Bull IASH Xe AnneAcirc e No 4 pp 57plusmn 62Strangeways I C and Smith S W (1985) Devel-

opment and use of automatic weather stationsWeather 40 pp 277plusmn 285

Wexler A (1965) Dew-point hygrometry InWexler A (Ed) Humidity and moisture Vol 1Principles and methods of measuring humidity ofgases Chapman and Hall London pp 125plusmn 215

Correspondence to Dr I C Strangeways Terra-Data 7 Cherwell Close Thames St WallingfordOxon OX10 0HF wwwterradatacouke-mail ianstrangewaysntlworldcom Royal Meteorological Society 2003doi 101256wea8103A

The central European floods of August2002Part 2 ndash Synoptic causes and considerations withrespect to climatic change

Uwe Ulbrich Tim Brucher Andreas H Fink Gregor C LeckebuschAndreas Kruger and Joaquim G PintoInstitute for Geophysics and Meteorology University of Cologne Germany

In the first part of this paper (Ulbrich et al2003) we gave a description of the August2002 rainfall events and the resultant floods inparticular of the flood wave of the River ElbeThe extreme precipitation sums observed inthe first half of the month were primarilyassociated with two rainfall episodes The firstepisode occurred on 67 August 2002 Themain rainfall area was situated over LowerAustria the south-western part of the CzechRepublic and south-eastern Germany A severeflash flood was produced in the Lower AustrianWaldviertel ( forest quarterrsquo ) The secondepisode on 11plusmn 13 August 2002 most severelyaffected the Erz Mountains and western partsof the Czech Republic During this second epi-sode 312mm of rain was recorded between0600GMT on 12 August and 0600 GMT on 13August at the Zinnwald weather station in theErz Mountains which is a new 24-hour recordfor Germany The flash floods resulting fromthis rainfall episode and the subsequent Elbeflood produced the most expensive weather-related catastrophe in Europe in recentdecades

In this part of the paper we discuss themeteorological conditions and physical

mechanisms leading to the two main eventsSimilarities to the conditions that led to therecent summer floods of the River Oder in1997 and the River Vistula in 2001 will beshown This will lead us to a consideration oftrends in extreme rainfall over Europe whichare found in numerical simulations of anthro-pogenic climate change

Synoptic situation during 1ndash15 August2002

At the end of July 2002 an upper-tropospherictrough developed west of Ireland It formed acut-off low centred over north-western Franceand Belgium from 1 August A surface lowwhich developed below the upper-troposphericdepression led to an incursion of cool air fromthe Atlantic into western Europe replacing thewarm and humid airmasses previously residentover the Continent Widespread thunderstormactivity and local rainfall extremes in northernGermany were associated with a cold front anda secondary low which moved over Germanytowards the Baltic Sea at the beginning of themonth During the following days the regionnorth of the Alps remained under the influence

Weather Vol 58 November 2003

434

of the cool air and a temperature contrastbetween the areas north and south of thismountain range was established (16 8C inKonstanz Germany and 25 8C in Milan Italyat 1200GMTon 6 August)

On 6 August a weak low formed in the Gulfof Genoa in connection with the frontal systemof the surface low now over The NetherlandsThis new system moved across northern Italyand the Adriatic Sea towards southern Hun-gary (Fig 1) but it remained very shallowWhen rainfall maxima were observed in LowerAustria before midnight on 6 August andduring the afternoon of 7 August the systemrsquoscore was located at some distance from therainfall area south-east of Austria (Fig 1) On7 August surface charts show a convergenceline over central Europe at both 1200GMT

(Fig 2(a)) and 1800GMT (not shown) At thesame time the quasi-stationary depression atupper levels split into two parts located northand south-east of the Alps (Fig 2(b))

Evidence will be given later that this wasimportant in the development of the firstintense rainfall event

On 8 August only weak and very small-scale remnants of the upper-troposphericdepression over central Europe could be identi-fied A new cold air outbreak directed towardswestern Europe commenced on the same dayleading to a new large upper-troposphericdepression with a core over northern Franceon 10 August At the surface this depressionwas associated with a low named Ilse byDeutscher Wetterdienst (the German WeatherService) This cyclone appeared west of Irelandat 0000GMT on 8 August (Fig 1) and movedto southern England by 1200GMT on 9August This surface low remained over south-ern England until 0000GMT on 10 Augustand subsequently filled (not shown) At upperlevels however the cut-off low continued tomove southwards to reach central Italy by1200GMT on 11 August This upper-tropo-spheric development supported the formationof a secondary low (Ilse 2 Fig 1) over the Gulfof Genoa on 10 August while the old core(now renamed Ilse 1) was still present oversouthern England Ilse 2 slowly moved east-wards until 0000 GMT on 11 August subse-quently turning northwards (Fig 1) At thistime the first intense rainfalls of this periodwere observed east of Lake Constance (Bavar-ian Water Board 2002) and later on 12August in the Salzburg area the westernCzech Republic and the Erz Mountains At1200GMT on 12 August the surface analysisproduced by the Deutscher Wetterdienst showsa partly occluded frontal system stretchingnorthwards from the centre of the low alongthe Germanplusmn Polish border (Fig 3(a)) North-erly surface winds of 15plusmn 25 kn between easternGermany and the Austrian Alps were inducedby the strong pressure gradient on the lowrsquoswestern flank The northward movement ofthe surface low was associated with the north-eastward displacement of the upper-tropo-spheric depression mentioned earlier whichhad a position south-east of Germany and theCzech Republic on 12 August (Fig 3(b))Both the surface low (Fig 1) and the upper-airdepression advanced eastwards on thefollowing day

Fig 1 Tracks of the surface lows relevant to the extremerainfall events Positions are given at 12-hourlyintervals partly giving reference to both day (of August2002) and time (GMT) Grey unnamed low movingeastwards from 0000 GMT on 6 August to 1200GMT on7 August Black low named Ilse with an initial positionwest of Ireland at 0000 GMT on 8 August The apparentmovement from a position over southern England at1200 GMT on 9 August to the Alps (dashed line) isconnected to the formation of a secondary low (namedIlse 2) at 0000GMT on 10 August The former core(renamed Ilse 1) was still identified over south-eastEngland at the same time The last position shown forIlse 2 is at 1200 GMTon 13 August

435

Weather Vol 58 November 2003

Physical mechanisms associated withthe two main rainfall events

Our first focus is the second more intense rain-fall event of 12 August The record-breaking

rainfalls were associated with a broad tongue ofextremely warm and humid lower-troposphericair extending from Greece to Denmark Thetemperature and moisture content of thelower-tropospheric air involved in the event

Fig 2 1200 GMT on 7 August 2002 (a) Surface analysis (source European Meteorological Bulletin) (b) 300mbar analysis (units dam) (c) Lower-tropospheric moisture flux (intensity indicated by shading units kgmplusmn 1s plusmn 1 direc-tion given by arrows) computed from the operational ECMWF analysis Isolines indicate areas of more than 50 mm dayplusmn 1

total tropospheric moisture flux convergence (approximately equivalent to rainfall intensity) Maxima of moisture fluxesand flux convergence are labelled Qx and Cx respectively (d) Vertical profile of vertical velocity (solid line units10plusmn 4 mbar splusmn 1 upward directed winds have negative values) and horizontal wind divergence (dashed line units 10plusmn 6splusmn 1)for the area 48plusmn 498N 14plusmn 158E

Weather Vol 58 November 2003

436

were very high The importance of advectionfrom the western Mediterranean Sea in theformation of this tongue is confirmed by ananalysis of back trajectories (Fig 4 inside frontcover) Note that the back trajectories indicateeastward advection over the MediterraneanSea prior to the formation of Ilse 2 It is obviousthat the cyclonersquos movement from the Gulf of

Genoa towards eastern central Europe played asignificant role in the course of the moist airaround the Alps At 1200GMT on 12 August2002 the lowrsquos frontal system (with westwardadvection of warm and humid air) and thenortherly flow west of the cyclone core weretwo essential factors leading to the extremerainfall event in the Erz Mountains area

Fig 3 As Fig 2 but for 1200 GMTon 12 August 2002 Isolines plotted in (c) are 50 70 and 90mm dayplusmn 1 The areaused in (d) is 51plusmn 528N 13plusmn 148E

437

Weather Vol 58 November 2003

Weather radar echoes (not shown) suggest apredominance of easterly flow and thus of front-al lifting during the early morning of 12August while later the strong southwardadvection (and thus orographic lifting towardsthe Erz Mountains) became important Bothmajor flow directions are reflected in the inte-grated low-level moisture fluxes between thesurface and 700 mbar (shown in Fig 3(c))

Observed rain rates and analysed moistureconvergence in the troposphere (indicated bythe isolines in Fig 3(c)) appear to be consist-ent Maximum convergence rates in the analy-sis for 1200 GMT are equivalent to a rainintensity of up to 97mmdayplusmn 1 ie about07mm in 10 minutes Observed rainfall in-tensities for this time period (Fig 4 in Ulbrichet al 2003) are in a range between 3mm(Zinnwald) and about 03mm (Fichtelberg) in10 minutes

The European Centre for Medium-RangeWeather Forecasts (ECMWF) analysis for thesame time shows upward vertical windsthroughout the troposphere with a maximumof 0015mbar splusmn 1 at 700 mbar (Fig 3(d))They are associated with low-level conver-gence and divergence maxima between 500and 250mbar The large divergence near thetropopause is a peculiarity of the profile Thisoutflow could be produced by latent heatrelease below but this would be more typical oftropical convection than a midlatitude depres-sion We thus prefer the view that horizontaldivergence was enhanced by directional diver-gence of the upper-air flow north-west of thecut-off low which is also evident from the 300mbar radiosonde winds encircled in Fig 3(b)Also both local human observations and mapsof sferics (eg Deutscher Wetterdienst 2002)suggest that there was hardly any thunderstormactivity in the main rainfall regions with a fewexceptions (eg Prague 1200GMT on 12August Fig 3(a))

With respect to the first rainfall event on 7August 2002 there was also a mid-tropo-spheric inflow of warm air from the Mediterra-nean Sea but as revealed by the backtrajectories in Fig 5 (inside front cover) thehumid air at lower levels approached central

Europe from eastern to northern originsLower-tropospheric water-vapour flow into themain rainfall area was similar to the situationof 12 August with a prevailing north to north-easterly flux (Fig 2(c)) Thus orographic lift-ing associated with the upward slope of theWaldviertel certainly contributed to theextreme rainfall amounts Moisture flux con-vergence in the troposphere (Fig 2(c)) wassomewhat less than on 12 August (Fig 3(c))It was associated with a line of convergence(Fig 2(b)) in this area Uplift above the 500mbar level was less intense on 7 August (Fig2(d)) than on 12 August (Fig 3(d)) This canbe understood from upper-tropospheric diver-gence which was weaker than during thesecond event even though wind arrows inFig 2(a) also reveal diffluent upper-air flow

In summary we suggest four main synopticcauses for the extreme summer rainfalls ofAugust 2002

(i) Very humid air advected at low- to mid-tropospheric levels from the westernMediterranean basin around the Alps

(ii) Frontal uplift or a convergence linestalling over the affected areas for a sig-nificant length of time (about 24 hours)

(iii) Orographic lifting associated with air-flow up the northern slopes of thecentral European mountain ranges tothe north-west or west of the lowrsquos core

(iv) The occurrence of these conditions isconnected with an upper-level trough ordepression over central Europe whichcontributed to the dynamically inducedupper-level divergence and which alsoprovided the large-scale steering flow forthe surface low Both the upper-leveltrough and the surface low moved onlyslowly or remained quasi-stationary sothat the area remained under atmos-pheric conditions conducive to extremerainfall for some time

Comparison with recent floods in thesame area

Severe summer floods have affected the RiversOder and Vistula in recent years In 1997 boththe Oder and the Vistula swelled to levels that Locations of lightning flashes

Weather Vol 58 November 2003

438

had return periods of the order of 1000 years atsome gauging stations (Kundzewicz et al1999) The floods were produced by asequence of two intense rainfall periods (4plusmn 9and 17plusmn 22 July 1997) affecting southernPoland Austria and eastern parts of the CzechRepublic (Fuchs and Rapp 1998 Figs 2 and3) Cumulative rainfall amounts for July 1997were of the order of 300 to 500 of the averageJuly precipitation At the mountain station ofLysa Hora (located at 49833rsquoN 18827rsquoEBeskides Mountains Czech Republic) 585mmof rain was reported between 4 and 9 July Themonthly sum of 811mm at this station is moretypical of monsoon rainfall The first of therainfall periods can be assigned to the lownamed Xolska by the Deutscher Wetterdienstwhich moved from Italy into central Europe(Fig 6) The second rainfall event was asso-ciated with the low named Zoe which took asomewhat more westerly path from theBalearic Islands towards the Alps forming a

secondary low on 18 July 1997 (Zoe 2) whichslowly progressed into southern Poland(Fig 6) Both surface lows were associatedwith upper-tropospheric cut-off lows oversouth-eastern Europe

The flooding of the Vistula in summer 2001also originated from two extreme rainfallperiods (1617 and 21plusmn 27 July 2001) Againlarge areas in southern Poland and neighbour-ing countries were affected by torrential rainsObserved rainfall amounts and spatial exten-sion of the rainfall area in 2001 were not aslarge as for the 1997 flood but monthly rainfallsums at several stations in southern Polandamounted to about three times as much as theclimatological mean For example 290 mm(337 of average July precipitation) wasreported at Kielce (50849rsquoN 20842rsquoE) for July2001 and 610mm (292) was reported atKasprowy-Wierch (49814rsquoN 19859rsquoE) (Fuchsand Rudolf 2002) The cyclone track asso-ciated with the first rainfall episode (Axel 1Fig 6) came from the western Mediterraneanforming a secondary low on 16 July 2001 (Axel2 Fig 6) The development at the surface tookplace at the eastern flank of a large-scaleupper-tropospheric trough over westernEurope The second rainfall period of the 2001flood was associated with a low that movedinto south-eastern Europe from the Black Seathus differing from the others It was howeveralso connected with a cut-off upper low oversouth-eastern Europe

The synoptic situations of the August 2002floods and those of the two other recent floodsdescribed above are rather similar They weregenerally associated with surface lows whichmoved from the Mediterranean Sea to centralEurope on a path over (or near) the easternAlps The importance of this specific class oftracks for the generation of summer floods incentral Europe has long been recognised (VanBebber 1891) In his classification of typicalcyclone paths over Europe Van Bebber calledthem `Vb-tracksrsquo a notation still used amongGerman meteorologists The Vb-tracks have arather high potential for leading to extremerainfall by contributing to the main synopticcauses identified for the rainfall events of 2002advection of humid air frontal uplift and air-flow up the northern slopes of the mountains

Fig 6 Tracks of the surface lows relevant to the floods of1997 and 2001 Positions are given in 12-hourly inter-vals Dark grey Low Xolska moving north-eastwardsfrom 0000 GMT on 4 July 1997 (Italy) to 0000 GMT on9 July 1997 (Ukraine) Light grey Low Zoe movingnorth-eastwards from the Balearic Islands (0000 GMTon16 July 1997) A secondary low (Zoe 2) forms over thesouthern Czech Republic at 0000 GMT on 18 July 1997Black Low Axel moving from the Iberian peninsula(0000 GMT on 15 July 2001) to a position south of theAlps within 12 hours A secondary core (Axel 2) is gen-erated further north-east at 1200 GMT on 15 July 2001and reaches Gotland at 0000 GMTon 17 July 2001 Thelow of the second rainfall event associated with theVistula flood of 2001 is not included in the figure

439

Weather Vol 58 November 2003

Climatic change simulations

The recent summer floods in central Europeled to a public discussion of whether theseevents were indicative of ongoing anthropo-genic climate change Numerical simulationsperformed using global circulation models(GCMs) provide a basis for estimating climatechange due to increasing greenhouse-gas con-centrations These models generally produce adecrease of summer rainfall in the Mediter-ranean region and Parry (2000 Fig B8) sug-gests that in most models the region withdecreasing summer precipitation reachesnorthwards into central Europe with a reduc-tion of 7plusmn 8 in the 2050s and 11plusmn 12 in the2080s at the respective grid points At firstglance these results seem to suggest a decreas-ing risk of summer floods in a changingclimate Changes in seasonal mean precipi-tation may however be inappropriate forestimating a changing risk of extreme rainfallTherefore we explored the probable change ofdaily rainfall in central Europe using twoGCMs ECHAM4OPYC3 (Bacher et al 1998)and HadCM3 (Carson 1999 Hulme et al1999) The models were programmed toassume economic development without majoremission reductions (`business as usualrsquo)

A comparison of present-day climate withclimate towards the end of the present century(two 30-year periods considered) suggests ashift towards heavy precipitation events in bothmodels (Fig 7) while total summer rainfalldecreases This result is consistent with otherrecent publications (Booij 2002 Christensenand Christensen 2003) The simulatedincrease of days with intense summer rainfall isassociated with an increasing number ofmeridional weather situations comprising atrough over western or central Europe This isthe weather situation also observed during therecent floods

Concluding remarks

The Elbe flood of August 2002 was caused byextreme rainfall in a large area The associatedcirculation pattern was rather typical of pre-vious flood events both in recent years and inthe more distant past (see Van Bebber 1891)

This situation is characterised by a surface lowmoving on a so-called Vb-track from theMediterranean basin into central Europe Withrespect to the summer 2002 events a positiveinterference of several factors is suggestedincluding advection of humidity from theMediterranean basin a quasi-stationary posi-tion of the tropospheric trough which inducesupper-level divergence and orographic liftingdue to northerly flow west of the surface low

Extreme summer rainfall and resultingfloods have always occurred in this region (VanBebber 1891) The model simulations consid-ered suggest that their frequency mightincrease in the future According to Fricke andKaminski (2001) an increase of days withextreme precipitation can already be observedin the long station time-series of the Hohen-peissenberg Observatory in southern Germany(1881plusmn 2001) For summer this increase isassociated with an increasing frequency ofweather types which are in turn associated withVb-tracks (Fricke and Kaminski 2002) Thisseems to agree with the trends produced insimulations of anthropogenic climate changeIt must however be stressed that it is difficultto determine trends of very rare events (Freiand SchaEgrave r 2001) and we suggest that moredetailed studies are performed on both thesimulated and the observed synoptic develop-ment leading to extreme summer precipitationevents in central Europe

Acknowledgements

We wish to thank Deutscher Wetterdienst forpermission to use ECMWF analysis data and theECMWF trajectory package The work was sup-ported by the EU project MICE (Modelling theImpacts of Climate Extremes) under grantEVK2-CT-2001-00118

References

Bacher A Oberhuber J M and Roeckner E(1998) ENSO dynamics and seasonal cycle in thetropical Pacific as simulated by the ECHAM4OPYC3 coupled General Circulation ModelClim Dyn 1414 pp 431plusmn 450

Bavarian Water Board (Bayerisches Landesamt fuEgrave rWasserwirtschaft Hochwassernachrichtendienst)(2002) Hochwasser im August 2002 mit

Weather Vol 58 November 2003

440

Physical mechanisms associated withthe two main rainfall events

Our first focus is the second more intense rain-fall event of 12 August The record-breaking

rainfalls were associated with a broad tongue ofextremely warm and humid lower-troposphericair extending from Greece to Denmark Thetemperature and moisture content of thelower-tropospheric air involved in the event

Fig 2 1200 GMT on 7 August 2002 (a) Surface analysis (source European Meteorological Bulletin) (b) 300mbar analysis (units dam) (c) Lower-tropospheric moisture flux (intensity indicated by shading units kgmplusmn 1s plusmn 1 direc-tion given by arrows) computed from the operational ECMWF analysis Isolines indicate areas of more than 50 mm dayplusmn 1

total tropospheric moisture flux convergence (approximately equivalent to rainfall intensity) Maxima of moisture fluxesand flux convergence are labelled Qx and Cx respectively (d) Vertical profile of vertical velocity (solid line units10plusmn 4 mbar splusmn 1 upward directed winds have negative values) and horizontal wind divergence (dashed line units 10plusmn 6splusmn 1)for the area 48plusmn 498N 14plusmn 158E

Weather Vol 58 November 2003

436

were very high The importance of advectionfrom the western Mediterranean Sea in theformation of this tongue is confirmed by ananalysis of back trajectories (Fig 4 inside frontcover) Note that the back trajectories indicateeastward advection over the MediterraneanSea prior to the formation of Ilse 2 It is obviousthat the cyclonersquos movement from the Gulf of

Genoa towards eastern central Europe played asignificant role in the course of the moist airaround the Alps At 1200GMT on 12 August2002 the lowrsquos frontal system (with westwardadvection of warm and humid air) and thenortherly flow west of the cyclone core weretwo essential factors leading to the extremerainfall event in the Erz Mountains area

Fig 3 As Fig 2 but for 1200 GMTon 12 August 2002 Isolines plotted in (c) are 50 70 and 90mm dayplusmn 1 The areaused in (d) is 51plusmn 528N 13plusmn 148E

437

Weather Vol 58 November 2003

Weather radar echoes (not shown) suggest apredominance of easterly flow and thus of front-al lifting during the early morning of 12August while later the strong southwardadvection (and thus orographic lifting towardsthe Erz Mountains) became important Bothmajor flow directions are reflected in the inte-grated low-level moisture fluxes between thesurface and 700 mbar (shown in Fig 3(c))

Observed rain rates and analysed moistureconvergence in the troposphere (indicated bythe isolines in Fig 3(c)) appear to be consist-ent Maximum convergence rates in the analy-sis for 1200 GMT are equivalent to a rainintensity of up to 97mmdayplusmn 1 ie about07mm in 10 minutes Observed rainfall in-tensities for this time period (Fig 4 in Ulbrichet al 2003) are in a range between 3mm(Zinnwald) and about 03mm (Fichtelberg) in10 minutes

The European Centre for Medium-RangeWeather Forecasts (ECMWF) analysis for thesame time shows upward vertical windsthroughout the troposphere with a maximumof 0015mbar splusmn 1 at 700 mbar (Fig 3(d))They are associated with low-level conver-gence and divergence maxima between 500and 250mbar The large divergence near thetropopause is a peculiarity of the profile Thisoutflow could be produced by latent heatrelease below but this would be more typical oftropical convection than a midlatitude depres-sion We thus prefer the view that horizontaldivergence was enhanced by directional diver-gence of the upper-air flow north-west of thecut-off low which is also evident from the 300mbar radiosonde winds encircled in Fig 3(b)Also both local human observations and mapsof sferics (eg Deutscher Wetterdienst 2002)suggest that there was hardly any thunderstormactivity in the main rainfall regions with a fewexceptions (eg Prague 1200GMT on 12August Fig 3(a))

With respect to the first rainfall event on 7August 2002 there was also a mid-tropo-spheric inflow of warm air from the Mediterra-nean Sea but as revealed by the backtrajectories in Fig 5 (inside front cover) thehumid air at lower levels approached central

Europe from eastern to northern originsLower-tropospheric water-vapour flow into themain rainfall area was similar to the situationof 12 August with a prevailing north to north-easterly flux (Fig 2(c)) Thus orographic lift-ing associated with the upward slope of theWaldviertel certainly contributed to theextreme rainfall amounts Moisture flux con-vergence in the troposphere (Fig 2(c)) wassomewhat less than on 12 August (Fig 3(c))It was associated with a line of convergence(Fig 2(b)) in this area Uplift above the 500mbar level was less intense on 7 August (Fig2(d)) than on 12 August (Fig 3(d)) This canbe understood from upper-tropospheric diver-gence which was weaker than during thesecond event even though wind arrows inFig 2(a) also reveal diffluent upper-air flow

In summary we suggest four main synopticcauses for the extreme summer rainfalls ofAugust 2002

(i) Very humid air advected at low- to mid-tropospheric levels from the westernMediterranean basin around the Alps

(ii) Frontal uplift or a convergence linestalling over the affected areas for a sig-nificant length of time (about 24 hours)

(iii) Orographic lifting associated with air-flow up the northern slopes of thecentral European mountain ranges tothe north-west or west of the lowrsquos core

(iv) The occurrence of these conditions isconnected with an upper-level trough ordepression over central Europe whichcontributed to the dynamically inducedupper-level divergence and which alsoprovided the large-scale steering flow forthe surface low Both the upper-leveltrough and the surface low moved onlyslowly or remained quasi-stationary sothat the area remained under atmos-pheric conditions conducive to extremerainfall for some time

Comparison with recent floods in thesame area

Severe summer floods have affected the RiversOder and Vistula in recent years In 1997 boththe Oder and the Vistula swelled to levels that Locations of lightning flashes

Weather Vol 58 November 2003

438

had return periods of the order of 1000 years atsome gauging stations (Kundzewicz et al1999) The floods were produced by asequence of two intense rainfall periods (4plusmn 9and 17plusmn 22 July 1997) affecting southernPoland Austria and eastern parts of the CzechRepublic (Fuchs and Rapp 1998 Figs 2 and3) Cumulative rainfall amounts for July 1997were of the order of 300 to 500 of the averageJuly precipitation At the mountain station ofLysa Hora (located at 49833rsquoN 18827rsquoEBeskides Mountains Czech Republic) 585mmof rain was reported between 4 and 9 July Themonthly sum of 811mm at this station is moretypical of monsoon rainfall The first of therainfall periods can be assigned to the lownamed Xolska by the Deutscher Wetterdienstwhich moved from Italy into central Europe(Fig 6) The second rainfall event was asso-ciated with the low named Zoe which took asomewhat more westerly path from theBalearic Islands towards the Alps forming a

secondary low on 18 July 1997 (Zoe 2) whichslowly progressed into southern Poland(Fig 6) Both surface lows were associatedwith upper-tropospheric cut-off lows oversouth-eastern Europe

The flooding of the Vistula in summer 2001also originated from two extreme rainfallperiods (1617 and 21plusmn 27 July 2001) Againlarge areas in southern Poland and neighbour-ing countries were affected by torrential rainsObserved rainfall amounts and spatial exten-sion of the rainfall area in 2001 were not aslarge as for the 1997 flood but monthly rainfallsums at several stations in southern Polandamounted to about three times as much as theclimatological mean For example 290 mm(337 of average July precipitation) wasreported at Kielce (50849rsquoN 20842rsquoE) for July2001 and 610mm (292) was reported atKasprowy-Wierch (49814rsquoN 19859rsquoE) (Fuchsand Rudolf 2002) The cyclone track asso-ciated with the first rainfall episode (Axel 1Fig 6) came from the western Mediterraneanforming a secondary low on 16 July 2001 (Axel2 Fig 6) The development at the surface tookplace at the eastern flank of a large-scaleupper-tropospheric trough over westernEurope The second rainfall period of the 2001flood was associated with a low that movedinto south-eastern Europe from the Black Seathus differing from the others It was howeveralso connected with a cut-off upper low oversouth-eastern Europe

The synoptic situations of the August 2002floods and those of the two other recent floodsdescribed above are rather similar They weregenerally associated with surface lows whichmoved from the Mediterranean Sea to centralEurope on a path over (or near) the easternAlps The importance of this specific class oftracks for the generation of summer floods incentral Europe has long been recognised (VanBebber 1891) In his classification of typicalcyclone paths over Europe Van Bebber calledthem `Vb-tracksrsquo a notation still used amongGerman meteorologists The Vb-tracks have arather high potential for leading to extremerainfall by contributing to the main synopticcauses identified for the rainfall events of 2002advection of humid air frontal uplift and air-flow up the northern slopes of the mountains

Fig 6 Tracks of the surface lows relevant to the floods of1997 and 2001 Positions are given in 12-hourly inter-vals Dark grey Low Xolska moving north-eastwardsfrom 0000 GMT on 4 July 1997 (Italy) to 0000 GMT on9 July 1997 (Ukraine) Light grey Low Zoe movingnorth-eastwards from the Balearic Islands (0000 GMTon16 July 1997) A secondary low (Zoe 2) forms over thesouthern Czech Republic at 0000 GMT on 18 July 1997Black Low Axel moving from the Iberian peninsula(0000 GMT on 15 July 2001) to a position south of theAlps within 12 hours A secondary core (Axel 2) is gen-erated further north-east at 1200 GMT on 15 July 2001and reaches Gotland at 0000 GMTon 17 July 2001 Thelow of the second rainfall event associated with theVistula flood of 2001 is not included in the figure

439

Weather Vol 58 November 2003

Climatic change simulations

The recent summer floods in central Europeled to a public discussion of whether theseevents were indicative of ongoing anthropo-genic climate change Numerical simulationsperformed using global circulation models(GCMs) provide a basis for estimating climatechange due to increasing greenhouse-gas con-centrations These models generally produce adecrease of summer rainfall in the Mediter-ranean region and Parry (2000 Fig B8) sug-gests that in most models the region withdecreasing summer precipitation reachesnorthwards into central Europe with a reduc-tion of 7plusmn 8 in the 2050s and 11plusmn 12 in the2080s at the respective grid points At firstglance these results seem to suggest a decreas-ing risk of summer floods in a changingclimate Changes in seasonal mean precipi-tation may however be inappropriate forestimating a changing risk of extreme rainfallTherefore we explored the probable change ofdaily rainfall in central Europe using twoGCMs ECHAM4OPYC3 (Bacher et al 1998)and HadCM3 (Carson 1999 Hulme et al1999) The models were programmed toassume economic development without majoremission reductions (`business as usualrsquo)

A comparison of present-day climate withclimate towards the end of the present century(two 30-year periods considered) suggests ashift towards heavy precipitation events in bothmodels (Fig 7) while total summer rainfalldecreases This result is consistent with otherrecent publications (Booij 2002 Christensenand Christensen 2003) The simulatedincrease of days with intense summer rainfall isassociated with an increasing number ofmeridional weather situations comprising atrough over western or central Europe This isthe weather situation also observed during therecent floods

Concluding remarks

The Elbe flood of August 2002 was caused byextreme rainfall in a large area The associatedcirculation pattern was rather typical of pre-vious flood events both in recent years and inthe more distant past (see Van Bebber 1891)

This situation is characterised by a surface lowmoving on a so-called Vb-track from theMediterranean basin into central Europe Withrespect to the summer 2002 events a positiveinterference of several factors is suggestedincluding advection of humidity from theMediterranean basin a quasi-stationary posi-tion of the tropospheric trough which inducesupper-level divergence and orographic liftingdue to northerly flow west of the surface low

Extreme summer rainfall and resultingfloods have always occurred in this region (VanBebber 1891) The model simulations consid-ered suggest that their frequency mightincrease in the future According to Fricke andKaminski (2001) an increase of days withextreme precipitation can already be observedin the long station time-series of the Hohen-peissenberg Observatory in southern Germany(1881plusmn 2001) For summer this increase isassociated with an increasing frequency ofweather types which are in turn associated withVb-tracks (Fricke and Kaminski 2002) Thisseems to agree with the trends produced insimulations of anthropogenic climate changeIt must however be stressed that it is difficultto determine trends of very rare events (Freiand SchaEgrave r 2001) and we suggest that moredetailed studies are performed on both thesimulated and the observed synoptic develop-ment leading to extreme summer precipitationevents in central Europe

Acknowledgements

We wish to thank Deutscher Wetterdienst forpermission to use ECMWF analysis data and theECMWF trajectory package The work was sup-ported by the EU project MICE (Modelling theImpacts of Climate Extremes) under grantEVK2-CT-2001-00118

References

Bacher A Oberhuber J M and Roeckner E(1998) ENSO dynamics and seasonal cycle in thetropical Pacific as simulated by the ECHAM4OPYC3 coupled General Circulation ModelClim Dyn 1414 pp 431plusmn 450

Bavarian Water Board (Bayerisches Landesamt fuEgrave rWasserwirtschaft Hochwassernachrichtendienst)(2002) Hochwasser im August 2002 mit

Weather Vol 58 November 2003

440

were very high The importance of advectionfrom the western Mediterranean Sea in theformation of this tongue is confirmed by ananalysis of back trajectories (Fig 4 inside frontcover) Note that the back trajectories indicateeastward advection over the MediterraneanSea prior to the formation of Ilse 2 It is obviousthat the cyclonersquos movement from the Gulf of

Genoa towards eastern central Europe played asignificant role in the course of the moist airaround the Alps At 1200GMT on 12 August2002 the lowrsquos frontal system (with westwardadvection of warm and humid air) and thenortherly flow west of the cyclone core weretwo essential factors leading to the extremerainfall event in the Erz Mountains area

Fig 3 As Fig 2 but for 1200 GMTon 12 August 2002 Isolines plotted in (c) are 50 70 and 90mm dayplusmn 1 The areaused in (d) is 51plusmn 528N 13plusmn 148E

437

Weather Vol 58 November 2003

Weather radar echoes (not shown) suggest apredominance of easterly flow and thus of front-al lifting during the early morning of 12August while later the strong southwardadvection (and thus orographic lifting towardsthe Erz Mountains) became important Bothmajor flow directions are reflected in the inte-grated low-level moisture fluxes between thesurface and 700 mbar (shown in Fig 3(c))

Observed rain rates and analysed moistureconvergence in the troposphere (indicated bythe isolines in Fig 3(c)) appear to be consist-ent Maximum convergence rates in the analy-sis for 1200 GMT are equivalent to a rainintensity of up to 97mmdayplusmn 1 ie about07mm in 10 minutes Observed rainfall in-tensities for this time period (Fig 4 in Ulbrichet al 2003) are in a range between 3mm(Zinnwald) and about 03mm (Fichtelberg) in10 minutes

The European Centre for Medium-RangeWeather Forecasts (ECMWF) analysis for thesame time shows upward vertical windsthroughout the troposphere with a maximumof 0015mbar splusmn 1 at 700 mbar (Fig 3(d))They are associated with low-level conver-gence and divergence maxima between 500and 250mbar The large divergence near thetropopause is a peculiarity of the profile Thisoutflow could be produced by latent heatrelease below but this would be more typical oftropical convection than a midlatitude depres-sion We thus prefer the view that horizontaldivergence was enhanced by directional diver-gence of the upper-air flow north-west of thecut-off low which is also evident from the 300mbar radiosonde winds encircled in Fig 3(b)Also both local human observations and mapsof sferics (eg Deutscher Wetterdienst 2002)suggest that there was hardly any thunderstormactivity in the main rainfall regions with a fewexceptions (eg Prague 1200GMT on 12August Fig 3(a))

With respect to the first rainfall event on 7August 2002 there was also a mid-tropo-spheric inflow of warm air from the Mediterra-nean Sea but as revealed by the backtrajectories in Fig 5 (inside front cover) thehumid air at lower levels approached central

Europe from eastern to northern originsLower-tropospheric water-vapour flow into themain rainfall area was similar to the situationof 12 August with a prevailing north to north-easterly flux (Fig 2(c)) Thus orographic lift-ing associated with the upward slope of theWaldviertel certainly contributed to theextreme rainfall amounts Moisture flux con-vergence in the troposphere (Fig 2(c)) wassomewhat less than on 12 August (Fig 3(c))It was associated with a line of convergence(Fig 2(b)) in this area Uplift above the 500mbar level was less intense on 7 August (Fig2(d)) than on 12 August (Fig 3(d)) This canbe understood from upper-tropospheric diver-gence which was weaker than during thesecond event even though wind arrows inFig 2(a) also reveal diffluent upper-air flow

In summary we suggest four main synopticcauses for the extreme summer rainfalls ofAugust 2002

(i) Very humid air advected at low- to mid-tropospheric levels from the westernMediterranean basin around the Alps

(ii) Frontal uplift or a convergence linestalling over the affected areas for a sig-nificant length of time (about 24 hours)

(iii) Orographic lifting associated with air-flow up the northern slopes of thecentral European mountain ranges tothe north-west or west of the lowrsquos core

(iv) The occurrence of these conditions isconnected with an upper-level trough ordepression over central Europe whichcontributed to the dynamically inducedupper-level divergence and which alsoprovided the large-scale steering flow forthe surface low Both the upper-leveltrough and the surface low moved onlyslowly or remained quasi-stationary sothat the area remained under atmos-pheric conditions conducive to extremerainfall for some time

Comparison with recent floods in thesame area

Severe summer floods have affected the RiversOder and Vistula in recent years In 1997 boththe Oder and the Vistula swelled to levels that Locations of lightning flashes

Weather Vol 58 November 2003

438

had return periods of the order of 1000 years atsome gauging stations (Kundzewicz et al1999) The floods were produced by asequence of two intense rainfall periods (4plusmn 9and 17plusmn 22 July 1997) affecting southernPoland Austria and eastern parts of the CzechRepublic (Fuchs and Rapp 1998 Figs 2 and3) Cumulative rainfall amounts for July 1997were of the order of 300 to 500 of the averageJuly precipitation At the mountain station ofLysa Hora (located at 49833rsquoN 18827rsquoEBeskides Mountains Czech Republic) 585mmof rain was reported between 4 and 9 July Themonthly sum of 811mm at this station is moretypical of monsoon rainfall The first of therainfall periods can be assigned to the lownamed Xolska by the Deutscher Wetterdienstwhich moved from Italy into central Europe(Fig 6) The second rainfall event was asso-ciated with the low named Zoe which took asomewhat more westerly path from theBalearic Islands towards the Alps forming a

secondary low on 18 July 1997 (Zoe 2) whichslowly progressed into southern Poland(Fig 6) Both surface lows were associatedwith upper-tropospheric cut-off lows oversouth-eastern Europe

The flooding of the Vistula in summer 2001also originated from two extreme rainfallperiods (1617 and 21plusmn 27 July 2001) Againlarge areas in southern Poland and neighbour-ing countries were affected by torrential rainsObserved rainfall amounts and spatial exten-sion of the rainfall area in 2001 were not aslarge as for the 1997 flood but monthly rainfallsums at several stations in southern Polandamounted to about three times as much as theclimatological mean For example 290 mm(337 of average July precipitation) wasreported at Kielce (50849rsquoN 20842rsquoE) for July2001 and 610mm (292) was reported atKasprowy-Wierch (49814rsquoN 19859rsquoE) (Fuchsand Rudolf 2002) The cyclone track asso-ciated with the first rainfall episode (Axel 1Fig 6) came from the western Mediterraneanforming a secondary low on 16 July 2001 (Axel2 Fig 6) The development at the surface tookplace at the eastern flank of a large-scaleupper-tropospheric trough over westernEurope The second rainfall period of the 2001flood was associated with a low that movedinto south-eastern Europe from the Black Seathus differing from the others It was howeveralso connected with a cut-off upper low oversouth-eastern Europe

The synoptic situations of the August 2002floods and those of the two other recent floodsdescribed above are rather similar They weregenerally associated with surface lows whichmoved from the Mediterranean Sea to centralEurope on a path over (or near) the easternAlps The importance of this specific class oftracks for the generation of summer floods incentral Europe has long been recognised (VanBebber 1891) In his classification of typicalcyclone paths over Europe Van Bebber calledthem `Vb-tracksrsquo a notation still used amongGerman meteorologists The Vb-tracks have arather high potential for leading to extremerainfall by contributing to the main synopticcauses identified for the rainfall events of 2002advection of humid air frontal uplift and air-flow up the northern slopes of the mountains

Fig 6 Tracks of the surface lows relevant to the floods of1997 and 2001 Positions are given in 12-hourly inter-vals Dark grey Low Xolska moving north-eastwardsfrom 0000 GMT on 4 July 1997 (Italy) to 0000 GMT on9 July 1997 (Ukraine) Light grey Low Zoe movingnorth-eastwards from the Balearic Islands (0000 GMTon16 July 1997) A secondary low (Zoe 2) forms over thesouthern Czech Republic at 0000 GMT on 18 July 1997Black Low Axel moving from the Iberian peninsula(0000 GMT on 15 July 2001) to a position south of theAlps within 12 hours A secondary core (Axel 2) is gen-erated further north-east at 1200 GMT on 15 July 2001and reaches Gotland at 0000 GMTon 17 July 2001 Thelow of the second rainfall event associated with theVistula flood of 2001 is not included in the figure

439

Weather Vol 58 November 2003

Climatic change simulations

The recent summer floods in central Europeled to a public discussion of whether theseevents were indicative of ongoing anthropo-genic climate change Numerical simulationsperformed using global circulation models(GCMs) provide a basis for estimating climatechange due to increasing greenhouse-gas con-centrations These models generally produce adecrease of summer rainfall in the Mediter-ranean region and Parry (2000 Fig B8) sug-gests that in most models the region withdecreasing summer precipitation reachesnorthwards into central Europe with a reduc-tion of 7plusmn 8 in the 2050s and 11plusmn 12 in the2080s at the respective grid points At firstglance these results seem to suggest a decreas-ing risk of summer floods in a changingclimate Changes in seasonal mean precipi-tation may however be inappropriate forestimating a changing risk of extreme rainfallTherefore we explored the probable change ofdaily rainfall in central Europe using twoGCMs ECHAM4OPYC3 (Bacher et al 1998)and HadCM3 (Carson 1999 Hulme et al1999) The models were programmed toassume economic development without majoremission reductions (`business as usualrsquo)

A comparison of present-day climate withclimate towards the end of the present century(two 30-year periods considered) suggests ashift towards heavy precipitation events in bothmodels (Fig 7) while total summer rainfalldecreases This result is consistent with otherrecent publications (Booij 2002 Christensenand Christensen 2003) The simulatedincrease of days with intense summer rainfall isassociated with an increasing number ofmeridional weather situations comprising atrough over western or central Europe This isthe weather situation also observed during therecent floods

Concluding remarks

The Elbe flood of August 2002 was caused byextreme rainfall in a large area The associatedcirculation pattern was rather typical of pre-vious flood events both in recent years and inthe more distant past (see Van Bebber 1891)

This situation is characterised by a surface lowmoving on a so-called Vb-track from theMediterranean basin into central Europe Withrespect to the summer 2002 events a positiveinterference of several factors is suggestedincluding advection of humidity from theMediterranean basin a quasi-stationary posi-tion of the tropospheric trough which inducesupper-level divergence and orographic liftingdue to northerly flow west of the surface low

Extreme summer rainfall and resultingfloods have always occurred in this region (VanBebber 1891) The model simulations consid-ered suggest that their frequency mightincrease in the future According to Fricke andKaminski (2001) an increase of days withextreme precipitation can already be observedin the long station time-series of the Hohen-peissenberg Observatory in southern Germany(1881plusmn 2001) For summer this increase isassociated with an increasing frequency ofweather types which are in turn associated withVb-tracks (Fricke and Kaminski 2002) Thisseems to agree with the trends produced insimulations of anthropogenic climate changeIt must however be stressed that it is difficultto determine trends of very rare events (Freiand SchaEgrave r 2001) and we suggest that moredetailed studies are performed on both thesimulated and the observed synoptic develop-ment leading to extreme summer precipitationevents in central Europe

Acknowledgements

We wish to thank Deutscher Wetterdienst forpermission to use ECMWF analysis data and theECMWF trajectory package The work was sup-ported by the EU project MICE (Modelling theImpacts of Climate Extremes) under grantEVK2-CT-2001-00118

References

Bacher A Oberhuber J M and Roeckner E(1998) ENSO dynamics and seasonal cycle in thetropical Pacific as simulated by the ECHAM4OPYC3 coupled General Circulation ModelClim Dyn 1414 pp 431plusmn 450

Bavarian Water Board (Bayerisches Landesamt fuEgrave rWasserwirtschaft Hochwassernachrichtendienst)(2002) Hochwasser im August 2002 mit

Weather Vol 58 November 2003

440

Weather radar echoes (not shown) suggest apredominance of easterly flow and thus of front-al lifting during the early morning of 12August while later the strong southwardadvection (and thus orographic lifting towardsthe Erz Mountains) became important Bothmajor flow directions are reflected in the inte-grated low-level moisture fluxes between thesurface and 700 mbar (shown in Fig 3(c))

Observed rain rates and analysed moistureconvergence in the troposphere (indicated bythe isolines in Fig 3(c)) appear to be consist-ent Maximum convergence rates in the analy-sis for 1200 GMT are equivalent to a rainintensity of up to 97mmdayplusmn 1 ie about07mm in 10 minutes Observed rainfall in-tensities for this time period (Fig 4 in Ulbrichet al 2003) are in a range between 3mm(Zinnwald) and about 03mm (Fichtelberg) in10 minutes

The European Centre for Medium-RangeWeather Forecasts (ECMWF) analysis for thesame time shows upward vertical windsthroughout the troposphere with a maximumof 0015mbar splusmn 1 at 700 mbar (Fig 3(d))They are associated with low-level conver-gence and divergence maxima between 500and 250mbar The large divergence near thetropopause is a peculiarity of the profile Thisoutflow could be produced by latent heatrelease below but this would be more typical oftropical convection than a midlatitude depres-sion We thus prefer the view that horizontaldivergence was enhanced by directional diver-gence of the upper-air flow north-west of thecut-off low which is also evident from the 300mbar radiosonde winds encircled in Fig 3(b)Also both local human observations and mapsof sferics (eg Deutscher Wetterdienst 2002)suggest that there was hardly any thunderstormactivity in the main rainfall regions with a fewexceptions (eg Prague 1200GMT on 12August Fig 3(a))

With respect to the first rainfall event on 7August 2002 there was also a mid-tropo-spheric inflow of warm air from the Mediterra-nean Sea but as revealed by the backtrajectories in Fig 5 (inside front cover) thehumid air at lower levels approached central

Europe from eastern to northern originsLower-tropospheric water-vapour flow into themain rainfall area was similar to the situationof 12 August with a prevailing north to north-easterly flux (Fig 2(c)) Thus orographic lift-ing associated with the upward slope of theWaldviertel certainly contributed to theextreme rainfall amounts Moisture flux con-vergence in the troposphere (Fig 2(c)) wassomewhat less than on 12 August (Fig 3(c))It was associated with a line of convergence(Fig 2(b)) in this area Uplift above the 500mbar level was less intense on 7 August (Fig2(d)) than on 12 August (Fig 3(d)) This canbe understood from upper-tropospheric diver-gence which was weaker than during thesecond event even though wind arrows inFig 2(a) also reveal diffluent upper-air flow

In summary we suggest four main synopticcauses for the extreme summer rainfalls ofAugust 2002

(i) Very humid air advected at low- to mid-tropospheric levels from the westernMediterranean basin around the Alps

(ii) Frontal uplift or a convergence linestalling over the affected areas for a sig-nificant length of time (about 24 hours)

(iii) Orographic lifting associated with air-flow up the northern slopes of thecentral European mountain ranges tothe north-west or west of the lowrsquos core

(iv) The occurrence of these conditions isconnected with an upper-level trough ordepression over central Europe whichcontributed to the dynamically inducedupper-level divergence and which alsoprovided the large-scale steering flow forthe surface low Both the upper-leveltrough and the surface low moved onlyslowly or remained quasi-stationary sothat the area remained under atmos-pheric conditions conducive to extremerainfall for some time

Comparison with recent floods in thesame area

Severe summer floods have affected the RiversOder and Vistula in recent years In 1997 boththe Oder and the Vistula swelled to levels that Locations of lightning flashes

Weather Vol 58 November 2003

438

had return periods of the order of 1000 years atsome gauging stations (Kundzewicz et al1999) The floods were produced by asequence of two intense rainfall periods (4plusmn 9and 17plusmn 22 July 1997) affecting southernPoland Austria and eastern parts of the CzechRepublic (Fuchs and Rapp 1998 Figs 2 and3) Cumulative rainfall amounts for July 1997were of the order of 300 to 500 of the averageJuly precipitation At the mountain station ofLysa Hora (located at 49833rsquoN 18827rsquoEBeskides Mountains Czech Republic) 585mmof rain was reported between 4 and 9 July Themonthly sum of 811mm at this station is moretypical of monsoon rainfall The first of therainfall periods can be assigned to the lownamed Xolska by the Deutscher Wetterdienstwhich moved from Italy into central Europe(Fig 6) The second rainfall event was asso-ciated with the low named Zoe which took asomewhat more westerly path from theBalearic Islands towards the Alps forming a

secondary low on 18 July 1997 (Zoe 2) whichslowly progressed into southern Poland(Fig 6) Both surface lows were associatedwith upper-tropospheric cut-off lows oversouth-eastern Europe

The flooding of the Vistula in summer 2001also originated from two extreme rainfallperiods (1617 and 21plusmn 27 July 2001) Againlarge areas in southern Poland and neighbour-ing countries were affected by torrential rainsObserved rainfall amounts and spatial exten-sion of the rainfall area in 2001 were not aslarge as for the 1997 flood but monthly rainfallsums at several stations in southern Polandamounted to about three times as much as theclimatological mean For example 290 mm(337 of average July precipitation) wasreported at Kielce (50849rsquoN 20842rsquoE) for July2001 and 610mm (292) was reported atKasprowy-Wierch (49814rsquoN 19859rsquoE) (Fuchsand Rudolf 2002) The cyclone track asso-ciated with the first rainfall episode (Axel 1Fig 6) came from the western Mediterraneanforming a secondary low on 16 July 2001 (Axel2 Fig 6) The development at the surface tookplace at the eastern flank of a large-scaleupper-tropospheric trough over westernEurope The second rainfall period of the 2001flood was associated with a low that movedinto south-eastern Europe from the Black Seathus differing from the others It was howeveralso connected with a cut-off upper low oversouth-eastern Europe

The synoptic situations of the August 2002floods and those of the two other recent floodsdescribed above are rather similar They weregenerally associated with surface lows whichmoved from the Mediterranean Sea to centralEurope on a path over (or near) the easternAlps The importance of this specific class oftracks for the generation of summer floods incentral Europe has long been recognised (VanBebber 1891) In his classification of typicalcyclone paths over Europe Van Bebber calledthem `Vb-tracksrsquo a notation still used amongGerman meteorologists The Vb-tracks have arather high potential for leading to extremerainfall by contributing to the main synopticcauses identified for the rainfall events of 2002advection of humid air frontal uplift and air-flow up the northern slopes of the mountains

Fig 6 Tracks of the surface lows relevant to the floods of1997 and 2001 Positions are given in 12-hourly inter-vals Dark grey Low Xolska moving north-eastwardsfrom 0000 GMT on 4 July 1997 (Italy) to 0000 GMT on9 July 1997 (Ukraine) Light grey Low Zoe movingnorth-eastwards from the Balearic Islands (0000 GMTon16 July 1997) A secondary low (Zoe 2) forms over thesouthern Czech Republic at 0000 GMT on 18 July 1997Black Low Axel moving from the Iberian peninsula(0000 GMT on 15 July 2001) to a position south of theAlps within 12 hours A secondary core (Axel 2) is gen-erated further north-east at 1200 GMT on 15 July 2001and reaches Gotland at 0000 GMTon 17 July 2001 Thelow of the second rainfall event associated with theVistula flood of 2001 is not included in the figure

439

Weather Vol 58 November 2003

Climatic change simulations

The recent summer floods in central Europeled to a public discussion of whether theseevents were indicative of ongoing anthropo-genic climate change Numerical simulationsperformed using global circulation models(GCMs) provide a basis for estimating climatechange due to increasing greenhouse-gas con-centrations These models generally produce adecrease of summer rainfall in the Mediter-ranean region and Parry (2000 Fig B8) sug-gests that in most models the region withdecreasing summer precipitation reachesnorthwards into central Europe with a reduc-tion of 7plusmn 8 in the 2050s and 11plusmn 12 in the2080s at the respective grid points At firstglance these results seem to suggest a decreas-ing risk of summer floods in a changingclimate Changes in seasonal mean precipi-tation may however be inappropriate forestimating a changing risk of extreme rainfallTherefore we explored the probable change ofdaily rainfall in central Europe using twoGCMs ECHAM4OPYC3 (Bacher et al 1998)and HadCM3 (Carson 1999 Hulme et al1999) The models were programmed toassume economic development without majoremission reductions (`business as usualrsquo)

A comparison of present-day climate withclimate towards the end of the present century(two 30-year periods considered) suggests ashift towards heavy precipitation events in bothmodels (Fig 7) while total summer rainfalldecreases This result is consistent with otherrecent publications (Booij 2002 Christensenand Christensen 2003) The simulatedincrease of days with intense summer rainfall isassociated with an increasing number ofmeridional weather situations comprising atrough over western or central Europe This isthe weather situation also observed during therecent floods

Concluding remarks

The Elbe flood of August 2002 was caused byextreme rainfall in a large area The associatedcirculation pattern was rather typical of pre-vious flood events both in recent years and inthe more distant past (see Van Bebber 1891)

This situation is characterised by a surface lowmoving on a so-called Vb-track from theMediterranean basin into central Europe Withrespect to the summer 2002 events a positiveinterference of several factors is suggestedincluding advection of humidity from theMediterranean basin a quasi-stationary posi-tion of the tropospheric trough which inducesupper-level divergence and orographic liftingdue to northerly flow west of the surface low

Extreme summer rainfall and resultingfloods have always occurred in this region (VanBebber 1891) The model simulations consid-ered suggest that their frequency mightincrease in the future According to Fricke andKaminski (2001) an increase of days withextreme precipitation can already be observedin the long station time-series of the Hohen-peissenberg Observatory in southern Germany(1881plusmn 2001) For summer this increase isassociated with an increasing frequency ofweather types which are in turn associated withVb-tracks (Fricke and Kaminski 2002) Thisseems to agree with the trends produced insimulations of anthropogenic climate changeIt must however be stressed that it is difficultto determine trends of very rare events (Freiand SchaEgrave r 2001) and we suggest that moredetailed studies are performed on both thesimulated and the observed synoptic develop-ment leading to extreme summer precipitationevents in central Europe

Acknowledgements

We wish to thank Deutscher Wetterdienst forpermission to use ECMWF analysis data and theECMWF trajectory package The work was sup-ported by the EU project MICE (Modelling theImpacts of Climate Extremes) under grantEVK2-CT-2001-00118

References

Bacher A Oberhuber J M and Roeckner E(1998) ENSO dynamics and seasonal cycle in thetropical Pacific as simulated by the ECHAM4OPYC3 coupled General Circulation ModelClim Dyn 1414 pp 431plusmn 450

Bavarian Water Board (Bayerisches Landesamt fuEgrave rWasserwirtschaft Hochwassernachrichtendienst)(2002) Hochwasser im August 2002 mit

Weather Vol 58 November 2003

440

had return periods of the order of 1000 years atsome gauging stations (Kundzewicz et al1999) The floods were produced by asequence of two intense rainfall periods (4plusmn 9and 17plusmn 22 July 1997) affecting southernPoland Austria and eastern parts of the CzechRepublic (Fuchs and Rapp 1998 Figs 2 and3) Cumulative rainfall amounts for July 1997were of the order of 300 to 500 of the averageJuly precipitation At the mountain station ofLysa Hora (located at 49833rsquoN 18827rsquoEBeskides Mountains Czech Republic) 585mmof rain was reported between 4 and 9 July Themonthly sum of 811mm at this station is moretypical of monsoon rainfall The first of therainfall periods can be assigned to the lownamed Xolska by the Deutscher Wetterdienstwhich moved from Italy into central Europe(Fig 6) The second rainfall event was asso-ciated with the low named Zoe which took asomewhat more westerly path from theBalearic Islands towards the Alps forming a

secondary low on 18 July 1997 (Zoe 2) whichslowly progressed into southern Poland(Fig 6) Both surface lows were associatedwith upper-tropospheric cut-off lows oversouth-eastern Europe

The flooding of the Vistula in summer 2001also originated from two extreme rainfallperiods (1617 and 21plusmn 27 July 2001) Againlarge areas in southern Poland and neighbour-ing countries were affected by torrential rainsObserved rainfall amounts and spatial exten-sion of the rainfall area in 2001 were not aslarge as for the 1997 flood but monthly rainfallsums at several stations in southern Polandamounted to about three times as much as theclimatological mean For example 290 mm(337 of average July precipitation) wasreported at Kielce (50849rsquoN 20842rsquoE) for July2001 and 610mm (292) was reported atKasprowy-Wierch (49814rsquoN 19859rsquoE) (Fuchsand Rudolf 2002) The cyclone track asso-ciated with the first rainfall episode (Axel 1Fig 6) came from the western Mediterraneanforming a secondary low on 16 July 2001 (Axel2 Fig 6) The development at the surface tookplace at the eastern flank of a large-scaleupper-tropospheric trough over westernEurope The second rainfall period of the 2001flood was associated with a low that movedinto south-eastern Europe from the Black Seathus differing from the others It was howeveralso connected with a cut-off upper low oversouth-eastern Europe

The synoptic situations of the August 2002floods and those of the two other recent floodsdescribed above are rather similar They weregenerally associated with surface lows whichmoved from the Mediterranean Sea to centralEurope on a path over (or near) the easternAlps The importance of this specific class oftracks for the generation of summer floods incentral Europe has long been recognised (VanBebber 1891) In his classification of typicalcyclone paths over Europe Van Bebber calledthem `Vb-tracksrsquo a notation still used amongGerman meteorologists The Vb-tracks have arather high potential for leading to extremerainfall by contributing to the main synopticcauses identified for the rainfall events of 2002advection of humid air frontal uplift and air-flow up the northern slopes of the mountains

Fig 6 Tracks of the surface lows relevant to the floods of1997 and 2001 Positions are given in 12-hourly inter-vals Dark grey Low Xolska moving north-eastwardsfrom 0000 GMT on 4 July 1997 (Italy) to 0000 GMT on9 July 1997 (Ukraine) Light grey Low Zoe movingnorth-eastwards from the Balearic Islands (0000 GMTon16 July 1997) A secondary low (Zoe 2) forms over thesouthern Czech Republic at 0000 GMT on 18 July 1997Black Low Axel moving from the Iberian peninsula(0000 GMT on 15 July 2001) to a position south of theAlps within 12 hours A secondary core (Axel 2) is gen-erated further north-east at 1200 GMT on 15 July 2001and reaches Gotland at 0000 GMTon 17 July 2001 Thelow of the second rainfall event associated with theVistula flood of 2001 is not included in the figure

439

Weather Vol 58 November 2003

Climatic change simulations

The recent summer floods in central Europeled to a public discussion of whether theseevents were indicative of ongoing anthropo-genic climate change Numerical simulationsperformed using global circulation models(GCMs) provide a basis for estimating climatechange due to increasing greenhouse-gas con-centrations These models generally produce adecrease of summer rainfall in the Mediter-ranean region and Parry (2000 Fig B8) sug-gests that in most models the region withdecreasing summer precipitation reachesnorthwards into central Europe with a reduc-tion of 7plusmn 8 in the 2050s and 11plusmn 12 in the2080s at the respective grid points At firstglance these results seem to suggest a decreas-ing risk of summer floods in a changingclimate Changes in seasonal mean precipi-tation may however be inappropriate forestimating a changing risk of extreme rainfallTherefore we explored the probable change ofdaily rainfall in central Europe using twoGCMs ECHAM4OPYC3 (Bacher et al 1998)and HadCM3 (Carson 1999 Hulme et al1999) The models were programmed toassume economic development without majoremission reductions (`business as usualrsquo)

A comparison of present-day climate withclimate towards the end of the present century(two 30-year periods considered) suggests ashift towards heavy precipitation events in bothmodels (Fig 7) while total summer rainfalldecreases This result is consistent with otherrecent publications (Booij 2002 Christensenand Christensen 2003) The simulatedincrease of days with intense summer rainfall isassociated with an increasing number ofmeridional weather situations comprising atrough over western or central Europe This isthe weather situation also observed during therecent floods

Concluding remarks

The Elbe flood of August 2002 was caused byextreme rainfall in a large area The associatedcirculation pattern was rather typical of pre-vious flood events both in recent years and inthe more distant past (see Van Bebber 1891)

This situation is characterised by a surface lowmoving on a so-called Vb-track from theMediterranean basin into central Europe Withrespect to the summer 2002 events a positiveinterference of several factors is suggestedincluding advection of humidity from theMediterranean basin a quasi-stationary posi-tion of the tropospheric trough which inducesupper-level divergence and orographic liftingdue to northerly flow west of the surface low

Extreme summer rainfall and resultingfloods have always occurred in this region (VanBebber 1891) The model simulations consid-ered suggest that their frequency mightincrease in the future According to Fricke andKaminski (2001) an increase of days withextreme precipitation can already be observedin the long station time-series of the Hohen-peissenberg Observatory in southern Germany(1881plusmn 2001) For summer this increase isassociated with an increasing frequency ofweather types which are in turn associated withVb-tracks (Fricke and Kaminski 2002) Thisseems to agree with the trends produced insimulations of anthropogenic climate changeIt must however be stressed that it is difficultto determine trends of very rare events (Freiand SchaEgrave r 2001) and we suggest that moredetailed studies are performed on both thesimulated and the observed synoptic develop-ment leading to extreme summer precipitationevents in central Europe

Acknowledgements

We wish to thank Deutscher Wetterdienst forpermission to use ECMWF analysis data and theECMWF trajectory package The work was sup-ported by the EU project MICE (Modelling theImpacts of Climate Extremes) under grantEVK2-CT-2001-00118

References

Bacher A Oberhuber J M and Roeckner E(1998) ENSO dynamics and seasonal cycle in thetropical Pacific as simulated by the ECHAM4OPYC3 coupled General Circulation ModelClim Dyn 1414 pp 431plusmn 450

Bavarian Water Board (Bayerisches Landesamt fuEgrave rWasserwirtschaft Hochwassernachrichtendienst)(2002) Hochwasser im August 2002 mit

Weather Vol 58 November 2003

440

Climatic change simulations

The recent summer floods in central Europeled to a public discussion of whether theseevents were indicative of ongoing anthropo-genic climate change Numerical simulationsperformed using global circulation models(GCMs) provide a basis for estimating climatechange due to increasing greenhouse-gas con-centrations These models generally produce adecrease of summer rainfall in the Mediter-ranean region and Parry (2000 Fig B8) sug-gests that in most models the region withdecreasing summer precipitation reachesnorthwards into central Europe with a reduc-tion of 7plusmn 8 in the 2050s and 11plusmn 12 in the2080s at the respective grid points At firstglance these results seem to suggest a decreas-ing risk of summer floods in a changingclimate Changes in seasonal mean precipi-tation may however be inappropriate forestimating a changing risk of extreme rainfallTherefore we explored the probable change ofdaily rainfall in central Europe using twoGCMs ECHAM4OPYC3 (Bacher et al 1998)and HadCM3 (Carson 1999 Hulme et al1999) The models were programmed toassume economic development without majoremission reductions (`business as usualrsquo)

A comparison of present-day climate withclimate towards the end of the present century(two 30-year periods considered) suggests ashift towards heavy precipitation events in bothmodels (Fig 7) while total summer rainfalldecreases This result is consistent with otherrecent publications (Booij 2002 Christensenand Christensen 2003) The simulatedincrease of days with intense summer rainfall isassociated with an increasing number ofmeridional weather situations comprising atrough over western or central Europe This isthe weather situation also observed during therecent floods

Concluding remarks

The Elbe flood of August 2002 was caused byextreme rainfall in a large area The associatedcirculation pattern was rather typical of pre-vious flood events both in recent years and inthe more distant past (see Van Bebber 1891)

This situation is characterised by a surface lowmoving on a so-called Vb-track from theMediterranean basin into central Europe Withrespect to the summer 2002 events a positiveinterference of several factors is suggestedincluding advection of humidity from theMediterranean basin a quasi-stationary posi-tion of the tropospheric trough which inducesupper-level divergence and orographic liftingdue to northerly flow west of the surface low

Extreme summer rainfall and resultingfloods have always occurred in this region (VanBebber 1891) The model simulations consid-ered suggest that their frequency mightincrease in the future According to Fricke andKaminski (2001) an increase of days withextreme precipitation can already be observedin the long station time-series of the Hohen-peissenberg Observatory in southern Germany(1881plusmn 2001) For summer this increase isassociated with an increasing frequency ofweather types which are in turn associated withVb-tracks (Fricke and Kaminski 2002) Thisseems to agree with the trends produced insimulations of anthropogenic climate changeIt must however be stressed that it is difficultto determine trends of very rare events (Freiand SchaEgrave r 2001) and we suggest that moredetailed studies are performed on both thesimulated and the observed synoptic develop-ment leading to extreme summer precipitationevents in central Europe

Acknowledgements

We wish to thank Deutscher Wetterdienst forpermission to use ECMWF analysis data and theECMWF trajectory package The work was sup-ported by the EU project MICE (Modelling theImpacts of Climate Extremes) under grantEVK2-CT-2001-00118

References

Bacher A Oberhuber J M and Roeckner E(1998) ENSO dynamics and seasonal cycle in thetropical Pacific as simulated by the ECHAM4OPYC3 coupled General Circulation ModelClim Dyn 1414 pp 431plusmn 450

Bavarian Water Board (Bayerisches Landesamt fuEgrave rWasserwirtschaft Hochwassernachrichtendienst)(2002) Hochwasser im August 2002 mit

Weather Vol 58 November 2003

440

Berichtigungen vom 28102002 Available onlinefrom http wwwbayernde lfw hndhw110802titelseitehtm

Booij M J (2002) Extreme daily precipitation inwestern Europe with climate change at appropri-ate spatial scales Int J Climatol 2222 pp 69plusmn 85

Carson D J (1999) Climate modelling Achieve-ments and prospects Q J R Meteorol Soc 125125pp 1plusmn 27

Christensen J H and Christensen O B (2003)Severe summer flooding in Europe Nature 421421pp 805plusmn 806

Deutscher Wetterdienst (German Weather Ser-vice) (2002) Extreme precipitation amounts in centralEurope from 1plusmn 13 August 2002 Available onlinefrom httpwwwdwddeenFundEKlimaKLISprod spezial regenrr_extreme_200208_germanypdf

Frei C and SchaEgrave r C (2001) Detection probabilityof trends in rare events theory and application toheavy precipitation in the Alpine region J Clim1414 pp 1568plusmn 1584

Fricke W and Kaminski U (2001) GAW Brief desDeutschen Wetterdienstes No 5 Available online

Fig 7 Climate change signal for the rainfall intensity distribution during summer (June July August) Precipitationsums for different intensity categories are accumulated and the difference between two 30-year periods (representing agreenhouse-gas forcing of the periods assigned to the years 2060plusmn 89 and 1930plusmn 59 according to the respective scenarios) isgiven (a) at a central European grid point of the low-resolution coupled model ECHAM4OPYC3 and (b) as (a) butfor HadCM3 averaged for the four grid points surrounding Dresden Germany

441

Weather Vol 58 November 2003

from httpwwwdwddedeFundEObservatorMOHPhp2gawgaw_briefegaw_briefehtm

ETH ETH (2002) GAW Brief des Deutschen WetterdienstesNo 12 Available online from httpwwwdwddedeFundEObservatorMOHPhp2gawgaw_briefegaw_briefehtm

Fuchs T and Rapp J (1998) Zwei auucirc erge-woEgrave hnlich starke Regenepisoden als Ursache desOderhochwasers im Juli 1997 Klimastatusbericht1997 pp 27plusmn 30 available online from GermanWeather Service httpwwwdwdderesearchklisksbksb97ksb0997pdf

Fuchs T and Rudolf B (2002) Niederschlagsana-lyse zum Weichselhochwasser im Juli 2001 mitVergleich zum Oderhochwasser 1997 Klimastatus-bericht 2001 pp 268plusmn 272 available online fromGerman Weather Service httpwwwdwddedeFundEKlimaKLISprodKSBksb01indexhtm

Hulme M Mitchell J Ingram W Lowe JJohns T New M and Viner D (1999) Climatechange scenarios for global impacts studies GlobalEnviron Change 99 pp S3plusmn S19

Kundzewicz Z W Szamalek K and Kowalczak P(1999) The great flood of 1997 in Poland HydrolSci J 4444 pp 855plusmn 870

Parry M L (Ed) (2000) Assessment of potentialeffects and adaptations for climate change in EuropeThe Europe ACACIA project Jackson EnvironmentInstitute University of East Anglia Norwich

Ulbrich U BruEgrave cher T Fink A H LeckebuschG C KruEgrave ger A and Pinto J G (2003) Thecentral European floods of August 2002 Part 1 plusmnRainfall periods and flood development Weather5858 pp 371plusmn 377

Van Bebber W J (1891) Die Zugstraucirc en der baro-metrischen Minima nach den BahnenkartenDeutschen Seewarte fuEgrave r den Zeitraum von 1875plusmn1890 Meteorol Z 88 pp 361plusmn 366

The online documents referenced above aremirrored under httpwwwmeteouni-koelndecontentforschungklimadiagnosefloodshtml

Correspondence to Dr U Ulbrich Institute forGeophysics and Meteorology University ofCologne 50923 KoEgrave ln Germany e-mail ulbrichmeteouni-koelnde Royal Meteorological Society 2003doi 101256wea6103B

Readersrsquo ForumReaders are invited to contribute short questions onany meteorological topic We will endeavour toobtain answers to all submitted questions

Sunshine records and broken records

In the May 2003 issue of Weather 58 p 170 in thenews item ` Sunniest March since records beganrsquorsquowe are told that March 2003 was the sunniestacross England and Wales since records began in1961 with an average of 541 hours per daySince the Campbellplusmn Stokes (CS) sunshine record-er has existed since the mid-1800s it looked to meas if 1961 should be 1861 but Grant Bigg thenWeatherrsquos Editor informed me that 1961 wasindeed correct being the time that formalrsquorecords of sunshine began However 1960 seemsvery late in the day to start formal records of sun-shine Why the delay

Looking at the Met Office website (Weather andClimate gt Past Weather gt UK gt Station Data)monthly data from ten stations are given free ofcharge for monthly mean maximum and mini-mum temperature mean daily sunshine hoursmean ground minimum temperature rainfalltotals for the month and the number of snowydays Data from other sites are also available ifpaid for Each of the ten records starts in different

years plusmn one Southampton back in 1855 plusmn sun-shine records often starting later than the rain andtemperature records as follows Armagh 1959Bradford 1903 Durham 1890 Lerwick 1930Long Ashton 1930 Oxford 1930 Sheffield 1959Southampton 1895 Stornoway 1930 and Valley1930 all but two dating from 1930 or earlier

Several questions arise

(i) What makes records formalrsquo(ii) Why is 1960 the start date for formal

sunshine records It is very late in the day(iii) Five of the above listed stations started rain

and temperature measurements in thenineteenth century Are these records alsoonly formal from 1961 or since their start

(iv) How many of the formal records today aregenerated by electrical sunshine sensorsrather than by CS recorders and are anybased on `pseudorsquo sunshine measurementsderived from solar radiometers (pyran-ometers) using the World MeteorologicalOrganization (WMO) threshold of 120Wmplusmn 2

The Met Office website data also show thatsunshine records have been stopped during thelast decade or so at six of the ten stations If thiscutback is reflected widely in the data the recordof 541 hours in March this year might be basedon fewer (or different) data than those used to cal-

Weather Vol 58 November 2003

442

Fig 4 Back trajectories computed from the ECMWF operational analysisTrajectories start at 1200 GMT on 12 August2002 near the station of Zinnwald at pressure levels 400 500 600 and 700 mbar and are computed for a time span of 5daysVertical displacement of the back trajectories is shown by the changing colours (See article on p 434)

Fig 5 As Fig4 but for 1200GMT on 7 August 2002 starting point near the station of Rohrbach Starting levels used are300 775 850 and 925 mbar (See article on p 434)