation type is less frequent in these events. Figures 6, 7, (Radinovic,´ 1987). 8 and 9 show the obtained Ec using time series for stations During the winter, anticyclones over eastern and north- Crkvice and Herceg Novi in the period 1951–2007. ern Europe (Siberian anticyclone) and Genoa cyclones cause The highest values of Ec for days with R < 1.0 mm in a depression that moves from the Adriatic Sea towards the Crkvice were calculated for types HM, WA, SA, SWA, NWA southeast and east, and brings in humid weather conditions. (Fig. 6). During these types, area of high pressure is over This is the main reason why Crkvice and Herceg Novi have South and Central Europe and depressions are over western their maximum precipitation rates during the winter. parts of North Atlantic, West and North Europe. The frontal Brzovic´ and Jurcecˇ (1997) found that the majority of zones move from west towards the interior of the continent, Adriatic cyclones do not show significant surface-pressure while stable conditions are over Mediterranean. All of these changes in their centres during their movement along the have the anticyclonic features in relation to Central Europe. Adriatic. They suggested that the orographic influence that Also, they belong to westerly, south/southwesterly and north- occurs in such a small area is of great importance for cyclone westerly circulation. Most of these types show positive lin- development and that cyclone development is too rapid to ear trend of frequency in period 1951–2007 (Table 6). The allow for a latent heat release to contribute significantly to number of dry days (R < 1.0 mm) also shows positive trend cyclone intensification. for Crkvice in the investigated period. The cyclones generated over the Gulf of Genoa are im- The GWLs which had occured more frequently during portant for annual precipitation as well as for extreme pre- wet (R75p) and very wet (R95p) precipitation events in sta- cipitation events. Therefore, a case study has been iso- tion Crkvice are TM, HFZ, HNFZ, TB, TRM, NEZ, SEZ lated and analysed for the day when the maximum daily (Fig. 7). These types show the cyclonic features in rela- precipitation was recorded in Crkvice for 1951–2007. The tion to Central Europe and belong to southerly, easterly and analysis for the maximum single-day precipitation amount northerly circulation. For most of them are calculated neg- (recorded on 28 December 2000, at 452.5 mm) shows strong ative linear trends for period 1951–2007 (Table 7). During activity Genoa cyclogenesis. On that day there was a pre- these types there is an active Siberian high over North and vailing Grosswetterlagen Cental European Low (GWL TM). East Europe and Azores high over the Atlantic. At the same This weather pattern shows the highest value of Ec for wet time the lows move from North Atlantic toward Southwest and very wet events. The HNZ type was preceded by and South Europe. There are depressions coming from the this events. The analysis of composite sea-level pressure west over Biscay Bay and Garonne River moving towards (hPa) and 500 hPa geopotential fields (Fig. 10) shows syn- Ligurian Sea. Under these conditions the impact of Icelandic optic conditions, during which the areas of low pressure and depression is presented in the area of Krivosije.ˇ the direction of frontal zones were over the North Atlantic Similar results of Ec are for station Herceg Novi (Figs. 8, and Western and Central Europe towards the Mediterranean. 9). The westerly, southwesterly and northwesterly circula- Genoa cyclogenesis and frontal zones slowly moving east- tion types with the anticyclonic features are favourable for wards have caused humid and unstable air mass from south- dry days and the northerly, easterly and southerly circulation southwestern direction bringing heavy precipitation at the types with cyclonic conditions over Central Europe are more slopes of Orijen Mountain. frequent for days with heavy precipitation events. In most cases, the results for Herceg Novi are in agreement with those obtained for Crkvice. Local factors such as mor- 4 Conclusions phology (in terms of steep slopes over the relatively small area of the Krivosijeˇ region) have a considerable impact on The analysis presented here revealed relationships between precipitation totals. In Krivosijeˇ region the most frequent are indices of precipitation extreme and circulation types in the orographic and frontal type of rainfall. During the summer Krivosijeˇ region during the period of 1951–2007. Number convectional type of rainfall occurs locally. Mediterranean of dry days, including R75p, R95p and R95pTOT indices, area as source of moisture, rising air over the Orijen Moun- showed statistically significant trends for Crkvice, as did the tain and very low dew point are the factors that led Krivosijeˇ number of dry days and R75p index for Herceg Novi. The region to be the wettest Mediterranean area. results for Crkvice suggest that the number of days with www.nat-hazards-earth-syst-sci.net/12/687/2012/ Nat. Hazards Earth Syst. Sci., 12, 687–697, 2012 696 V. Ducic´ et al.: Precipitation extremes in the wettest Mediterranean region precipitation significantly decreased. For Herceg Novi, the temperatures and precipitation in Spain, Contribution to Science, amount of precipitation due to dry days decreased as well. 3, 331–342, 2007. Regarding relation to atmospheric circulation, westerly, Brzovic,´ N. and Jurcec,ˇ V.: Numerical simulation of the Adriatic southwesterly and northwesterly circulation types with an- cyclone development, Geofizika, 14, 29–46, 1997. ticyclonic features are more frequent for dry days, and Caesar, J., Alexander, L. V., Trewin, B., Tse-ring, K., Sorany, L., Vuniyayawa, V., Keosavang, N., Shimana, A., Htay, M. M., Kar- northerly, easterly and southerly types for wet and very wet macharya, J., Jayasinghearachchi, D. A., Sakkamart, J., Soares, days (R75p and R95p indices). Also, the types with cyclonic E., Hung, L. T., Thuong, L. T., Hue, C. T., Dung, N. T. T., Hung, conditions over central Europe show a large proportion of P. V., Cuong, H. D., Cuong, N. M., and Sirabaha S.: Changes R75p and R95p. in temperature and precipitation extremes over the Indo-Pacific Krichak and Alpert (2005) suggested that the precipitation region from 1971 to 2005, Int. J. Climatol., 31, 791–801, 2011. decline over the western part of the Mediterranean over the Cony, M., Hernandez, E., and Del Teso, T.: Influences of synoptic last several decades of the 20th century has been linked to the scale in generation of extremly cold days in Europe, Atmosera, positive trend in the North Atlantic Oscillation (NAO). At the 21, 389–401, 2008. same time, the precipitation decline over the eastern region Costa, A. C. and Soares, A.: Homogenization of Climate Data: Re- of the Mediterranean may be explained by the positive trend view and New Perspectives Using Geostatistics, Math. Geosci., 41, 291–305, doi:10.1007/s11004-008-9203-3, 2009. in the East Atlantic Western Russia (EAWR) pattern. Investi- Gerstengarbe, F. W. and Werner, P.: Katalog der Grosswetterlagen gating the relationship between the NAO index and precipita- Europes (1881–2004), PIK Report nr.100, Postdam-Institut fur¨ tion variability in Serbia during the period 1951–2000, Ducic´ Klimafolgenforschung, 153 pp., 2005. et al. (2006) also noticed the impact of NAO for stations in Ducic,´ V., Milovanovic,´ B., and Lukovic,´ J.: Connection between the southeastern part of the country. ENSO index, NAO index and decadal-scale variability of pre- Kostopoulou and Jones (2007a) found that during the win- cipitation in Serbia, in: Proceedings of the Third International ter season, large scale features situated in Northern Europe Conference- Global changes and regional challenges, Sofia, Bul- affect atmospheric circulation over the eastern part of the garia, 28–29 April 2006, 137–142, 2006. Mediterranean, whereas the Atlantic (Azores) anticyclone Dunkeloh, A. and Jacobeit, J.: Circulation dynamics of Mediter- and the Asian Thermal Low define atmospheric circulation ranean precipitation variability 1948–98, Int. J. Climatol., 23, in the summer. 1843–1866, 2003. Durao,˜ R., Pereira, M. J., Costa, A. C., Corte-Real,ˆ J. M., and Orographic influence over the small investigated area of Soares, A.: Indices of precipitation extremes in Southern Por- Montenegro is of the greatest importance for the high precip- tugal – a geostatistical approach, Nat. Hazards Earth Syst. Sci., itation amounts recorded in the Krivosijeˇ region (Crkvice). 9, 241–250, doi:10.5194/nhess-9-241-2009, 2009. Considering the fact that the Mediterranean is charac- Durao,˜ R. M, Pereira, M. J., Costa, A. C., Delgado, J., del Barriod, terised by very low precipitation totals during the summer, G., and Soares, A.: Spatial–temporal dynamics of precipitation future work should consider the seasonal scale of the indices extremes in southern Portugal: a geostatistical assessment study, of precipitation extremes, especially for the winter season or Int. J. Climatol., 30, 1526–1537, 2010. the period from October to April. In addition, it could be Gonzalez-Hidalgo,´ J.C., De Luis, M., Raventos, J., and Sanchez, J. of interest to investigate the relationships between extreme R.: Spatial distribution of seasonal rainfall trends in a western indices and circulation types in other parts of the Mediter- Mediterranean area, Int. J. Climatol., 21,843–860, 2001. Goodess, C. M. and Jones, P.D.: Links between circulation and ranean region. changes in the characteristics of Iberian rainfall, Int. J. Clima- tol., 22, 1593–1615, 2002. Acknowledgements. This study was supported by the Serbian James, P. M.: An objective classification method for Hess and Bre- Ministry of Education and Science, under Grants III 47007 and III zowsky Grosswetterlagen over Europe, Theor. Appl. Climatol., 43007. 88, 11–42, 2007. Kioutsioukis, I., Melas, D., and Zerefos, C.: Statistical assessment Edited by: F. Guzzetti of changes in climate extremes over Greece (1955–2002), Int. J. Reviewed by: M. Brunet-India and two other anonymous referees Climatol., doi:10.1002/joc.2030, 2010. Klein Tank, A. M. G., and Konnen,¨ G. P.: Trends in In- dices of Daily Temperature and Precipitation Extremes in References Europe, 1946–99, J. Climate, 16, 3665, doi:10.1175/1520- 442(2003)016<3665:TIIODT>2.0.CO;2, 2003. Alpert, P., Ben-Gai, T., Baharad, A., Benjamini, Y., Yeku- Kostopoulou, E. and Jones, P. D.: Comprehensive analysis of tieli, D., Colacino, M., Diodato, L., Ramis, C., Homar, V., the climate variability in the eastern Mediterranean. Part I: Romero, R., Michaelides, S., and Manes, A.: The paradox- map-pattern classification. Int. J. Climatol., 27, 1189–1214, ical increase of Mediterranean extreme daily rainfall in spite doi:10.1002/joc.1467, 2007a. of decrease in total values, Geophys. Res. Lett., 29, 1536, Kostopoulou, E. and Jones, P. D.: Comprehensive analysis doi:10.1029/2001GL013554, 2002. of the climate variability in the eastern Mediterranean. Part Brunet, M., Sigro,´ J., Jones, P. D., Saladie,´ O., Aguilar, E., Moberg, II: relationships between atmospheric circulation patterns and A., Lister, D., and Walther, A.: Long-term changes in extreme

Nat. Hazards Earth Syst. Sci., 12, 687–697, 2012 www.nat-hazards-earth-syst-sci.net/12/687/2012/ V. Ducic´ et al.: Precipitation extremes in the wettest Mediterranean region 697

surface climatic elements, Int. J. Climatol., 27, 1351–1371, Szentimrey, T.: Statistical problems connected with the homoge- doi:10.1002/joc.1466, 2007b. nization of climatic time series, edited by: Heino, R., Climate Krichak, O. S. and Alpert, P.: Signatures of the NAO in variations in Europe, Proceedings of the European workshop the atmospheric circulation during wet winter months over held in Kirkkonummi (Majvik), Finland, May 1994, Publications the Mediterranean region, Theo. Appl. Climatol., 000, 1–13, of the Academy of Finland, 330–339, 1994. doi:10.1007/s00704-004-0119-7, 2005. Szentimrey, T.: Multiple analysis of series for homogenization Kysely,´ J. and Huth, R.: Relationships of surface air temperatures (MASH), in: Proceedings of the second seminar for homogeniza- anomalies over Europe to persistence of atmospheric circulation tion of surface climatological data, Budapest, Hungary, WMO- patterns conductive to heat waves, Adv. Geosci., 14, 243–249, TD No. 962, WCDMP No. 41, 27–46, 1999. 2008 Szentimrey, T.: Multiple analysis of series for homogenization Maheras, P., Flocas, H. A., Patrikas, I., and Anagnostopoulou, C.: A (MASH); Verification procedure for homogenized time series, 40 year objective climatology of surface cyclones in the Mediter- in: Fourth seminar for homogenization and quality control in cli- ranean region: Spatial and temporal distribution, Int. J. Climatol., matological databases. Budapest, Hungary, WMO-TD No. 1236, 21, 109–130, 2001. WCDMP No. 56, 193–201, 2003. Moberg, A., and Jones, P. D.: Trends in indices for extremes in Unkaseviˇ c,´ M., Tosiˇ c,´ I., and Vujovic,´ D.: Variability and probabil- daily temperature and precipitation in Central and Western Eu- ity of annual and extreme precipitation over Serbia and Montene- rope, 1901–99, Int. J. Climatol., 25, 1149–1171, 2005. gro, Theor. Appl. Climatol., 79, 103–109, doi:10.1007/s00704- Moberg, A., Jones, P. D., Lister, D., Walther, A., Brunet, M., Ja- 004-0060-9, 2004. cobeit, J., Alexander, L. V., Della-Marta, P. M., Luterbacher, Tolika, K., Anagnostopoulou, C., Maheras, P., and Kutiel H.: Ex- J., Yiou, P., Chen, D., Klein Tank, A. M. G., Saladie,´ O., treme precipitation related to circulation types for four case stud- Sigro,´ J., Aguilar, E., Alexandersson, H., Almarza, C., Auer, ies over the Eastern Mediterranean, Adv. Geosci., 12, 87–93, I., Barriendos, M., Begert, M., Bergstrom, H., Bohm, R., But- 2007, ler, C. J., Caesar, J., Drebs, A., Founda, D., Gestengarbe, F. http://www.adv-geosci.net/12/87/2007/. W., Micela, G., Maugeri, M., Osterle, H., Pandzic, K., Petrakis, Toreti, A., Xoplaki, E., Maraun, D., Kuglitsch, F. G., Wanner, H., M., Srnec, L., Tolasz, R., Tuomenvirta, H., Werner, P. C., Lin- and Luterbacher, J.: Characterisation of extreme winter precip- derholm, H., Philipp, A., Wanner, H., and Xoplaki, E.: Indices itation in Mediterranean coastal sites and associated anomalous for daily temperature and precipitation extremes in Europe anal- atmospheric circulation patterns, Nat. Hazards Earth Syst. Sci., ysed for the period 1901–2000, J. Geophys. Res., 111, D22106, 10, 1037-1050, doi:10.5194/nhess-10-1037-2010, 2010. doi:10.1029/2006JD007103, 2006. Tsonis, A., Swanson, K., and Kravtsov, S.: A new dynamical mech- Osborn, T. J., Hulme, M., Jones, P. D., and Basnett, T. A.: Observed anism for major climate shifts, Geophys. Res. Lett., 34, L13705, trends in the daily intensity of United Kingdom precipitation, Int. doi:10.1029/2007GL030288, 2007. J. Climatol., 20, 347–364, 2000. Zhang, X., Vincent, L. A., Hogg, W. D., and Niitsoo, A.: Temper- Peterson, T. C., Folland, C., Gruza, G., Hogg, W., Mokssit, A., and ature and precipitation trends in Canada during the 20th century, Plummer, N.: Report on the activities of the Working Group on Atmos. Ocean, 38, 395–429, 2000. Climate Change Detection and Related Rapporteurs 1998– 2001. Zhang, X., Aguilar, E., Sensoy, S.,. Melkonyan, H., Tagiyeva, U., World Meteorological Organisation Rep. WCDMP-47, WMO- Ahmed, N., Kutaladze, N., Rahimzadeh, F., Taghipour, A., Han- TD 1071, Geneva, Switzerland, 143 pp., 2001. tosh, T. H, Albert, P., Semawi, M., Karam Ali, M., Halal Said Planchon, O., Quenol,´ H., Dupont, N., and Corgne, S.: Appli- Al-Shabibi, M., Al-Oulan, Z., Zatari, T., Al Dean Khelet, I., cation of the Hess-Brezowsky classification to the identifica- Hamoud, S., Sagir, R., Demircan, M., Eken, M., Adiguzel, M., tion of weather patterns causing heavy winter rainfall in Brit- Alexander, L., Peterson, T. C., and Wallis, T.: Trends in Mid- tany (France), Nat. Hazards Earth Syst. Sci., 9, 1161–1173, dle East climate extreme indices from 1950 to 2003, J. Geophys. doi:10.5194/nhess-9-1161-2009, 2009. Res., 110, D22104, doi:10.1029/2005JD006181, 2005. Radinovic,´ D.: Mediterranean Cyclones and their Influence on the Zhang, X., Zwiers, F. W., and Hegerl, G: The influences of data pre- Weather and Climate, Programme on Short and Medium Range cision on the calculation of temperature percentile indices, Int. J. Weather Prediction Research (PSMP), W.M.O Sofia 24, 1987. Climatol., 29, 321–327, doi:10.1002/joc.1738, 2009. Radovanovic,´ M., Ducic,´ V., and Milovanovic,´ B.: The climate of Zolina, O., Simmer, C., Kapala, A., Bachner, S., Gulev, S., Orjen mountain. Advances in the studies of of the Balkan and Maechel, H.: Seasonally dependent changes of pre- Peninsula, Monograph, No 22, Institute for Nature Conservation cipitation extremes over Germany since 1950 from a very of Serbia, Belgrade, 73–78, 2008. dense observational network, J. Geophys. Res., 113, D06110, Sen, P. K.: Estimates of the regression coefficient based on doi:10.1029/2007JD008393, 2008. Kendall’s tau. J. Amer. Statist. Assoc., 63, 1379-1389, 1968. WMO: Technical Document 1125, World Meteorological Organi- Sousa, P. M., Trigo, R. M., Aizpurua, P., Nieto, R., Gimeno, L., zation, GCOS-76, Geneva, Switzerland, 2002. and Garcia-Herrera, R.: Trends and extremes of drought indices throughout the 20th century in the Mediterranean, Nat. Haz- ards Earth Syst. Sci., 11, 33–51, doi:10.5194/nhess-11-33-2011, 2011.

www.nat-hazards-earth-syst-sci.net/12/687/2012/ Nat. Hazards Earth Syst. Sci., 12, 687–697, 2012