The flood of June 2013 in Germany: how much do we know about its impacts?

Introduction Conclusions References

beings, infrastructures and buildings to flooding (also referred to as elements at risk or damage potential), as well as the susceptibility of these elements to inundation. The extent of vulnerability and risk is strongly influenced by the resilience of the affected society or its ability to resist: the better the preventive and protective measures, early warning systems and emergency response have been developed, the less severe the 5 resulting damage will be. In general, adverse effects of floods are divided into direct and indirect damage (Smith and Ward, 1998). While direct damage, such as fatalities and injured people as well as damaged or destroyed buildings, are directly caused by a physical contact of the element at risk with the flood water, indirect damage occur in space and time 10 outside the actual event. Among these effects are traffic and business disruptions, but also migration or long-term psychological illnesses.
Accounting for all impacts and costs of a particular event is complicated for many reasons (Downton and Pielke, 2005). To begin with, damage to buildings seems to be monetised easily since the goods concerned are traded on the market (Merz et al., 15 2010). Thus, the damage costs can be estimated on the basis of the necessary repair works and materials in a first instance. For some applications such as cost-benefit analyses, however, the financial damage that is based on repair and replacement costs has to be depreciated by the betterment that the damaged structures underwent during reconstruction; taxes also have to be excluded (see Merz et al., 2010). A monetary 20 estimate can also be put on disruptions of operations, turnover losses or costs incurred by delivery detours. However, further indirect costs of disasters along production chains are difficult to measure and can often only be assessed by models (Greenberg et al., 2007;Meyer et al., 2013). Moreover, many losses (and benefits) associated with a flood event are intangible and difficult to monetarise or even to observe. Health Groeve et al. (2013), as well as on the overall context of the analysis and its underlying monetary assessment.
The true costs of flood events may hence include hidden costs, such as health effects and long term societal impacts, and hidden benefits caused by, e.g. extra compensation payments, which are difficult to identify and quantify (Downton and Pielke, 2005). Due 10 to this complexity, there is currently a clear focus on accounting direct damage costs or primary effects of actual events (Pielke and Landsea, 1998) by using economic and/or human indicators (IRDR, 2015). While human indicators such as the number of people killed, injured or evacuated can be determined fairly reliable shortly after the event, a reliable estimate of the direct economic or financial costs of an event can often only be 15 made after several years when all repair works and compensation payments have been completed. Using flood damage data provided by the National Weather Service (NWS) in the USA, Downton and Pielke (2005) demonstrated that reliable loss figures require regular data updates and consistent definitions of the damage components included. Data consistency is, however, difficult to assess if sub-amounts such as damage in Introduction Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | standardized datasets have been constantly demanded since an accurate, comparable and consistent impact database is required for many applications, among others: to assess the influences of climate, population growth, land use and policies on trends in losses and damage (Downton et al., 2005), to improve risk assessment methods by calibrating and validating loss models 5 with real data (De Groeve et al., 2013), to identify drivers and root causes of disasters and to deepen our understanding of damaging processes (disaster forensic; DKKV, 2012), to set priorities between competing demands for national and international budget allocations (Guha-Sapir and Below, 2002), 10 to evaluate policy successes and failures on the basis of trends and spatial patterns of damage, to think about new policies (insurance, climate policies), to set priorities of research funding, and to evaluate contributions of science to real-world outcomes (Downton and Pielke,15 2005).
Since damage information is assumed to be collected more systematically and comprehensively for a major flood than for a small event and information is more likely to be shared among different agencies and institutions (Downton and Pielke, 2005), this paper explores what data is currently available to describe the impacts of the flood 20 event of June 2013 and what can be learnt from them about the types and severities of flood impacts in different sectors. Finally, it will be discussed how good current data and information are and what could be done to create better impact data.
In consistency with the European Floods Directive (2007/60/EC) that aims to establish a framework for the assessment and management of flood risks in Europe Introduction and to reduce adverse consequences of flooding for human health, economic activities (and assets), cultural heritage, and the environment, predominantly data and impacts on these domains are explored in this paper. Further, we mainly concentrate on direct flood impacts on different scales, i.e. from the national down to the property (asset) scale, due to the above-mentioned complexities and problems that are associated 5 with indirect and long-term effects. In the next section, the used data sources are introduced, before the actual flood impacts are presented per damage type and scale (if applicable) in Sect. 3. This part of the paper is accompanied by an overall evaluation of the data content and quality in comparison to recently published guidelines on recording disaster losses (Corbane et al., 2015;IRDR, 2015), which will 10 be introduced in each section dealing with a damage category. The paper ends with recommendations on future event documentation and loss data collection.

Data sources
Three main data sources were used for this study: (i) governmental reports on the flood in June 2013, (ii) communications on disruptions of road and railway traffic, 15 and (iii) computer-aided telephone interviews among flood-affected residents and companies.

Governmental reports
General information on the flood impacts was collected from official governmental reports on the flood on the federal/national level ( (2015) on the national and the subnational level. In addition, the reports provided insight into expenses for emergency services as well as into impacts on cultural heritage and the environment. 15 Since the Floods Directive addresses impacts on economic activities, disruption of transportation plays an important role. Therefore, communications on the disruption of road and railway traffic were analysed.

Road traffic
In order to capture the impact of the directions due to flooding, traffic obstruction due to flood, both directions of traffic closed, a detour has been instated" (source: Saxon Police, 2013, own translation). All situations that posed an obstruction to road traffic, such as a closed road on one side or on both sides, narrowing of lanes, obstructions by traffic (e.g. by emergency vehicles) as well as dangers (e.g. an increase in game crossing the road due to the 5 flood) were further considered as traffic obstruction. Repeated identical reports were merged so that they counted as one traffic obstruction. However, should a piece of information in the report change, for example the stated section of the affected road, then the report was captured as a new traffic obstruction. A traffic obstruction was deemed to have ended, as soon as 10 information in the report changed so that this could be captured as a new traffic obstruction, it had been reported that the street was traversable once again or that the danger on the road had passed, the traffic obstruction did not appear in the police traffic reports any longer.

Railway transportation
The German Railways Corporation (Deutsche Bahn AG; DB) provided several internal communication maps, in which the railway segments that were interfered due to extreme weather conditions or flooding are shown. The maps cover the time period between 3 June and 1 July 2015 with, however, some days without any information. Introduction

Computer-aided telephone interviews
To capture more detailed flood effects on the level of individual properties (assets, households), information from flood-affected residents and companies was systematically gathered. 5 Computer-aided telephone interviews (CATI) were conducted among households in the flood-affected regions of Germany nine months after the event. On the basis of information from affected municipalities, flood reports or areas experiencing flooding, street lists were compiled and the telephone numbers of residents potentially affected by the flood were searched. For the survey on the 2013 event, a comprehensive survey 10 was conducted, i.e. all the searched telephone numbers were contacted. In total, 1652 interviews were completed between 18 February and 24 March 2014 with affected households. In the survey, the term "affected" was defined as a household that had suffered (financial) flood damage in May or June 2013. Similarly to former surveys (see Thieken et al., 2005;Kienzler et al., 2015), the main 15 objective was to investigate how financial flood losses are influenced by other factors, for example flood characteristics or private mitigation. However, after the June 2013 flood, some questions were posed regarding flood effects on health and wellbeing as well as on the assessment of the (governmental) aid for reconstruction. Overall, the questionnaire addressed the following topics (in the order of appearance): physical and financial flood damage to the building and the household contents;

Flood-affected residents
building ownership and further information on the residential building (or the rented apartment);

5
previously experienced flood events and flood awareness; long-term preventive and protective measures undertaken by the affected household and motivation (not) to do so; aid and financial compensation; socio-demographic information. 10 Information on health effects and the perceived severity of different damage types presented in Sect. 3.1.2 of this paper are based on this survey.
The above-mentioned surveys that were conducted a few months after the floods in 2002, 2005 and 2006 (see Kienzler et al., 2015) were complemented by a follow-up household survey in autumn 2012 (n = 910 households), i.e. ten years after the flood 15 in 2002. The survey follow-up focused on long-term (health) effects of the floods as well as property-level mitigation measures. These data are used in Sect. 3.1.2 to illustrate short and long-term flood effects on affected residents.

Flood-affected companies
Companies that had been affected by the flood in June 2013 were surveyed with 20 regard to the losses incurred and the circumstances influencing the type and amount of damage. For the sampling procedure, street lists were compiled on the basis of information obtained from municipalities, flood reports or mapped inundation areas and were further used to determine the telephone numbers of companies potentially 13 Introduction affected by the flood. To include some large-sized companies in the random sampling as well, these were searched additionally from flood reports. Affected companies were surveyed from mid of May to mid of July 2014. Again, the term "affected" was defined as an enterprise that had suffered (financial) flood damage. The information was gathered through CATI with the individual in the company who 5 was best placed with providing information on the flood. In total, 557 interviews were completed. The interviews lasted 15 to 35 min on average; the questionnaire covered approx. 90 questions on the following topics (in the order of appearance): company description (sector, size, number of buildings, assets, perceived vulnerability with regard to flooding, etc.);

Flood impacts on human health
The effects of flooding on health can be significant and may concern both, physical and mental health. Physical health effects are deaths due to drowning, electrocution, heart attacks, vehicle-related accidents etc. as well as injuries, illnesses and infections 5 that require medical assistance and result directly from the flood, for example due to a lack of sanitation, contaminated water, chemical hazards or mildew (within wet or insufficiently reconstructed buildings; IRDR, 2015). Mental health effects might be acute or long-term due to a loss of family members or friends, displacement, destruction of homes, delayed recovery and water shortages (Menne and Murray, Human indicators in disaster loss databases are commonly related to physical health or the displacement and movements of people caused by the flooding; mental health effects are usually not explicitly reported. For example, IRDR (2015) proposes the 20 numbers of dead, missing, injured and exposed people as primary human impact indicators, while the numbers of homeless, evacuated, relocated and affected people are regarded as secondary. Some indicators, e.g. dead and missed people, are mutual exclusive, others, e.g. homeless, evacuated and relocated people, are not since they correspond to consecutive management phases of a damaging event (IRDR, 2015). 25 In the European guidance for recording disaster losses (Corbane et al., 2015), the number of deaths, missing people as well as directly affected people are recommended as minimum information that should be recorded with regard to human losses. All  information should be provided on the NUTS 2 or NUTS 3 level (NUTS stands for Nomenclature of Territorial Units for Statistics). In Germany, these mainly correspond to the 38 (former) administrative regions (Regierungsbezirke) as well as 402 urban and rural administrative districts (kreisfreie Städte und Landkreise), respectively. Since almost no information was provided for administrative levels below the federal states, 5 an overview of human loss indicators (see Table 1) can currently be provided for this level (NUTS 1) only.   Table 1). However, the term "affected" is not clearly defined, nor is its relation to the categories "injured" and "evacuated" in terms of ex-/inclusiveness. Due to this ambiguity, IRDR (2015) recommends using exposed people, defined as the number of people who permanently or temporarily reside in the hazard area before or 20 during the event, in a first instance, as this number can be more reliably determined from census data and geographic information on the flooded area. So far, the number of people exposed to the June 2013 flood has not been determined. As a further human-related indicator, the number of helpers in emergency services and relief or aid organisations totalling to more than 1 million person days Germany 25 was often reported in governmental reports dealing with the June 2013 flood and hence added to Table 1  Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | flood (damage) without being organised in an emergency service or an aid or relief organisation. With regard to the 2013 flood, the numbers illustrate that the magnitude of responders and helpers is similar to the amount of people directly affected by this widespread flood event. Table 1 clearly demonstrates that the reporting of the federal states was not focussed 5 on human losses. Only in the Bavarian report (Annex 6 in BMF, 2013), numbers for all categories of Table 1 were mentioned. In most of the states, no (accessible) numbers were reported. Therefore, more transparent and systematic reporting procedures are needed to evaluate the quality of the aggregated data and to reach a comprehensive report on human losses that fulfils the minimum requirements proposed by Corbane

Flood impacts on affected residents and perceived severity
To obtain more insights of the variety and severity of flood impacts on affected residents, the surveyed households (see Sect. 2.3.1) were asked to indicate from a list of ten possible flood damages, which of these had affected them in June 2013 and 15 how seriously they perceived each of the witnessed damage type. The answers could be graded on a scale of 1 (= damage was not serious at all) to 6 (= damage was very serious). The perceptions of all respondents to a particular damage type resulted in average assessments between 3.0 and 4.6 ( Fig. 2). The damage types that were assessed on average with 4.0 or worse -and were thus evaluated as serious -cited 20 as a priority and included psychological stress or other stresses, reinstatement works (e.g. cleaning or repairs), supply problems (e.g. no electricity, water etc.) as well as damage to buildings and household contents (Fig. 2). This highlights that mental health issues and disruption of daily life are of great importance for affected people. In an open question regarding the type of stress undergone and grievances in detail, uncertainty about the future, worries with regard to family, existence and subsistence, and the future, fears of loss, panic, trauma, shock, crying fits or nervous breakdowns were cited most frequently. In addition to these, sleep disorders or nightmares were mentioned, as well as feeling restless, tense and nervous 5 or agitated. Physical symptoms manifested themselves most frequently in the form of states of exhaustion or lack of sleep; joint, bone, muscle or nervous complaints; infections, inflammation, (skin) irritations or the exacerbation of pre-existing illnesses or conditions. It is noteworthy that the management of the flood situation aggravated in 10 the case of persons with chronic illnesses or conditions. Psychological stress is, however, not limited to the period of the actual flood event, but can still remain in existence a long time afterwards. The above-mentioned surveyed group of affected persons with health impairments was therefore additionally asked about the extent to which they were still stressed by the flood event at the point in time 15 of the interview (answer scale from 1 = "I am not stressed by it any longer/I feel like I did before the event" to 6 = "I am still very stressed by it"). Slightly more than a third of the respondents (35 %) were still very or extremely stressed as a result of the flood as much as nine months after the event (answers 5 and 6); by contrast, a further third hardly felt stressed any longer or not at all (answers 1 and 2). 20 However, the 2013 flood was still very prevalent in the minds of all the residents affected. This is clearly evident from the results to the question: "How often have you thought about the June 2013 flood over the past six months?" At the point in time of the survey, i.e. approximately nine months after the event, 35 % of all the affected persons still thought about the 2013 flood once or several times a day, 50 % still at least once 25 a month to several times a week (Fig. 3). This distribution of answers clearly differs from the answers of affected persons who were asked the same question in autumn 2012, i.e. ten years after having witnessed the severe flood of August 2002 (Fig. 3). On the one hand the comparison illustrates the extent to which a flood can change daily life and thinking, on the other hand the long-term and ongoing impression that an extreme flood can leave behind is evidenced: ten years after the event of August 2002, only 20 % of respondents stated that they never thought of the event in the six months preceding the interview. 8 % still thought of it approximately daily. However, it is worth noting that the thoughts about the flood were not negative throughout: the experience 5 of solidarity and a sense of community were often positively highlighted. According to Kuhlicke et al. (2014), affected households in Saxony that had been flooded up to three times in recent years (i.e. in 2002, 2006 or 2010, and 2013), perceived the flood impacts more severe than households that had been affected by flooding in 2013 for the first time. In addition, households that already suffered flood 10 damage several times thought considerably more often about resettlement which might have severe consequences for flood-prone communities that do not get flood protection (Kuhlicke et al., 2014).
In conclusion, the survey among flood-affected residents highlights the importance of physical and particularly mental health issues caused by flooding. This is contrasted 15 by the little attention this domain received in official governmental flood documentations and reports in Germany.

Overview of impacts on economic activities (and assets) on the regional and national scale
In industrialised countries, economic or financial losses caused by natural hazards 20 are a major concern and achieve a lot of attention during and after disastrous events. Quick and reliable loss estimates are requested by the (re-)insurance industry as well as by governmental institutions. However, data on economic or financial losses are fairly uncertain Downton and Pielke, 2005;Downton et al., 2005 that physical damage indicators should deliver information on the number of damaged or destroyed houses, educational centres (e.g. schools, kindergartens) and health facilities (e.g. hospitals). Optionally, further aggregated damage indicators can be provided, i.e. on the total area of destroyed or affected crops and woods (in hectares), the number of lost four-legged livestock, the number of damaged or destroyed 5 governmental and administrative buildings, the number of damaged or destroyed industrial and commercial facilities as well as the length of damaged or destroyed roads and railways (in kilometres) and the number of damaged or destroyed transportation infrastructure such as bridges, airports and marine ports (Corbane et al., 2015). These physical damage indicators are further translated into economic monetary indicators, in particular into the overall direct tangible loss, i.e. the monetary value of the physical damage to capital assets. This loss should ideally be disaggregated over all sectors or loss owners and accompanied by information on the loss bearer. Expenditures for emergency services and clean-up are further costs to be recorded optionally (Corbane et al., 2015). 15 As a minimum requirement the overall direct damage should be reported on NUTS 2or NUTS 3-levels (see Sect. 3.1). As outlined above, almost no information was provided for administrative levels below the state-level. Therefore, the overview of damage and losses can currently be provided for the NUTS 1-level only. Table 2 summarises the information that was collected for the flood of June 2013 on the 20 minimum indicators on direct damage and economic loss as proposed by Corbane et al. (2015).

Overview of financial losses
Twelve out of 16 federal states were affected by flooding between 18 May and 4 July 2013; in parts of eight federal states a state of emergency was declared 25 (BMI, 2013). Table 2 illustrates that data on the physical damage indicators are so fragmentary that they do not allow a sound interpretation. Only from the Saxon report (Annex 14 in BMF, 2013;Saxon State Chancellery, 2013), information for all minimum 20 estimated at EUR 2.699 billion, was reduced to between EUR 1.5 and 2 billion (written communication from the Saxony-Anhalt Ministry of Finance dated 15 April 2014). In addition, the damage to the infrastructure of the Federal Government -this involves damage to the federal assets regarding railways, motorways and navigable waterways, as well as to the administrative buildings of the Federal Government -clearly lies 15 below the EUR 1.484 billion estimated initially (Table 2). In its response to a small parliamentary enquiry concerning the flood relief funds, the Federal Government recently assumed that only a sum of approximately EUR 114 million was in question (Federal Parliament, 2015). The same enquiry also provides recent amounts of losses that have been claimed to governmental relief funds by the end of June 2015 (see 20   Table 2). In contrast to the application to the European Union Solidarity Fund, these numbers, however, seem not to include expenses for emergency response, nor insured losses. Losses that property owners bear themselves are probably also neglected in these figures. With these corrections and considerations, the total direct loss will probably not exceed EUR 6 billion. 25 Even the most recent numbers indicate that the compilation of the overall financial losses is still preliminary. Many of the damage claims have not been resolved conclusively and to some extent unforeseeable losses that had been incurred but have not been reported may still appear. According to the administrative arrangement for the NHESSD Act to Establish Reconstruction Funds passed in 2013, applications for reconstruction aid could be submitted until 30 June 2015. The period for final approval was recently extended to 30 June 2016 (Federal Parliament, 2015). Only thereafter will it be possible to compile a conclusive loss statement. Nevertheless, it is already possible to look at the spatial and sector-wise distribution 5 of losses.
. It is noteworthy that flood losses in Germany are generally divided into the sectors private households, industry and commerce, agriculture and forestry, state and municipal infrastructure as well as costs for emergency services in loss statements of the Federal Government and the federal states. To some extent, losses to cultural 25 facilities, sport and recreational centres, churches and research institutions are also provided. Unfortunately not all the sectors are systematically dealt with in every flood event and state, and the definitions, which damage should be reported in which category, are not managed uniformly over space and time. Moreover, in the case of changes to the overall loss estimates, the distribution among the sectors is often not updated. This considerably hinders a comparison of the overall financial or economic losses of different flood events and in different federal states . Of all losses, the insurance industry in Germany has covered around EUR 1.65 billion (GDV, 2015). With EUR 900 million, the most insured damages occurred in 5 the Freestate of Saxony, followed by Saxony-Anhalt (EUR 310 million), Bavaria (EUR 270 million) and Thuringia (EUR 140 million; GDV, 2014). 142 major claimsthis equals to individual claims exceeding EUR 500 000 -were reported to the Association of the German Insurance Industry (GDV) with an overall damage total of EUR 257 million (GDV, 2015).
In addition, the Federal Government and all federal states launched flood relief funds containing a total amount of EUR 8 billion. The parties agreed that losses of private households can be compensated up to 80 %, whereas repair costs for damaged state and municipal infrastructure can be covered up to 100 %. Further, private donations of EUR 108 million have been available (BMF, 2013). Altogether, the funds available for 15 reconstruction excel the total damage. Therefore, more reliable methods for first and immediate damage estimates are required. In order to evaluate the reasonability of first loss estimates reported by the federal states to the Federal Government, not only the estimation methods applied should be documented, but the numbers of physically damaged (or destroyed) items should also be reported by default as is suggested by 20 loss data guidelines (e.g. Corbane et al., 2015). For a first estimate, the number of damaged or destroyed items could be combined with standard repair costs per item. Further, damage indicators should be clearly defined and agreed upon so that the loss documentation of different states and events can be better compared. 25 With the 2013 flood situation, distinctive needs arose for disaster response and appropriate support by personnel and technical resources in the affected federal states. While Baden-Wuerttemberg and Bavaria managed the flood situation predominantly 23 Introduction with their own teams and resources, as well as via bilateral cooperation, states like Saxony, Thuringia and Saxony-Anhalt used the coordination service offered by the German Federal Joint Information and Situation Centre (GMLZ). Aside from support by the federal states not affected, the states worst affected also organised themselves into providing joint support for one another according to the Federal Ministry of the 5 Interior (BMI, 2013). The assistance requests of individual federal states had already been brought to the attention of the GMLZ on 2 June 2013. In total, the GMLZ processed 43 assistance requests from five affected states. Subsequently, around 5.15 million sandbags, 5700 emergency rescue personnel and transport services for 1000 tonnes of material were 10 arranged in the course of the flood situation by 15 June 2013(GMLZ, 2014. Material shortages occurred in the number of available sandbags. To meet the demand of the affected areas, the GMLZ arranged for five million sandbags from other federal states and Germany's European neighbours (BMI, 2013).

Expenses for disaster response and emergency services
In total, 1.7 million voluntary workers are organized in (volunteer) fire brigades, relief 15 and aid organisations as well as the German Agency for Technical Relief (THW). They form the cornerstone of Germany's disaster response. By 5 July 2013, the deployment of local fire brigades and aid organisations added up to around 871 000 person days (GMLZ, 2014; Table 1). Additionally, the Federal Government supported affected municipalities and states with its own resources. In the process, the Ministry  Table 1). Through this, the Federal Government incurred additional costs to the tune of EUR 59.9 million (BMF, 2013). 25 In general, the costs for emergency services and response are included in the overall loss estimates shown in Table 2 the total amount of the other direct damage costs as at July 2015, the response costs of these federal states amount to 1.2, 12.1 and 0.5 % of the direct damage, respectively, which considerably differs from the 2 % of the direct damage that is often used to estimate response costs ex-ante (see Penning-Rowsell and Wilson, 2006;Pfurtscheller and Thieken, 2013). Since publicly accessible data in this domain is 5 scarce despite well-established costing and reporting procedures, explicitly reporting of costs for emergency services and disaster response in loss documentations is highly recommended.

Impacts on economic activities -traffic disruptions
Apart from the direct damage to assets presented in Sect. 3.2, floods can have damage to the transportation system is only considered in terms of physical damage (see Sect. 3.2). Traffic disruption include complete interruption of operations due to route closures as well as restrictions to normal operations on damaged routes, for example: on dual track/carriage routes only one track/lane is usable, low-speed routes or diverted routes are implemented, or the transportation system is replaced by another 20 mode of transport (e.g. railways are replaced by buses). In the longer term, disruption of a particular mode of transport might lead to a loss of customers or a decline in customer satisfaction, for example with railway services. Such effects of flood events are, however, difficult to separate from other influencing factors. Therefore, the analysis focusses on traffic disruptions and interferences. Introduction

Disruption of navigation
If rivers are used as waterways, river reaches will be closed for navigation when a specified water level, i.e. the highest navigable water level, is exceeded at the respective reference gauge. As summarized in Table 3, such water levels were observed in June 2013 at several gauges on different Federal waterways and lasted 5 for 15 consecutive days at maximum. The internationally important waterway at the Lower Rhine was, however, not affected by this flood (BfG, 2014). Disruption of the shipping traffic might last longer than the durations given in Table 3, since the Federal Waterways and Shipping Administration first has to screen for new obstacles in the navigation channels before these can be regularly navigated again. 10 So far, no monetary assessments of the disruption of waterways have been undertaken (BfG, 2014). Related costs are therefore not included in the figures of Table 2.

Disruption of road traffic
The flood event of 2013 led to flooding, dangerous situations and closures of streets in 15 city centres, closures of regional roads and even of a Federal motorway (Autobahn). In total, 700 km of roads and 150 bridges were damaged in Germany (BMF, 2013). These impacts resulted in interferences of road traffic across almost the whole of Germany.
The chronological sequence of traffic obstructions on German roads is illustrated in Fig. 5  More than 75 % of reported traffic obstructions can be traced back to the actual flooding of streets or to flood danger (Fig. 5). In addition to these, landslides especially in Baden-Wuerttemberg (Keller and Atzl, 2014) together with numerous uprooted trees 5 contributed to approximately 20 % of obstructions in road traffic. In more than 60 % of the events, the roads had to be closed completely in both directions. Of the traffic obstructions, 10 % occurred in city centres and on other urban roads. The federal transregional road network was affected by more than 50 % of the traffic obstructions. Figure 6 illustrates the spatial distribution and duration of the traffic obstructions on 10 an administrative district level. With traffic obstructions lasting more than 14 500 h in total, traffic in Saxony was the most curtailed. The Saxon administrative districts of Meissen, Leipzig district, city of Dresden, Saxony's Swiss-East Ore Mountains and Central Saxony were equally affected by a very high incidence of traffic obstructions, as was the Hanover administrative region in Lower Saxony. However, it took only days 15 to remove most of these after the flood had been cleared. In the administrative districts of Traunstein (Bavaria) and Tuebingen (Baden-Wuerttemberg) extensive construction work to damaged roads had to be conducted, which to some extent still affected regional traffic months afterwards. The administrative districts denoted in red in Fig. 6 therefore all display a high overall 20 duration of traffic obstruction. This information does not, however, indicate any decisive conclusions arrived at as to the actual indirect cost due to detours etc. incurred.

Disruption of railway operations
One company that has been considerably affected by the flood event of 2013 is the German Railways Corporation (Deutsche Bahn AG). In June 2013, mudslides as well 25 as the submergence or under-washing of tracks led to a variety of interferences of the normal rail traffic (Fig. 7).  Table 2. 20 Although the EU Floods Directive explicitly addresses the effects flooding has on economic ACTIVITIES, current loss guidelines and reporting emphasise adverse effects on ASSETS. In order to further complement the nationwide data that was presented in Sect. 3.2 and focused on losses to assets, this section looks at the diverse impacts floods can have on individual companies. The data from the survey described 25 in Sect. 2.3.2 were used as basis for the analysis. Flood impacts on companies comprise direct damage to buildings or merchandise, losses due to operational disruptions as well as indirect damage caused by delivery 28 Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | difficulties of suppliers (Fig. 8). Most of the companies surveyed, i.e. 88 %, indicated that they had been affected by operational disruptions (Fig. 8). This led to a similarly large percentage of turnover losses. Thus losses due to business interruption might be equally important than direct asset losses. Except for the amount on insured losses, they are not included in the overall losses given in Table 2 since they are more difficult 5 to assess than repair costs. In general, different methods are available for the estimation of business interruption costs. The most prevalent approaches are (1) to apply a sector-specific reference value per unit affected or per day of interruption to estimate the loss of added value, (2) to compare production output between hazard and non-hazard years, and (3) to calculate 10 production losses as a fixed share of direct damages (Meyer et al., 2013). Since the first approach is the most reliable, the companies surveyed about the 2013 flood were further asked about the period of interruption of operations in their company, as well as how long it took for normal operations without any restrictions to resume afterwards (period of restricted operations). The median value of downtime, through complete 15 interruption of operations or restriction of operations, accounted for two to eight weeks, respectively. In the case of the 2013 flood, there were, however, a number of companies that experienced far longer downtimes through interruption of operations or restrictions of operations: the 75 %-percentile of downtime through the interruption of operations due to the 2013 flood amounts to 60 days, the duration with restrictions of operation 20 to 150 days. The average loss caused by interruption (or restriction) of operations amounted to EUR 137 287 (n = 358; median: EUR 15 000). They only exceeded the losses due to damaged equipment or buildings (see Table 4). Significant differences might, however, occur between different economic sectors, both in terms of downtimes through interruption of operations and in terms of loss share as was shown for the 2002 25 flood by Kreibich et al. (2007).

Impacts on cultural heritage
Although research and data on flood impacts predominantly deal with the impacts on economic assets, the European Floods Directive also addresses effects on cultural heritage. In fact, the flood in August 2002 severely damaged, for example the historic Semper opera house in Dresden (Saxony) and the Garden Kingdom in Dessau-

5
Woerlitz (Saxony-Anhalt) approved as UNESCO world heritage and almost destroyed the flower gardens of the castle Weesenstein at the river Müglitz (Saxony; see DKKV, 2003). Since the repair and reconstruction work of such assets is often very specific and sometimes undoable and since the value that people attribute to such places is beyond financial accounting, this category is treated differently in the European guidance for recording disaster losses: not necessarily financial losses, but lists of damaged cultural, historical and UNESCO world heritage assets are proposed as indicators for loss databases (Corbane et al., 2015). Despite the difficulties of the monetarization of damages to cultural heritage, financial losses related to cultural assets are given in the report of the German government 15 to the European Union Solidarity Fund. The overall amount of initially EUR 56 million (BMF, 2013) accounts for only 1 % of the overall financial losses listed in Table 2. Saxony-Anhalt, Thuringia and Saxony reported losses to cultural assets of more than EUR 10 million each, Bavaria more than EUR 6 million, Baden-Wuerttemberg around EUR 1 million and Schleswig-Holstein EUR 350 000. A detailed list of affected cultural, 20 historic or heritage assets is, however, missing. Thuringia explicitly mentions several damages to parks and gardens, for example the historic Greizer Landscape Park, a cultural place of national importance (Annex 16 in BMF, 2013). Schleswig-Holstein mentions inundation of the historic and listed centre of the city of Lauenburg on the Elbe. Furthermore, it is known that the (historic) city centres of Passau (Bavaria) as 25 well as Grimma and Meißen (Saxony) were flooded. In contrast to 2002, the Garden Kingdom in Dessau-Woerlitz (Saxony-Anhalt) was rarely inundated despite higher water levels due to the meantime betterment of the flood protection. Nevertheless,

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | rising groundwater damaged one castle of this UNESCO world heritage site so that the restoration of the Garden Kingdom is with more than EUR 22 million one of the most expensive projects of the governmental relief funds (Federal Parliament, 2015). This demonstrates the high importance cultural heritage might have in individual cases. 5 Similar to the impacts on cultural heritage, it is difficult to quantify flood impacts on the environment (see Meyer et al., 2013). In fact, some impacts that are adverse at first sight might be ambiguous due to the fact that floods are natural phenomena and ecosystems in floodplains are adapted to flooding. Nevertheless, the environment can be damaged, especially due to inorganic and organic harmful substances that 10 have dissolved or are transported with sediments and floodwater and enter freshwater systems. With regard to contaminants that are deposited in meadows, pastures and agriculturally used lands there is the risk of organisms absorbing them so that the contamination may sustain in food chains. Therefore, pollution must be regarded as the main indicator of adverse environmental flood impacts. Flooded protected ecosystem 15 habitats and formation of new water bodies are further items that are considered by the European guidance for recording disaster losses (Corbane et al., 2015). Adverse environmental effects might also occur, if floodplains or flood retention areas have not been used in a flood-adapted manner. For example, in August 2002, the intended flooding of the Havelpolders at the confluence of the rivers Elbe and Havel 20 caused widespread fish deaths. The flooding submerged the agriculturally used areas and caused the sensitive crops, i.e. maize, to die off. The ensuing decay processes lowered the oxygen content in the water to such an extent that fish could no longer survive (DKKV, 2003). In June 2013, 430 000 ha of agricultural land was flooded (BMF, 2013), but according to the State Office of Nature Conservation and Landscape 25 Management in the Free State of Saxony no fish deaths were noticed (LFULG, 2013).

Environmental impacts
In the framework of river monitoring programmes, sediment load and water quality are frequently measured. Measurements are augmented during and after (extreme) 31 Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | flooding, particularly along the middle reach of the river Elbe due to past mining and industrial activities, particularly in the catchment of the river Mulde (Böhme et al., 2005;BfG, 2014). The measurement programme along the middle reach of the river Elbe provides a wealth of data on the quality of sediments, suspended matter and floodwater (BfG, 2014).

5
In June 2013, an increased sediment load was observed in all main rivers, i.e. Rhine, Danube, Elbe and Weser (BfG, 2014). In the rivers Rhine and Weser, the total load during the flood each amounted to about 20 % of the average annual load. At many gauges on the rivers Elbe and Danube, even higher loads were measured with a maximum of two thirds of the average annual load at the river Danube and even 80 % of the average annual load at a spot on the river Elbe (BfG, 2014).
In a few samples of water and suspended matter, increased concentrations of heavy metals and arsenic were detected, most probably originating from the Ore Mountains (Erzgebirge). The loads of heavy metals amounted occasionally to more than 100 % of the annual load in 2012, reaching a maximum at the Magdeburg gauge. The loads 15 were, however, comparable to those during former flood events, i.e. in 2002(BfG, 2014. In the suspended matter, greatly enhanced amounts of organic pollutants such as hexachlorocyclohexane (HCH) and derivatives of DDT (1,1,1-trichloro-2,2-bis-(pchlorophenyl)ethane), i.e. DDD and DDE, were measured here and there (BfG, 2014). 20 These pesticides had been produced in chemical plants in Bitterfeld-Wolfen until 1973 (DDT) and 1982 (HCH) and process wastes had been dumped nearby in abandoned open pit mines causing severe pollution of soil and groundwater (e.g. Thieken, 2001;Böhme et al., 2005).
In water bodies in Saxony, aggravating pollution was not experienced in June 2013; 25 all the samples inspected were not toxic (LFULG, 2013). In contrast to this, the Bavarian State Office for the Environment (LfU, 2014) reported a high incidence of contamination by heating oil, especially in the area affected by the breach of the embankment at Deggendorf-Fischerdorf (see Fig. 1c). Leaking heating oil from damaged tanks has repeatedly been observed during flood events in Germany and had already been identified as a major source for environmental damage during the Whitsun Flood of 1999 in Bavaria. As a consequence, one-off mandatory testing was introduced at that time for heating oil storage facilities with storage volumes of 1000-10 000 L located in flood-prone areas; a measure which came into effect on 1 January 2001 (LfU, 2014). 5 It is noteworthy that oil leakage and contamination not only harms the environment, it also aggravates damage of flooded buildings considerably (see Kreibich et al., 2005;Thieken et al., 2005). In the administrative district of Deggendorf, up to 150 buildings have to be destroyed and newly erected because of oil contamination (Bavarian Parliament, 2014). Since cost-effective and efficient technical fail-safety systems exist 10 that counteract the floating of oil tanks (e.g. Kreibich et al., 2011), homeowners should be better informed about them. Since 2005, the Federal Water Act states that homeowners are obliged to mitigate damage according to their means. Additionally the implementation of fail-safe measures ought to be monitored more consistently by public authorities (LfU, 2014). the United Nations (UN). In the SFDRR, seven targets to be achieved by 2030 are listed, among others a substantial reduction of (1) (global) disaster mortality, (2) the number of affected people, (3) direct economic losses as well as (4) damage to critical infrastructure and disruption of basic services such as health and educational facilities (UN-ISDR, 2015). Apart from these targets, four priority areas for action are defined, 5 in which systematically recorded, evaluated, shared and publicly accessible loss data play a vital role to understand and consequently mitigate the impacts of such events (UN-ISDR, 2015).
In case of the flood event in June 2013, data and information on the flood impacts could be presented for all four domains considered as relevant by the European Floods 10 Directive and the Sendai Framework SFDRR. The range of impacts portrayed (from direct to indirect damage caused by operational and traffic interruptions through to health and environmental effects) does convey an impression of the diversity of the impacts that flood events can have at different scales. The used data sets have, however, some strengths and weaknesses. It should be noted that the availability 15 of survey data and traffic disruptions is rather an exception than a rule and does require high efforts and resources for data collection and data processing that may not be available for many events. Such data do, however, provide detailed insights into impacts on the property scale, their perception etc. The analysis shows that affected residents perceive psychological stress, reinstatement works and supply problems 20 more seriously than damage to buildings or household contents. With regard to economic activities, traffic and business disruptions are more widespread than damage to economic assets or infrastructure elements. The costs attached to these impacts are, however, currently not assessed and hence not included in the overall damage figures. Therefore, more efforts are needed to include such impacts in loss documentation -by 25 indicators or in monetary terms.
On the contrary, rough data and information from governmental reports, media articles etc. are generally available for many events. For the flood of 2013, there is a clear emphasis of the national and regional reports on the cost assessment of damaged assets for an application to the European Union Solidarity Fund and for the creation of a national reconstruction fund. With regard to human losses common loss indicators such as the number of dead, missing, injured, and directly affected (or exposed) people are only entirely reported on the national level. In the reports of the subnational levels to the Federal Ministry of Finance (BMF, 2013), many specifications 5 are lacking. It is obvious that human losses are not in the focus of an application to the European Union Solidarity Fund. This lack of information is, however, contrasted by the importance that this damage type is given in European and international agreements and that affected residents attribute to physical and particularly mental health problems caused by the flooding. A more comprehensive documentation of human losses 10 together with information on their contexts would be helpful to prevent such losses in future. This was partly undertaken by GMLZ (2014). Most of this information is, however, not publicly accessible as is requested by the Sendai Framework (SFDRR). Damage to economic activities and particularly to economic assets is the domain for which the most information is available, also on a subnational level since this 15 was the focus of the report by BMF (2013). However, the annexes to BMF (2013) as well as Tables 1 and 2 illustrate that the reports of the affected states to the Federal government differ in length, content and comprehensiveness. Some of the differences can be explained by the different relevance the flood had for the respective state, some by previously experienced flooding or a lack of experience with event documentation. 20 In the future, economic indicators should by default be accompanied by information on the number of damaged or destroyed items (physical damage indicators), such as damaged buildings, enterprises, schools and health facilities as proposed by Corbane et al. (2015). On the one hand, such information will reveal further impacts on the affected population (e.g. supply problems, access to health facilities). On the other 25 hand, it will allow a better comparison and evaluation of the quality of the financial loss estimates reported by the states. The reasonability of first estimates could roughly be appraised by multiplying the numbers of damaged or destroyed item by an average loss per item. Reasonable average losses could be derived from the survey data used in this paper. In addition, such an approach could help to balance different experiences with event documentation between states. It is of course not applicable to complex damages. Furthermore, economic sectors should be clearly defined and agreed upon so that the loss documentation of different states and for different events can be better 5 compared. Finally, costs for emergency services and disaster response should be explicitly reported since costing and reporting procedures have been well established in civil protection. The loss reports on the 2013 floods suggest that this potential has not been fully exploited. Only some federal states explicitly provided costs of emergency services of the flood of 2013, others included them to infrastructure losses. This 10 practise should be avoided since in the scientific literature these costs are sometimes regarded as indirect costs (e.g. van der Veen et al., 2003).
Further indirect costs due to traffic or business interruption are currently not included in the overall losses, but might be substantial. Our analysis reveals that traffic disruptions were widespread in 2013 and lasted partly for several weeks and even 15 months. The descriptive assessment of the information gathered on traffic disruptions and interferences illustrate the consequences of the flood without going into the further effects this had on travel times, cancellations of trips or the monetisation of these impacts. The obstruction of shipping, road and railway traffic that is portrayed here was not taken into account in the loss specifications of the states that was presented 20 in Sect. 3.2 and therefore complements the description of flood impacts on (economic) activities. Still, more efforts are needed to derive financial losses of such impacts.
The survey among flood-affected companies further reveals that disruption of production processes and other operations is the most frequently reported flood impact. Since methods to estimate the costs attached to this are in their infancies, this 25 domain requires more attention in research. Data collected on the scale of individual companies can help to derive more reliable estimation models.
In June 2013, damage to cultural assets and heritage accounted only for a small share of the overall financial losses. In consistency with the guideline of Corbane  (2015), damaged historic, cultural and heritage places should be explicitly listed together with the scale of their importance, i.e. for the regional, national or international heritage. With regard to environmental impacts, many measurements of sediment loads as well as of water and sediment quality are available. In 2013, a particular monitoring 5 programme was launched at the middle reaches of the river Elbe due to past mining and industrial activities in the catchment of the river Mulde. In order to better evaluate these measurements, indicators should be developed which also assess the consequences of such contaminations. It is striking that environmental impacts were only addressed in the reports of the water authorities, although contamination by leaking oil tanks is a frequently observed and important driver for building damage (DKKV, 2015). In Germany, the number of floating and leaking oil tanks could thus serve as an important indicator for environmental damage. Since cost-effective and efficient technical fail-safety systems exist that counteract the floating of oil tanks (e.g. Kreibich et al., 2011), homeowners should be better informed about them. Their obligation 15 to mitigate loss should be emphasized. Additionally, the implementation of fail-safe measures ought to be monitored more consistently by public authorities (LfU, 2014).
Altogether, it has to be concluded that the information provided in governmental reports from Germany hardly meet the requirements of European (Corbane et al., 2015) or international (IRDR, 2015) guidelines for disaster loss documentation 20 and databases. Tables 1 and 2 illustrate that more efforts are needed to reach comprehensive loss documentations that are also required for reporting on the progress of the implementation of the SFDRR. Present data and information on flood impacts in Germany appear to be fragmentary, incomplete, partly still preliminary and more often than not publicly inaccessible even for an extreme event such as the flood 25 in June 2013. Since floods are the second most damaging natural hazard in Germany and insurance penetration is still low, transparent and systematic reporting procedures of flood impacts and a related database should be developed. As a minimum effort, a template should be created that is not only usable for applications to the European Union Solidarity Fund but also fulfils minimum requirements of Corbane et al. (2015) and the SFDRR. This template should be generated and agreed upon before the next flood happens and should be accompanied by more robust methods and procedures for first loss estimations. 5 Ideally, such efforts should be embedded in a broader risk management context in order to not only monitor, but to reduce losses in the longer term. Investment decisions on risk reduction should be combined with an integrated risk management and their effects should be monitored and evaluated. Therefore, an information system on flood impacts and costs should ideally include all relevant cost categories including 10 costs for response and prevention (see Meyer et al., 2013;Kreibich et al., 2014). Hazard information should be clearly linked to data on damage and losses, preferably on an event basis with sub-national spatial resolution. Data collection and provision should be established as a continuous task and enforced by national legislation as (potential) data providers are often non-governmental entities, e.g. with regard to 15 infrastructures. To ensure quality, data collection should be based on transparent rules and methodologies. The set-up of such a system can be done stepwise, but data gaps should be closed gradually. Needed research efforts should be systematically identified and funded and good/best practise examples should be studied and maintained. Only then, event impacts and the effectiveness of the risk reduction measures in place can 20 be reliably evaluated.

Conclusions
At present, a lack of adequate cost assessment approaches and data on flood impacts limits our knowledge and understanding of appropriate prevention and risk management measures. In comparison to other scientific fields related to the hydrologic 25 system, impact data are still scarce and methods on assessing losses and damage are in their infancies. Therefore, this paper explored what data is available to describe and NHESSD doi: 10.5194/nhess-2015-324 The flood of June 2013 in Germany: how much do we know about its impacts?
A. H. Thieken et al. quantify the impacts of the flood in June 2013, which was the most widespread flooding Germany witnessed over at least the past 60 years (Merz et al., 2014). The analysis shows that information about impacts in all four domains that are addressed by the European Floods Directive, i.e. human health, economic activities (and assets), cultural heritage, and the environment, is available, but considerably differs in detailedness, completeness and accuracy. The analysis further reveals that drawing up a balance sheet for the impacts of the event in June 2013 has not yet been completed in its entirety. It is further evident that the information currently available does not meet the standards for loss documentation that were proposed by Corbane et al. (2015) for member states of the European Union. Therefore, the establishment of national (and regional) disaster-related accounting systems should be further encouraged. In such an information system, all relevant cost categories including expenditures for risk reduction and response should be included. Furthermore, the system should enable a linkage of flood event indicators with (various) impact indicators in order to evaluate the success of (flood) risk management strategies and measures on the long run. Such an evaluation is required, for example, in progress reports on the Sendai Framework for Disaster Risk Reduction 2015-2030 (SFDRR) that was agreed in Sendai, Japan in March 2015. Only accurate, consistent and comparable databases will allow Germany to substantially and seriously contribute to these internationally agreed targets and commitments. 20 The range of damages portrayed (from direct to indirect damages from operational and traffic interruptions through to health and environmental effects) conveys an impression of the diversity of the impact that flood events can have. Data collected on the scale of individual properties reveal that business disruption is the most frequently reported damage by affected companies and mental health issues as well as supply 25 problems are perceived more seriously by affected residents than building damage or other forms of financial damage. These damage types receive, however, only little attention in governmental reports on the flood of 2013 as well as in research. However, in the case of evaluating and accepting preventive and protective strategies, these can NHESSD Risk Reduction Technology (CEDIM) in Potsdam and Karlsruhe and partly in the framework of the project "Hochwasser 2013" funded by the German Ministry of Education and Research (BMBF; funding contracts 13N13016 and 13N13017). Data provision by all ministries and organisations mentioned in the paper is gratefully acknowledged. We further acknowledge the support of Deutsche Forschungsgemeinschaft (German Research Foundation) and Open