Climate Change and Floods

Climate change and floods – findings and adaptation Technology & Science Water strategies for flood protection in Baden-Wu¨ rttemberg

W. Hennegriff Federal Institute for Environment, Measurements and Nature Protection Baden-Wu¨rttemberg, Landesanstalt fu¨r Umwelt, Messungen und Naturschutz Baden-Wu¨rttemberg (LUBW), Griesbachstraße 1, D–76185 Karlsruhe (E-mail: [email protected])

Abstract The climatic conditions in Southern Germany have changed noticeably in the 20th century, especially during the last three decades. Both in specific regions and interannually, the trends found 35–44 pp 4 No 56 Vol exceed the natural margins of deviation previously known from long measurement series for some measured quantities. The mean and also the extreme floods are expected to increase significantly, although the results of the model chain global model–regional climate models–water balance models are still uncertain. As a precaution an adaptation strategy has been developed for the field of flood protection which takes into consideration the possible development for the next decades and also takes into account the uncertainties.

Keywords Air temperature; climate development modelling; ﬂood protection; ﬂood runoff; precipitation; Q snow cover 2007 Publishing IWA

Initial position and cause The numerous extreme floods of the recent past involving extensive damage have pro- voked heated discussions, both in public and among experts, as to whether these flood events must be considered as part of natural climate variation or as a result of climate change already in progress with long-term future effects. According to current prognoses of climate researchers the large-scale climate in the European region will undergo changes, over and above the natural variations, due to anthropogenic influences and in particular due to the increasing CO2-concentration and other increasing greenhouse gas concentrations in the air. Climate researchers currently assume that the mean global air temperature will increase by 1.4 to 5.8 8C in the coming approximately 100 years due to the anthropogeni- cally caused greenhouse effect (Katzenberger, 2004). This global warming will have effects on the water cycle. In general, an increase in temperature leads to an intensification of the water cycle, which may result in increased evaporation, higher cloud formation and higher precipitation. The statements derived from global climate models for future climate change to date refer mainly to large-scale regions such as Europe. Detailed data on the effects on climate and water balance on a regional scale have not up to now been available at federal state level. Knowledge of possible effects of a changing climate on the water cycle, and in particular on a possibly increasing flood hazard, is in the original interest of the federal states, as these are responsible for flood protection. In the cooperation plan KLIWA (climate change and consequences for water management) of the federal states Baden-Wu¨rttemberg and Bavaria and the German Meterologi- cal Service, possible consequences of climate change on the water balance in the individual river areas of the federal states are assessed. The consequences are portrayed and recommendations are developed in terms of a precautionary water management doi: 10.2166/wst.2007.534 35 policy. The investigations, which were started in 1999, focused first on climate conditions up to now and subsequently on future climate conditions. The questions regarding water management take this up. The examinations to date placed special emphasis on a possible flood intensification.

Climatic development in the 20th century .Hennegriff W. The investigation of the long series of hydrometeorological and hydrological measurements available provides information about the natural variations observed to date and any noticeable changes. These investigations were systematically carried out for Baden- Wu¨rttemberg and Bavaria on the basis of a large data set within the context of KLIWA. Here the long-term behaviour of the ﬂood runoffs (KLIWA, 2002), the mean runoffs, the regional and heavy precipitation, the air temperature (KLIWA, 2005a), the evaporation and the snow cover period (KLIWA, 2005b) were analysed for time periods in the 20th century.

Increase in the air temperature The air temperature at ground level is of crucial importance for the water cycle, as it affects the absorbing capacity of an air mass for water vapour as well as the evaporation. The air temperature is a hydrometeorological variable which, in relation to climate change, can be best simulated in the climate models. Knowledge of the development of this variable is therefore of particular interest. After detailed examination of all available time series in the examined period from 1931–2000, 354 weather-stations proved to be suited for a further regionalisation of the air temperature. Time series of the daily mean temperatures in the catchment areas were determined from the regionalised measurement series for 33 investigation catchment areas, which cover Baden-Wu¨rttemberg and Bavaria. The following remarkable results were found: † increase in the annual mean temperature: between 0.5 and 1.2 8C † rise in the monthly mean temperature for August: between 0.7 and 1.7 8C † rise in the monthly mean temperature for December: between 1.8 and 2.7 8C † increases in temperature – although less signiﬁcant than in August and December were also found in the months January, February, March and October. † region-speciﬁc distinctive features: e.g. main focus of the temperature increase in winter at lower altitudes; higher temperature increase in the west of Baden-Wu¨rttemberg. The greatest increase in the monthly mean temperature occurs in most regions in December. The regional values can be seen in Figure 1.

Milder winters with less snowfall Long-term examinations of the snow cover conditions should be heeded because global warming will also probably result in changes in the frequency and duration of snow cover. Changes in snow cover regimes and their parameters have effects on water balance, especially on the soil water balance, the groundwater recharge and the regime of the catchment runoff (ﬂood formation). The parameters snow cover period, snow cover time, longest snow cover period (winter cover), start of the maximum snow cover height, constancy of snow cover, conservation of winter cover and maximum water equivalent values are best suited to describe the snow cover conditions and the water reserves stored in the snow cover. All the snow cover parameters mentioned correlate strictly with the elevation of the ground. The winter cover allows conclusions to be drawn about changes in the characteristics of winter periods. The trend towards winters with less snowfall with 36 less lasting snow cover is deﬁnitely apparent. .Hennegriff W.

Figure 1 Increase [8C] in the monthly mean temperature in December in the period 1931 – 2000 in the catchment areas examined in KLIWA

Up to moderate altitudes the durations decrease markedly in general. In the observed period, however, some regional distinctive features can be seen (Figure 2). In the western parts of the regions (Upper-Rhine plain and the western declivity of the Black Forest) the duration of snow cover decreases by approximately 50% and more on lower ground and decreases at moderate altitudes to 10 to 20%. In the higher regions mean values under 10% are observed. Here, too, the trend weakens with increasing altitude. However, only in isolated cases are values observed where the trend is reversed. At lower and moderate altitudes the number of days with snow cover has decreased markedly: † approximately 30–50% in lower regions ( , approximately 300 m above sea level), † approximately 10–20% at moderate altitudes (between 300 and 800 m above sea level), † less than 10% on high ground, or in some cases even increasing at higher altitudes ( . approximately 800 m above sea level).

Increase in precipitation Extensive and, as far as possible, homogeneous data on precipitation behaviour are funda- mental prerequisites for the better understanding of the interaction of climate and water cycle. For this purpose the long precipitation series on all available stations in Southern Germany were interpolated with a geostatistical method to yield grid point precipitation and daily catchment precipitation heights were calculated. Statistical parameters for the daily values for the catchment area in one month were then analysed as representative partial samples. The series of the monthly values formed in this way were analysed in detail in a time series analysis. Station time series were employed to examine heavy precipitation. The following changes in precipitation behaviour appear to be especially remarkable: † signiﬁcant decrease in the catchment precipitation in the summer half year, especially in North Wu¨rttemberg 37 .Hennegriff W.

Figure 2 Relative Trend [%] in mean duration of snow cover, series 1951/52 to 1995/96

† increase, in most cases signiﬁcant, in the catchment precipitation in the winter half year † regional clear increase in heavy precipitation by 30–35% in the winter half year; however, in summer only small changes † regional foci of heavy precipitation in winter can be found in the Black Forest, in the Northeast of Baden-Wu¨rttemberg (see Figure 3) † winter half year more humid, summer half year drier. In the winter half year, the precipitation-bearing Western weather fronts have increased in Southern Germany. These fronts that are particularly important for the formation of ﬂoods which may go some way towards explaining the changes found. This can be seen in Figure 3.

Figure 3 Increase [%] in heavy precipitation at individual stations with 24 h-duration in the winter half year 38 in the period from 1931-2000 Long-term behaviour of ﬂood runoffs The investigation of the long-term behaviour included determination of any linear trends present in the time series of the annual and monthly highest runoffs. The annual and monthly highest runoff values at 107 gauges, which have long observation series since at least 1931, formed the basis for the trend investigations. Furthermore 51 gauges with shorter time series, i.e. with an observation start after 1932, were included in the analysis.

The long-term behaviour of the highest runoffs can be characterised as follows. Hennegriff W. † When examining the annual series from 70 to 150 years duration, the majority of annual highest runoff levels do not show significant changes. † When examining the last approximate 30 years the highest runoffs show increasing trends at many gauges. † The frequency of winter floods has increased since the 1970s. † The monthly flood runoffs in the winter half year since the 1970s are higher than in the time before the 1970s. In summary, it can be established that only in the last 30 to 40 years does the examined runoff time series demonstrate regional increases in flood runoffs.

General conclusion The climate conditions in Southern Germany, which have an impact on the entire water balance, have changed noticeably in the past century, especially during the last three decades. In speciﬁc regions the trends found exceed the natural margin of deviation, derived from long measurement time series, for some of the variables examined. The results suggest the explanation that the global and regional climate is human-induced, a basic premise which the international climate research community no longer questions. The investigations carried out for Southern Germany in the context of KLIWA agree in this respect with the trends stated in comparable studies. In particular, the KLIWA results on the changes in heavy precipitation have caused the German Meterological Service to put the KOSTRA values on a new basis as KOSTRA-DWD-2000. This will be of importance for the assessment practice in water management.

Modelling climate development up to 2050 The trends examined to date in the measurement time series of climatological and hydrological parameters cannot be directly extrapolated into the future, as climate processes and their complex interactions are nonlinear and may vary over time. In order to make statements about possible climate changes in Southern Germany for the next decades, regional climate scenarios had to be developed. As an optimum method has not yet been devised for this purpose, different institutions were given the task of establishing regional climate scenarios as a part of KLIWA. They were required to develop three different methods, namely: † a statistical downscaling method using cluster analysis (Potsdam Institut fu¨r Klimafol- genforschung (PIK)), † a statistical dynamic downscaling using classiﬁcation of weather conditions (Fa. Meteo-Research (MR)) and † a regional dynamic climate model (REMO) (Max-Planck-Institut fu¨r Meteorologie (MPI)). In order to achieve comparable results, the KLIWA partners established conditions that were to a large extent identical: measurement data 1951-2000, veriﬁcation period 1971- 2000, global model ECHAM 4, IPCC emission scenario B2, scenario period 2021–2050. After comparison and evaluation of the results of the three methods, which, as expected, delivered a certain range of results, further evaluations were primarily made on 39 the basis of the results of the Meteo-Research-method. In summary, this scenario in the period 2021 – 2050 showed the following.

Air temperature The air temperature will continue to increase clearly in future in Baden-Wu¨rttemberg. The annual mean temperature increase amounts to 1.7 8C. In winter, the increase of approxi- .Hennegriff W. mately 2 8C is highest, in summer the increase amounts to 1.4 8C, as can be seen in Figure 4. The expected temperature increase in winter is of special importance, as the temperature has great inﬂuence on the temporary storage of precipitation as snow and thus may be decisive for the expected ﬂow conditions in future.

Precipitation For the selected future scenario the increase in the mean annual values of precipitation amounts to approximately 8% with a bandwidth of approximately 4% to approximately 17%. The large-scale precipitation will change by maximum -4% in summer in Southern Germany. On the other hand, it is to be expected that precipitation in winter will increase signiﬁcantly (see Figure 5). Depending on the region, the varying increase will amount to up to 35%. One area showing striking annual precipitation totals with relatively high changes is that on the southwestern-northeastern half-circle (Black Forest-Odenwald- Spessart-Franconian Forest). However the available regional climate models cannot currently provide quantitative data for the future development of convective short-period precipitation (thunderstorms), which are of importance for urban drainage and for ﬂoods in smaller catchment areas.

Weather conditions In winter, an increase is to be expected in the frequency and duration of west weather conditions (west condition cyclonic), which are important for ﬂood formation. In summer, great changes are not to be expected.

Figure 4 Change of the future air temperature [8C] in the winter half year compared to today with 40 differences between scenario (2021-2050) and current state (1971-2000) .Hennegriff W.

Figure 5 Changes in future precipitation values in the winter half year compared to today with differences between scenario (2021-2050) and current state (1971-2000)

General conclusion The results of the further development of climate change on the basis of regional climate models can generally be summarised as follows. † Warming continues; the air temperature will increase, in particular in winter. † Precipitation will increase in winter. † An increase in the duration and frequency of rain-laden west weather conditions (especially west condition cyclonic), which is important for ﬂood formation in winter, is to be expected. These changes will have considerable impacts on water balance, especially the runoffs into ﬂowing waters.

Adaptation of flood protection planning Climate change and flood runoff The data of the regional climate scenarios were employed as input quantities for the water balance models (WHM), in order to make statements on the impact of climate change on water balance (e.g. runoffs into flowing waters). The water balance models on the basis of LARSIM (Bremicker, 2000) are available as a 1 km grid for the whole of Baden-Wu¨rttemberg. The modelling of water balance focused at first on the possible future changes in runoff, with initial consideration of the effects of flood runoff. For this purpose the runoff values obtained from water balance modelling were analysed with methods of extreme value statistics. The results predict a marked increase in mean flood levels (MHQ), and also in the extreme runoff. Although the results of the model chain (global model – regional climate model – water balance models) and the model assumptions contain uncertainties, the results all point in the same direction. Thus for the period considered up to the year 2050 it can be assumed that floods will intensify due to climate change in Baden-Wu¨rttemberg.

Adaptation of flood protection planning Against this background, it was necessary to take precautions and develop an adaptation strategy which takes into consideration the possible development for the next decades, 41 but also takes into account the existing uncertainties. For this reason decisions should, at their core, be harmless in the long run while at the same time remaining adaptable should the need arise (e.g. as a result of new findings in climate research) according to the “flex- ible and no regret” strategy. The evaluations provided reasons to modify the method previously used to determine design runoff and, as a result of the climate change, to consider a “load case climate .Hennegriff W. change”. On the basis of practical case studies it has been proven that a consideration of the effects of climate change for technical flood protection measures in most cases would have led to a relatively moderate cost increase, if this load case had already been taken into consideration during planning and if at least appropriate precautions had been taken during construction for later adaptation. Later adaptations, by contrast, mostly involve very high costs. For this reason, when planning new flood protection measures in future, the load case climate change should also be examined. This should include demonstration of the consequences of the load case for the measures being planned and which additional costs are expected as a result. Decisions have then to be made, on the basis of available findings, to what extent the adaption necessary for future climate change should already be taken into consideration in current work. The possibilities for additional later adaptation should also be taken into account. The use of the load case climate change for flood protection conceptions, where the implementation has already started or had already been completed, is currently not planned.

Increase in the design runoff Increased design runoff has to be taken as the basis for the load case climate change. This is carried out with a supplement (“climate change factor”) to the currently

valid design value (e.g. HQ100). In Baden-Wu¨rttemberg, the climate change factors for the runoffs differ between regions depending on the recurrence time (recurrence

interval Tn). In order to assess the magnitude of the climate change factors, the results of the regional climate scenarios established in the context of KLIWA, were employed as input quantities for water balance models and the runoff determined in water balance modelling was evaluated using extreme value statistics. The results for the future scenarios were compared with those of the current state. This was used to determine regional climate change factors for runoff for different recurrence intervals. In summary, ﬁve areas result for Baden-Wu¨rttemberg, each having a different climate change factor, which can be seen in Figure 6. By spatial allocation to one of the ﬁve areas, climate change factors for any area in the country and for different recurrence intervals are available. The respective values are listed in Figure 7.

The runoff HQTn resulting from ﬂood regionalisation or hydrological model calculation can be directly increased with the climate change factor fT,K for the runoff in the load case climate change (HQTn,K):

HQTn;K ¼ f T;K·HQTn

In the middle and lower river catchments of the Neckar, for example, the climate change

factor has a value of 1.15 for the century ﬂood runoff (HQ100). Thus the value for HQ100 must be multiplied by the climate change factor 1.15 for the load case climate change 42 (see Figure 7). .Hennegriff W.

Figure 6 Areas in Baden-Wu¨rttemberg with uniform climate change factors fT,K

The procedure in the load case climate change for new flood protection planning has been included in the guideline “Determination of the design flood for technical flood protection equipment” (Landesanstalt fu¨r Umweltschutz Baden-Wu¨rttemberg, 2005). Organis- ations supporting non-governmental projects have been recommended to proceed in a comparable way.

Figure 7 Climate change factors fT,K to determine the design ﬂood for the areas or river catchments in Baden-Wu¨rttemberg 43 Examples The following examples should explain how plans can be implemented in case of increased design values, i.e. taking into consideration the load case climate change. † Planning of a ﬂood protection dam: The dam is built according to current guidelines. However, additional measures are taken, which would not be required according to previous planning regulations. For example, an additional strip of ground on the valley

.Hennegriff W. side is reserved, enabling a future increase of the dam, if necessary, without additional problems. † New constructions where a future alteration or adaptation is not possible or is only very expensive (e.g. bridges), should immediately be planned to take account of future increased calculation parameters relating to the level of water, if necessary. † New constructions where a future adaptation is less difﬁcult (e.g. river walls) should in view of the construction features (e.g. statics) be planned to a higher speciﬁcation than currently required so that any further adaptation necessary (e.g. increase in height using stationary or mobile elements) would be possible without high costs.

Outlook The findings to date have already led to concrete consequences, not least in terms of precautions. When designing flood protection plans the expected consequences of climate change have already been taken into consideration. The investigation of the effects of climate change in other areas of water management, primarily in view of the future development of low water discharge and groundwater recharge, is currently being started. 70 million euros are available for the development of technical flood protection per year by the Federal State of Baden-Wuerttemberg. The KLIWA partners are aware that the findings gained to date still include considerable uncertainties. With the progress of world-wide climate research and improving modelling instruments the findings to date will inevitably have to be developed further. Further investigations can be continued relatively easily by establishing water balance models for the individual river basins. The water management has already proven that it can adapt to changed conditions. Finally, climate change is a great challenge.

References Bremicker, M. (2000). Das Wasserhaushaltsmodell LARSIM – Modellgrundlagen und Anwendungsbeispiele, Freiburger Schriften zur Hydrologie, Vol. 11, Institut fu¨r Hydrologie der Universita¨t Freiburg. Katzenberger, B. (2004). Bisherige Erkenntnisse aus KLIWA-Handlungsempfehlungen, KLIWA-Berichte, No. 4, Mu¨nchen. pp 197–204. KLIWA (2002). Langzeitverhalten der Hochwasserabﬂu¨sse in Baden-Wu¨rttemberg und Bayern, KLIWA- Berichte, No. 2, Karlsruhe. KLIWA (2005a). Langzeitverhalten der Lufttemperatur in Baden-Wu¨rttemberg und Bayern, KLIWA- Berichte, No. 5, Mu¨nchen. KLIWA (2005b). Langzeitverhalten der Schneedecke in Baden-Wu¨rttemberg und Bayern, KLIWA-Berichte, No. 6, Mu¨nchen. Landesanstalt fu¨r Umweltschutz Baden-Wu¨rttemberg (2005). Leitfaden ”Festlegung des Bemessungshochwassers fu¨r Anlagen des technischen Hochwasserschutzes“, Karlsruhe.