Abstract
Blackouts aggravate the situation during an extreme river-flood event by affecting residents and visitors of an urban area. But also rescue services, fire brigades and basic urban infrastructure such as hospitals have to operate under suboptimal conditions. This paper aims to demonstrate how affected people, critical infrastructure, such as electricity, roads and civil protection infrastructure are intertwined during a flood event, and how this can be analysed in a spatially explicit way. The city of Cologne (Germany) is used as a case study since it is river-flood prone and thousands of people had been affected in the floods in 1993 and 1995. Components of vulnerability and resilience assessments are selected with a focus of analysing exposure to floods, and five steps of analysis are demonstrated using a geographic information system. Data derived by airborne and spaceborne earth observation to capture flood extent and demographic data are combined with place-based information about location and distance of objects. The results illustrate that even fire brigade stations, hospitals and refugee shelters are within the flood scenario area. Methodologically, the paper shows how criticality of infrastructure can be analysed and how static vulnerability assessments can be improved by adding routing calculations. Fire brigades can use this information to improve planning on how to access hospitals and shelters under flooded road conditions.
Similar content being viewed by others
1 Introduction conceptual problem
1.1 Relevance of the topic
Floods impact society, but additional cascading effects of infrastructure failure such as blackouts aggravate the situation. It is therefore paramount to identify the possible exposure of a city to flood (or storm/earthquake) impacts, and more specifically, to impacts of critical infrastructure (CI) failure (US Government 1996; European Commission 2008). Water, electricity, food and information supply and many other sectors, goods and services are termed CI, often in relation to technical failures, attacks, but also to natural hazards.
The recent Sendai Framework for Disaster Risk Reduction (SFDRR) (United Nations 2015) calls for such risk assessments and urges for more integration of infrastructure effects, while programmes on the resilience of cities are already state of the art and advanced by the United Nations (UNISDR 2012) and private foundations (for example, the Rockefeller Foundation programme on 100 resilient cities). In a civil protection context, research increases also with a focus on urban areas and critical infrastructure. Urban areas and infrastructure contribute to a large extent to overall insured disaster losses (Munich 2015). Riverine areas with high concentrations of human population are especially prone to disastrous losses, and modern technology such as remote sensing can help identify such risk hot spots (Kuenzer et al. 2015; Aubrecht et al. 2015; Taubenböck and Strunz 2013). Flood risk assessments have been advanced in terms of critical assessments on their usefulness as well as on limitations (Ward et al. 2015). Using earth observation, usage of radar data has improved usability also under clouded conditions (Kussul et al. 2011). Reviews of flood risk assessments also consider different scale effects (de Moel et al. 2015).
1.2 Challenges of spatial assessments
Identifying CI elements or human usage of CI services from space remains a challenge. It is complex and difficult to get a holistic picture of land cover in space and time and especially of critical infrastructure. Earth observation data are one means to achieve such spatial overviews and offer many advantages for natural hazard detection and related vulnerabilities (Taubenböck and Strunz 2013). The specifics and also the limitations of earth observation and other spatial data and mapping methods are also well documented (Taubenböck and Geiß 2014); a combination with other data sources such as field observations, interviews and statistical data is necessary (de Sherbinin 2014).
In order to address these challenges, we accordingly combine spatial and non-spatial data for a more holistic and updated overview and realisation of the social and CI vulnerability of the area of Cologne. Certain elements or buildings such as power plants or refineries, water bodies for cooling can be extracted from earth observation data such as satellite imagery quite well (Wurm et al. 2009). Road detection and other line elements such as transmission lines work quite well in areas with no or low vegetation cover. Point features or small objects such as power poles can be detected using also spectral information about heat or shadowing. But many areal, linear and point features of CI are difficult to map by utilising remote sensing data alone. Altogether, this provides only a limited risk picture of a city and therefore needs to be amended by additional on-the-ground mapping and data collection.
In water systems, as one example of CI, it is possible to identify rivers from space or from airborne devices, sewage plants, but not the pipes running sub-surface. Again, it is not only remote sensing imagery that allows for risk assessment of water supply systems or other infrastructure, but it is additional information that can be integrated into a GIS (Taubenböck and Strunz 2013). This information can be gathered from the CI operators when they are willing to share, which often is a security concern and issue, especially in times of terror attacks or private-sector competition (Fekete et al. 2012). Another possible source is city planning, which could be asked to provide plans with the pipes, lines and stations. In some countries, open data policies help to include such information and its recent updates (Stevens 2007). However, there are few countries that allow for this. And these data still do not reveal directly the users and consumptions of services by the people. Supply and consumption data just as other risk and vulnerability information are often aggregated at a very coarse level (Welle and Birkmann 2015).
1.3 Advancing mapping by integrating several data types provided by earth observation
At the case of the city of Cologne, this paper shows the options of conducting a criticality assessment of CI objects and how this can be linked with traditional spatial vulnerability and risk assessments. The approach aims to address how large-scale spatial information such as remote sensing can be disaggregated, combined and enriched by case studies at city quarter level using local statistics, maps, specifically, open-access products and quantitative reports of car accidents, for example. Remote sensing data in a strict sense are used mainly in the form of providing the natural hazard flood layers. But many other points of interests, line and polygon features such as land use classifications are processed data originally derived from remote sensing data. From the perspective of civil protection and search and rescue teams on the ground, local data and overview maps need to be combined with positioning data (Stevens 2007). Furthermore, since the spatial data infrastructure (such as GPS) is now in place that enables individuals to position themselves and navigate to a chosen destination by multiple routes, identifying nearby places and services of interest, lives can be saved as a result of emergency response services reaching the right destination in a shorter time (Rajabifard and Coleman 2012). Routing has not only become indispensable as a key infrastructure service for navigating fire and rescue cars (Fekete et al. 2015), and also, utilising social media for deriving crowdsourced up-to-date information about actual damages, flood extent, etc. (Fohringer et al. 2015), is also remote sensing-related services that play an increasing role in civil protection and related planning (Taubenböck and Esch 2011) and allow for even better advancement of the applicability of satellite and airborne data on the ground.
1.4 Research question
The guiding general research question in this paper is: How can risk and resilience assessments at city quarter level be operationalised and advanced when integrating the critical infrastructure topic into traditional spatial risk assessments?
In order to address this broad general question, the paper will narrow down the scope possible to cover in one paper to the following set of sub-questions:
-
Which critical infrastructure objects (elements) are specifically relevant for supplying vulnerable populations from an emergency management or civil protection perspective?
-
Which of these objects can be mapped at city quarter level?
-
What aspects of exposure, vulnerability and resilience can be analysed in a flood scenario?
-
How can existing assessments be advanced regarding objects and their interdependencies?
Generally the paper aims to advance our understanding, from the prevailing focus of recent studies on ‘what can be damaged’ to ‘how can it operate best under suboptimal conditions’ at a minimal standard basis in order to restore capacities. This study will narrow down the focus from all possible CI sectors, branches and object classes (USA, German, etc.) to those CI primarily used by emergency managers in a crisis. It will also make a suggestion for a novel perspective by a focus on certain especially important shelters that should be supplied when restoration of power supply of the whole city after a power cut is not feasible immediately. From the whole suite of vulnerability and resilience features in existing frameworks, this study will select only very few in order to demonstrate feasibility of mapping and opportunities by GIS to advance the state of the art of static parameters and static situation mapping. Limitations of this approach will be discussed within the results.
2 Conceptual background and aim
2.1 Conceptual understanding of disaster risk reduction (DRR) and critical infrastructure protection (CIP)
From a disaster risk analyst perspective, the approach often is identifying the areas of highest risk and the most vulnerable elements or subjects (Birkmann 2013). In the field of critical infrastructure protection or resilience, this approach often faces problems in terms of data accessibility and data sensitivity (Moteff 2005; Fekete et al. 2015). Both fields, DRR and CIP, begin to merge, and international expression of the importance of integrating infrastructure into holistic disaster risk assessments and resilience programmes has been pronounced with the novel Sendai framework SFDRR (United Nations 2015). However, conceptual differences remain, and certain standard assessment paths in DRR are not easily applicable in CIP and vice versa. For example, explicit place-based (also termed spatial) vulnerability assessments (Cutter et al. 2008) may capture exposure of CI elements of certain types of CI, but this often does not allow an easy access to information regarding the technical susceptibility of the elements or the extent and composition of population affected by a power blackout, for example.
We argue that conceptual advancements are necessary in both DRR and CIP and that both can benefit greatly from each other. This is especially the case when this field is analysed from a civil protection perspective. This perspective merges the level of the technical infrastructure with the level of people affected and plays a negotiating role between both levels and stakeholders related to them.
These conceptual demands for advancement and extension of the existing view on CI include standard perceptions of what in DRR is termed lifelines, physical or structural infrastructure, or built environment sometimes. This extends the CI view on solely the technical elements and processes to the environment and land area. And according to the perspective of ecosystem services (MEA 2003), CI can also be regarded as key services they provide to people, ecosystems, economy, etc., rather than being just physical structures per se that are damaged or not. But also people in terms of staff operating and maintaining the infrastructure must be included as much as the customers and users, but also those unaware of being affected by a CI impairment or interruption. The hallmark of CI is interdependencies and cascading effects (Rinaldi et al. 2001); for example, a blackout also affects traffic lights and even non-residents of a certain region are affected, while not being aware they are passers-by utilising local electricity. At least, it is common that this dependency is not noticeable as long as systems function, and this is the case most of the time, which influences risk perception.
DRR, on the other hand, greatly benefits from CIP concepts and perspectives when it comes to the interrelation and interdependency between natural and man-made hazards and vulnerabilities. Particularly, DRR but also CCA often excluded certain man-made risks while promoting the human role in creating disasters themselves quite strongly in the past decades since the International Decade for Natural Disaster Reduction by the United Nations. The holistic approach and all-hazard approach in CIP are one addition that fits to DRR perspectives. Another is the benefit to be gained by interdependency analysis, and the advanced state of many CIP strategies, including not only a risk analysis phase, but also management and strategy aspects to be considered.
3 Model to identify exposure and dependencies between people, emergency management and critical infrastructure
The overall approach follows largely a deductive approach, where certain parameters from existing vulnerability and resilience frameworks are selected and applied in the case study. Similarly, certain CI objects and the overall concept of minimum supply centres are a deductive approach. The following chart provides an overview on the overall conceptual model and assessment steps in this paper. The main aspect of the methodology is a conceptual approach on vulnerability, resilience and critical infrastructure, showcasing how this rather abstract conceptual model can be applied at city quarter level using publicly available spatial and demographic data (Fig. 1).
3.1 Vulnerability and resilience model used
This paper adopts the general idea of differentiating disaster risk indices by hazard and vulnerability components (Davidson and Shah 1997), being aware of the ambiguity of vulnerability with coping capacities (Anderson and Woodrow 1998) and exposure dimensions, with a consensus seeing to establish on its main components (Adger 2006). The integration of resilience is more complicated, and this paper adopts the perspective of resilience as a sub-component of the overall approach. The current study acknowledges the vagueness of both resilience and vulnerability/risk being possible overall terms for the assessment (Cutter et al. 2008). At the moment, resilience has achieved a utilisation of being the overall term, while vulnerability more often is used in an applied and methodological way (Fekete et al. 2014). However, in order to merge perspectives and to apply resilience, in this paper we do not adopt resilience as the overall term, but we rather aim to answer the question, how resilience can be measured, too. Specifically, for this purpose, we adopt the resilience components in relation to CI from both an engineering and sociological perspective by a group of experts (Bruneau et al. 2003), which enables to fit it with vulnerability and risk components, suggested by another group of experts in the same year to make it consistent (Turner et al. 2003).
From both frameworks, we use the following (conceptual) components and indicate in brackets which of them we actually apply in this paper:
Vulnerability:
-
Exposure (exposure zone).
-
Susceptibility (special needs of shelter and health services).
-
Capacities (fire brigades and rescue services).
Resilience (4 R model):
-
Robustness (not considered here).
-
Redundancies (e.g. number of alternative hospitals, fire stations and routings).
-
Rapidity of repair (routing and accessibility).
-
Resourcefulness (not considered here).
Generic criticality assessment criteria useful for GIS assessments (Fekete 2011):
-
Critical amount (of elements such as number of hospitals flooded).
-
Critical timing (of flood phase, arrival times of fire cars; day and night traffic).
-
Critical quality (of service offered of emergency supply or modal change from car to boat).
3.2 Critical infrastructure types
Regarding critical infrastructure, in this paper we adopt traditional assessment perspectives of risk and criticality of infrastructure elements and processes of typical CI sectors and branches as depicted in national CI lists (for example, FMIG—Federal Ministry of the Interior of Germany 2009; US DHS 2013). However, in order to broaden the perspective and merge it with natural hazards and sustainability perspectives, the paper regards not only nodes and vectors in a network, but places the network topology and objects within their real spatial context on a map. In addition to technical and physical elements visible on a map, the current study includes place and environmental features, but also human beings both as staff operating and maintaining infrastructure services and people affected by CI impairment or failure. This is a methodological prototype to approach this complex intertwined situation, and the further steps in this paper show a way on how to put this into application (Tables 1, 2).
General element types of CI (Fekete 2011):
-
Technical elements.
-
Place/environment.
-
Staff.
For minimising the amount of CI sectors, branches and elements, only some typical infrastructure elements are selected. More specifically, the typical infrastructure elements that are selected are relevant to civil protection and emergency management purposes, in relation to the minimum supply scenario of a city-wide power cut lasting for several days. The categorisation of the infrastructure elements in different categories is as follows:
Infrastructure types and order.
-
Category A everyday basic needs supply infrastructure (water, food, energy).
-
Category B maintenance and service supply for Category A (fire stations).
-
Category C infrastructure enabling A and B (roads, electricity).
-
Category D shelters (hospitals, schools).
Please note This sequence is necessary for the model in order to separate different cascading dependencies between the infrastructures.
From this perspective, certain CI elements are Category A types, such as hospitals and schools, that are main infrastructures to provide shelter and supply for the people in the city. They are in the first category, since they represent the persons that should be taken care of and should be the prime target of a civil protection risk management concept (Fekete et al. 2012). Category B types of infrastructure encompasses infrastructure that helps to provide vital emergency aid to category A. Fire stations are selected since they are also responsible for rescue operations and civil protection at a city level in the special case of Cologne. Category C contains infrastructure which is the underlying basis for functioning of key processes within hospitals and fire stations (such as electricity) and enables accessibility and mobility (such as roads). These infrastructure components are at the same time critical for a certain supply service or human group, but at the same time are part of the overall vulnerability as well as resilience of a supply area such as a city or, of a human community.
The selected scenario consists of an extreme flood event (HQ500) in most predicted areas and a flood defence breach lasting 4 days. It has been selected, since it resembles a severe disaster scenario, a non-normal situation (above the flood defence preparedness level in Cologne), with degraded system functionalities and an additional blackout risk existing in the flood-exposed zone.
3.3 GIS methodology and data used
Based on the previously described guiding conceptual model of risk, vulnerability, resilience and critical infrastructure components, the following methods, tools and data sets are used.
Data sources used for the hazard zonation include an extreme flood scenario that includes an over 200-year flood return period river discharge with additional exceedance on top and including the assumption of dyke breaches and failure of flood defence in general. These data are obtained from the Rhine Atlas from 2001 (IKSR 2001) and its new online version with updates from 2015 (ICPR—International Commission for the Protection of the Rhine 2015). We use additional data, kindly provided by the city of Cologne, based on their own flood zonation maps, derived from LIDAR data, and openly available. The Cologne scenario is termed as HQ500, hence, a 500-year return period. All return period models undergo changes when additional flood events, such as in 1993 and 1995 and later, modify knowledge and calculations of the return periods. Certain assumptions of extreme flood extent calculations possibly have to be considered false. The estimated flood zone is the calculated maximum zone at any section of the river. In reality, a flood is dynamic and maximum flood wave heights differ largely between the river sections.
Accessing critical infrastructure data is difficult because of data sensitivity issues. This also influences the assessments carried out in this paper, since data privacy or security concerns because of possible saboteurs or terrorists have to be considered. Therefore, exclusively open-access data were used. OpenStreetMap data were used from certain providers such as geofabrik.de and FLOSM.de, but also open data archives from the city of Cologne (offenedaten-koeln.de) and from GIS companies (opendata.arcgis.com). Accuracy and data completeness were checked with other available data such as online maps and reports from the statistical offices of the city. Data for hospitals and their capacities were taken from annual quality reports available online. However, one constraint here is the inconsistency of data sets, since some reports were available for 2015, but many more for 2014 and older. However, we assume this is not a major problem, since hospital capacities largely remain in similar boundaries within just one year of difference. Still, this is a constraint of data availability and room for improvement for future follow-up studies. Shelter data are one of the most sensitive issues, since misuse by attackers from nationalist groups must be considered. In fact, some of the first to publish information about refugee shelters during the current in-migration from civil conflict states, such as Syria, Afghanistan, Iraq and others, were right-wing nationalists. Therefore, the addresses are not disclosed here and the maps generalised in scale. Additional data from Web Mapping Services (WMS) had been used during the process, but not in the maps provided in this paper. The geoinformation system (GIS) used is QGIS, while some analysis also ran on ESRI products.
4 Vulnerability and resilience assessment in the case study in Cologne
4.1 Exposure mapping of affected people at risk in a flood scenario
We have selected the city of Cologne, being a major German city exposed to flood hazard, permitting assessment of a large-scale event.
The first step in many risk mappings is to create an overview of the situation. In Fig. 2a–c, immediately visible is the city of Cologne area, the river flowing through the city and large areas of the city exposed in an extreme river-flood event. The city quarter level has been selected for this study. It represents a compromise between spatial resolution and the capability to provide a spatial overview and draw comparisons. On the other hand, the main purpose of this approach is to demonstrate the principal methodological approach and its feasibility for other wanting to repeat the approach in other cities. For German cities, higher-resolution data of demographics or airborne land use are often not available for free.
Additional information about the vulnerability of population in relation to floods can be captured using demographic and statistical data on number of people potentially affected by a flood; be it flooded property or difficulties in accessing other flooded parts of the city. Population density information improves exposure information and indicates possible hot spots for evacuation or serving people at home with information, emergency power or water bottles. Not all people evacuate even when commanded by officials, so it might be important to consider access routes to them within a flooded situation. One important aspect therefore is not to use risk maps only for knowing where people are exposed, but also, for civil protection authorities that need to know how to plan emergency provisioning of basic services in case of additional power cuts affecting whole parts of the city or individual houses (Fig. 3).
But of course, such demographic and spatial data alone only provide a limited picture of the whole suite of factors that constitute their vulnerability, and additional household surveys must be carried out in order to gain knowledge about individual households and their situation. Since previous aggregated vulnerability indices have been criticised for blurring the picture of real factors (Fekete 2012), the following maps only show individual demographic factors such as age, welfare dependency that might indicate higher or lower social vulnerability (Cutter et al. 2003). Previous studies have assessed and discussed the opportunities and limitations of general demographic variables to explain specific hazard-related vulnerabilities such as river floods (Fekete 2012).
4.2 Criticality of infrastructure
In some risk assessment approaches, critical infrastructure is analysed separately from other risk factors such as buildings, people (GFOCD 2010). For instance, social vulnerability assessments try to analyse the composition of a society or community in how far it is specifically prone to disastrous consequences triggered by a hazard scenario (Cutter et al. 2003; Rufat et al. 2015). CI assessments often focus on the technical susceptibilities of the elements and in their network or process chain interdependency (Rinaldi et al. 2001). While there is not really clarity in differentiating the terms criticality, vulnerability and resilience assessment, some risk frameworks like the ISO 31010:2009 (ISO—International Organization for Standardization 2009) as applied by the German Federal Office of Civil Protection and Disaster Assistance (BMI 2011) or the National Infrastructure Protection Plan of the USA (US DHS 2013) put an assessment step of essential asset identification related to their ‘criticality’ ahead of further ensuing steps of risk, hazard and vulnerability assessments. For the purpose of this paper, the following two examples shall suffice to explain how a criticality assessment can generally be carried out (Fig. 4).
Observing the rail tracks and rail track crossings already shows that the city of Cologne has a higher number of entry points of railway tracks into the city boundary, as compared to Bonn, and also has more alternative routing possibilities to access the city and to reroute rail traffic in case one crossing or connection is interrupted. For the city of Bonn, due to fewer railway lines running through the city, each of the main north–south connecting tracks is already ‘critical’ in terms of lack of redundancies to uptake all volume, if one track is impaired. Both railway tracks are heavily used, since they are main connectors between the industrial Ruhr area in the north and its customers in the south.
The question of criticality of a critical infrastructure facility to a community or city can also be addressed without spatial information. In order to illustrate this, another example is used here. Hospital capacities (Table 3) can be compared using general statistical data from the quality reports of public hospitals in Cologne that take the largest share of patients in daily and emergency situations. There exist a number of private hospitals too, but they are mainly for non-emergency services like psychiatry and other.
It is immediately observable that the hospital with the largest number of patient beds in the first row in Table 3 would put a heavy burden on the patients, the community and emergency management services in case it would have to be evacuated. This can be caused by many forms of events such as outbreak of an epidemic in the hospital, a blackout, flood or other scenario. Also, numbers such as the amount of doctors or annual ambulatory services allow for the hospitals criticality estimation within daily situations and daily emergencies, as well as in a severe crisis situation. While a single hospital listed beneath the largest hospital may not appear immediately critical in comparison, it can also become critical. This may occur when several hospitals are affected by the same event, need to be evacuated, or another major crisis in the city has fire brigades and emergency mangers already working at maximum capacity and additionally if certain very rare and specialised services such as fire burn treatment stations are affected. The table shows that only for two hospitals we found such data (oral information received by expert interview with the medical head of emergency management of city of Cologne, A. Lechleuthner, 3 March 2016). Special services such as fire burn stations are especially critical, and this extends also to a much larger region, since such fire burn stations are very rare, even regarding the entire federal state of North Rhine-Westphalia (Fig. 5).
4.3 Emergency management and civil protection at risk in a flood scenario
As a first result, it comes as a little surprise for those acquainted with the experience of river floods in Cologne that even fire stations are flooded, and quite many hospitals seem exposed and ‘at risk’. But it must be added, this is at an assumed extreme event of a HQ500. Cologne experienced riverine floods of the Rhine River in 1993 and 1995 with quite extensive damage, despite the city being ‘used’ to floods (Engel 1997). Following those events, the city of Cologne established a hallmark example of flood protection and preparedness within Germany; a flood protection centre was created, mobile wall elements stored, pumping stations along the Rhine erected which cost several million Euros, and so far, have not been in real case operation due to lack of significant events, and an active citizen NGO to ensure bottom-up participation. Extensive research has been carried out about diverse aspects of flooding risk (Grünthal et al. 2006), vulnerability assessments for floods including built up area, population, using remote sensing (Wurm et al. 2009a, b) or CI (GFOCD 2010; Birkmann et al. 2008) and in context to heat and other extreme stressors (BBK 2013; Depietri et al. 2013). Still, flash flood hazard and the interdependency of natural hazard risks with CI risks came into focus in recent years. But yet, new built houses in flood-prone areas such as the architecturally prestigious ‘crane houses’ or the fire stations partly located in exposed areas for a given rare extreme scenario raise certain questions we will address partly in this paper.
4.4 Exposure mapping of auxiliary supply infrastructure for enabling emergency management operations
In the next step, we want to demonstrate certain challenges occurring by mapping CI elements. In the previous steps, objects were identified to be within or without the flood scenario zone.
Figure 6 highlights two options for exposure zonation mapping. The blue area is the polygon resulting from connecting the blue dots representing the hypothetically exposed transformer substations. It is hypothetical, since (a) every flood is unique and will not necessarily affect the area designated a possible flood-exposed area. It is also hypothetical since (b) more information is needed about the transformer substations; are they elevated, water-proof, etc. Not having this information at hand, an exposure assessment could designate the blue area as the area where the transformer substations could be flooded and fail. It is therefore also an area of possible impact on the customers (people using electricity) and of maintenance teams from the grid operator who might need to check, protect and repair the transformer substations (TSS).
However, this is all still hypothetical, since the failure of those TSS could result in failure of power supply in the blue area, but it could also be the case that either certain islands within the blue area are still working, or the area affected by blackout is even wider than mapped. From a GIS analyst’ view, it might be arguable to extend the impact area to the orange area, where the blackout reaches the nearest, not flooded TSS. Therefore, this area is also exposed, although not lying within the flood zone area.
This map was shown to a local electricity operator expert (anonymous, pers. com. 17.03.2016) for feedback and a plausibility check on how realistic or flawed this spatial assessment is. They replied that the first problem is data accuracy; the open data (provided by flosm.de) are inaccurate and not complete.
The estimation of both the blue and orange hypothetical blackout zones affecting the residents is, however, not fully wrong in their opinion. Such estimations are not fully correct, but allow for an approximation, and are more likely to be accurate in less densely populated areas outside of the city centre. It does make sense, to provide different scenarios, they recommend.
The operator also expressed interest in further studies analysing modelling approaches also.
The next step will be to enrich the point data of the TSS with additional information to better estimate the exposure and further the susceptibility of the TSS themselves. This can be done by expert interviews with the CI operator, or by adding more information from field observation, photographs of the objects and streets from sources such as Google street view. It is also necessary to add more data on the affected people. These data could contain information about electricity usage during the day (exposure), special needs (sensitivity to blackouts), emergency backup measure (capacities), impact initiation speed of the blackout, time needed for recovery (resilience), etc.
4.5 From static exposure to additional vulnerability information and incorporating the temporal aspects
The above-presented exposure estimation for static objects such as transformer stations may be wrong in several aspects. For instance, there is incomplete information about each object and its real exposure, mainly due to its position on ground or being elevated. Additional technical and organisational susceptibility information is lacking as well. One organisational aspect is service maintenance: For example, when some transformer stations or other elements must be accessed and checked during a flood scenario. In this case, road accessibility in the affected areas is a key parameter to be taken into consideration. This also applies to the supply of hospitals and many other buildings within urban (highly) populated areas. Another key parameter that we take into consideration is the traffic congestion, which can disrupt emergency response. In a crisis situation, it can be considered as impedance in our model of dynamic network routing, which offers solutions: A GIS can be a useful tool for determining emergency vehicle response routing. The application of dynamic variables, like historical traffic count data, can help emergency response vehicles avoid traffic congestion and improve response times (Winn 2014). The paper will focus on one selected example to demonstrate, how dynamic information of routing can be utilised to improve the overall assessment—from being a static exposed object map to one that encompasses additional temporal information. The methodology followed is presented in the flowchart below.
Routing algorithms use a standard of measurement called a metric (i.e. path length, drive time) to determine the optimal route or path to a specified destination. Optimal routes are determined by comparing metrics, and these metrics can differ depending on the design of the routing algorithm used (Parker 2001).
In an overview of Karadimas et al. (2008) of different kind of algorithms for finding the optimal routes, several are proposed such as the simulated annealing (Kirkpatrick et al. 1983), the tabu search (Glover and Laguna 1997), genetic algorithms (Holland 1975), the ant colony optimisation algorithm (Kirkpatrick et al. 1983) and the Dijkstra’s algorithm, used by network analyst. For our study, we have selected the Dijkstra’s algorithm, because it is an algorithm which finds the shortest route from one of the vertexes of the graph to all the others and the values of the optimisation parameters. It is recommended because it works faster and uses memory more efficiently (PyQGIS 2016).
In this paper, the ArcGIS network analyst algorithm is introduced. The ArcGIS network analyst allows to solve common network problems, such as finding the best route across a city, finding the closest emergency vehicle or facility, identifying a service area around a location or facility of interest, servicing a set of orders with a fleet of vehicles or choosing the best facilities to open or close. Out of the six types or solvers which are used (route, closest facility, service areas, OD cost matrix (an origin–destination (OD) cost matrix from multiple origins to multiple destinations), vehicle routing problem, location allocation), we are processing the closest facility for identifying the closest routing to affected facilities (in our study the hospitals). Calculations on geographic accessibility are conducted by identifying the number of fire stations (not affected from the extreme flood) closest to the affected hospitals (in matters of time). We also examine the sensitivity to changes in travel speed settings through the creation of a scaled cost polygon barrier (which is the flood area). It basically allows to examine and compare travel times:
-
The route from the fire stations outside the flood area to the flooded hospitals before incident.
-
The route from the fire stations to the hospitals in the flood area after incident.
A scaled cost polygon barrier doesn’t restrict travel on the edges and junctions it covers; rather, it scales the cost of traversing the covered edges and junctions by a factor specified, such as 3.0 would mean it is expected to take three times longer than normal. At this stage, we do not have empirical data on how to derive this factor more realistically. However, 3.0 is not fully randomly selected, rather, it is based on own observations and is suited to compute a delay time not too much exaggerated. But it is evident that further studies need to be conducted to derive such factors. This type of barrier might be used to model storms that reduce travel speeds in specific regions or to find shortest paths to destinations avoiding areas of high criminality. Lacking in our study is an empirically derived factor of decrease in travel speed on roads flooded to a certain height.
Any kind of change barriers making the traversability or impedance of the network can also be accomplished through edits to the network data set. Yet barriers help to add and remove network changes quickly, which is ideal for modelling temporary impedance changes: For instance, a tree blocking the traffic will eventually be removed, and the flood will ultimately recede.
The map on the left shows the quickest routes, in means of time, that fire fighters of two fire stations should follow in a normal day so as to reach quicker the hospitals that are within the extreme flood zone scenario. The route goes through the city without regard to the effect flooded road conditions have on travel time. The map on the right shows the quickest routes that should be followed from fire fighters of two fire stations to one flooded hospital. On the right, a scaled polygon barrier triples the travel time of the roads covered by the flood. The routes still pass through the flood zone so as the fire fighters to be able to provide their emergency response to the flooded hospitals, but it’s altered indicating different routes that should be followed even if more time of driving will be spent driving slowly through the flood zone. To a next step, the combination of travel time and population exposure data gives an indication of the number of people potentially unable to evacuate in the available time, which facilitates planning of additional evacuation and emergency response solutions (Fraser et al. 2014). Additional useful information that should be taken into consideration is the traffic during specific time periods in a day. An expert from the flood protection centre in Cologne advised us to consider testing such spatial assessments in real situations (M. Willkomm, pers. com. 14.03.2016). M. Willkomm mentioned an event from the actual emergency response on the field. Even if the street was flooded with just 10 cm water height and the fire trucks should be able to drive through, they had to stop and wait until the fire men had examined the street for gully covers afloat due to buoyant force, so that the cars would not be damaged while driving over the open gully openings (Figs. 6, 7).
4.6 Shelters in operation and at risk
The following map contains schools and gyms as possible shelters, which are usually equipped or foreseen for evacuations in Germany. They are even used for this purpose, when bombs from World War II are found and the area has to be temporarily cleared until the bomb has been defused. In Cold War times, many schools had also been equipped with emergency hospital functionality, but this, as the bunkers, has been dismantled in the years past reunification in 1989.
In a future study, we will analyse the routing delay times of fire brigades among to serve the emergency shelters during a flood scenario. At this stage, it is sufficient to show shelters as another point of interest (POI) for disaster management and how this can be assessed as well as hospitals ort many other POI. It is interesting to be able to compare this hypothetical assumption of schools and gyms as a shelter with the recent situation. Some professionals described the current ongoing refugee influx from civil conflict states such as Afghanistan, Iraq, Syria and other countries, as a factual real case test at large-scale unimagined before (Figs. 8, 9).
5 Closing data and analysis gaps: opportunities and constraints of spatial and digital information
Shortcomings identified in our own work include assessments that carry only very rudimentary information about the vulnerability of infrastructure. For instance, in an index of aggregated social vulnerability for river floods at county scale, we added another layer of infrastructure, but all information we could derive by using spatial information such as found in open accessible maps and aerial images were locations of certain objects such as refineries, power plants and similar. Therefore, this information layer barely contained more information than the density of certain CI objects per county area.
Another challenge of matching CI asset assessments with affected population and their vulnerability is that traditional risk zoning is not enough. Risk zoning imposes a spatial area that is affected in this case the flooded transformer stations. However, as this paper has shown, this area does not necessarily resemble the area of affected people. That area of residents and mobile population groups such as visitors, commuters might lie outside of the flooded area, since cascading effects of power failure cascade over larger regions that are difficult to estimate without the expertise of a CI operator company possessing all monitoring information.
Underlying secondary infrastructure is energy, information, logistics, roads and other forms and services that enable other primary infrastructure such as hospitals, civil protection. As an example, electricity is also an underlying infrastructure enabling its own production and distribution, too, for example.
While roads are important for human use in normal situations, in the case of a certain disaster scenario, roads can become vital when people need to be transported to hospitals, or organs and medical supplies or other need to be delivered. Roads are one of the most important infrastructures for providing access to hospitals and any other kind of shelter. Research on roads and transport adopted vulnerability and criticality assessments was early taken on and applied in network analyses also with geographic context (Jenelius et al. 2006). Furthermore, Jenelius in 2010 studied the vulnerability of the road network under area-covering disruptions (such as flooding, heavy snowfall or forest fires) and found that in contrast to single link failures, the impacts of this kind of events are largely determined by the population concentration, more precisely the travel demand within, in and out of the disrupted area itself, while the density of the road network is of small influence. Additionally, spatial risk assessments address road and route interruptions due to extreme weather events, for example (Keller and Atzl 2014). Late examples of disasters such as Hurricane Sandy 2012 have also shown the importance also of other transport routes such as when subways were flooded. Meanwhile, big amount of such collected spatial information can be updated and utilised in addition to the conventional authoritative data (Adger 2006). People experiencing disasters may still be able to share messages and locations on social media websites, voluntarily supplying information regarding the affected areas via online social media, collaborative platforms or in situ and mobile sensors (Resch 2013; Albuquerque et al. 2015; Goodchild and Glennon 2010), especially GPS-enabled devices (Middleton et al. 2014) leading to the gathering of the so-called crowdsourced or volunteered geographic information (VGI) (Goodchild 2007; Horita et al. 2013). Such volunteered geographic information (VGI) has a varying quality; it can provide timely updates for estimating the disaster severity (De Longueville et al. 2010; Zook et al. 2013). Specifically in crisis situations, such as after the earthquake of Nepal in 2015, a timely information regarding the road network connectivity (i.e. whether a road segment is still accessible after a disaster or not) is very valuable for the decision makers since this type of information can be commonly observed and is critical for planning rescue routes (Hu and Janowicz 2016).
In Germany, social media use and crowdsourcing came up as a topic only fairly recently with the floods in 2013 (Kern and Zisgen 2014; Fohringer et al. 2015; Herfort et al. 2014), where volunteers organised themselves to help defend the dykes, often with a sceptical perspective of the traditional professional and traditional organised volunteer organisations in Germany. Traditional civil protection agencies and fire brigades alike, since then, carry a discussion on how and whether to integrate such bottom-up information. Analysts have soon picked up possibilities to analyse accuracy of information of people reporting flood heights or damages on the front line utilising ‘Big Data’ opportunities, for example a recent research project named sd-kama (http://www.sd-kama.de).
Regarding the robustness of the hypothetical blackout service failure areas, we have taken a very simplified approach on purpose; the ambition is to generate simple maps for a first-order estimation. In areas of data scarcity, data sensitivity concerns both operators and civil protection officials; such a simplified approach might have its usage for a first estimation. Also for urban planners and emergency managers that are lacking access to precise technical data on current electricity flows and detailed technical and organisational preparedness and backup measures, such first estimation zones are largely missing. Still, we are fully aware about the shortcomings and the possibilities already existing to verify our blackout map on a much more sound basis. For example, while our verification efforts in terms of discussing results with an electricity operator expert are in general valuable, they remain on a less quantitative level. In future research, we will make efforts to use advanced statistical models, which quantify uncertainties and materialise the knowledge of electricity operator expert into likelihoods of correctness. While there exist demonstration studies for such approaches based on voronoi polygons, for example, and service area estimation is a field of study (Pala et al. 2014), we did not have the data nor expertise at hand to conduct it at this stage. Together with the operator, we aim to approach this in the next step.
6 Conclusion
This paper has been driven by the ambition to move beyond the mere conceptual discussions around resilience and vulnerability (Taubenböck and Geiß 2014) and to respond the requests by fellow colleagues, students and readers of our previous, more conceptual papers. This will now provide examples on how to apply those concepts of generic criticality criteria (Fekete 2011) and conduct spatial CI assessments. Certainly, this one paper cannot satisfy all aspects of these requests. But it showcases certain possible assessments steps to be carried out by fellow analysts and addresses certain challenges.
From the whole suite of vulnerability and resilience components, which would have to be analysed for a holistic characterisation of the city, this paper managed only to address a small share. But the main objective of this paper is to show how such an assessment can be carried out in principle and, highlighting the multiple steps necessary to consider while analysing critical infrastructure, applied to emergency management and affected population in their multiple aspects of interconnection (termed escalation steps in Table 2).
The main research question of this paper is ‘How can risk and resilience assessments at city quarter level be operationalised and advanced when integrating the critical infrastructure topic into traditional spatial risk assessments?’ This paper illustrates Cologne city as a case study where vulnerability and resilience elements (Table 1) in general could be used to analyse risk. We operationalised these conceptual criteria by an explicit spatial assessment, in which, due to length of analysis, only few of those are practically demonstrated. However, this advances current understanding of critical infrastructure assessment by adding a spatial dimension often lacking in pure network analyses. Table 4 summarises conceptual components of Table 1 addressed in this paper.
The paper merges existing spatial vulnerability assessments with CI asset analysis and advances current understanding by showing in a five-step model (Table 2):
-
1.
How the assessment must differentiate between affected people’s vulnerability, the criticality of the infrastructure in general and in a flood situation.
-
2.
Which parts of civil protection are itself exposed to flood risk in this situation.
-
3.
How even the shelters would possibly be affected during a flood.
In this sense, the paper answered the first three sub-questions on critical infrastructure objects specifically relevant for supplying vulnerable populations from an emergency management or civil protection perspective. The feasibility of mapping certain objects at city quarter level is demonstrated in the figures, and some typical challenges are addressed which may help fellow analysts. One specific challenge is the difficult estimation of the area affected by a blackout. Mapping individual elements of the electricity transmission system has severe shortcomings, since individual households and their dependence on one or the other electric grid element cannot be derived at this coarse analysis level. Even more, CI such as electricity is renowned for their interdependencies and cascading effects (Rinaldi et al. 2001), and mapping such cascades in a scenario remains a challenge when relying on openly accessible spatial data only. Along with electricity operators, we have consulted what could be feasible from a ‘bird eyes view’ as a spatial analyst, without the exact energy grid system and its actual operating processes and dynamic states of the grid at hand. This remains, at least for us, still a research area to further explore.
For the third research sub-question of this paper, regarding exposure, vulnerability and resilience parameters this paper managed only to address a small share. But the main objective of this paper is to show how such an assessment can be carried out in principle and highlighting the multiple steps necessary to consider while analysing critical infrastructure, emergency management and affected population in their multiple aspects of interconnection (termed escalation steps in Table 2).
Regarding the fourth and last sub-question on how existing assessments can be advanced, this paper put objects of civil protection in focus and showed how city authorities responsible for emergency management, daily search and rescue and fire fighting services must take extreme scenarios into account, when certain underlying infrastructure, such as roads and electricity, is not functioning at 100% reliance as usual. Certain interdependencies have been highlighted, such as accessibility to hospitals impaired by floods and traffic deceleration on flooded but trafficable roads.
7 Outlook: demands for minimum supply centres
In future work as a direct follow-up advancements of the algorithms used for the routing calculations with other algorithms such as the A* search algorithm will be used. And for the blackout area estimation, alternative approaches for service areas will be investigated. There is a wealth of studies existing that estimate the blackout areas based on voronoi polygon modelling or similar approaches (Pala et al. 2014). However, for our purposes, this demands more data and modelling and will become part of a further study. At this stage, it was more important to illustrate a simple methodology that can be based on publicly available data and is easy to generate but also easy to understand which is an important factor for acceptance of such products by planners or emergency practitioners.
As an outlook on work beyond what was presented here, the application of the approach for minimum supply concepts will be analysed. Minimum supply concepts identify suitable shelters or locations of last resort, also termed ‘lighthouses’ (Ohder and Sticher 2013), that should be provisioned with emergency supplies for such rare cases of surprise crises. In the Cold War era shelter concepts existed in the form of bunkers, with extensive protection and air ventilation systems. In Germany, they are built back, since expectable disaster scenarios and also attack methodology and power have changed. Finally, costs for maintenance and emerging new security regulations such an environmental and fire safety standards all led to the end of bunker construction, at least in Germany. As alternatives, in the recent light of blackouts as a hazard of concern, certain projects look at minimum supply centres that serve as last resort, evacuation hot spots. Object types to be considered for such minimum supply centres vary, from hospitals, schools, fire stations, city halls, gyms and so on. Considerations also include gas stations, supermarkets (mini supermarkets in Japan), churches or cars in very remote areas. Recent research has analysed how technical sensors can improve information about diesel storage levels and how shelters could be provisioned in a large city such as the German capital Berlin (Ohder et al. 2014). Shelters and refugee camps both are closely related to the idea of minimum supply centres. Research and practical knowledge on shelter design have a much longer tradition. However, questions such as spatial risk assessments of potential natural hazards such as floods, landslides, earthquakes, but also questions around the possibility of blackouts or needs of backup systems of critical infrastructure are not standard procedures in setting up either shelters or refugee camps yet. Identification of such minimum supply centres requires assessment of basic needs of people and related infrastructure services that serve those needs (security, safety, water, food, heat, information, medical care, gas, etc.). This paper contributes ideas for planning such centres with its focus on exposure as an estimate for safe ground in a crisis situation. The aim is also to indicate the need for more research and consideration of spatial distribution of shelters, their location within hazard zones and their dependence on emergency infrastructure services that might be affected by a flood or blackout themselves. Spatial information and earth observation data can serve to obtain some key data and some novel ways to calculate routing through flooded roads in emergency or disaster situations, as this paper shows. There are still areas unexplored such as navigating through flooded zones or an awareness of flood-exposed fire brigade and shelter stations. While spatial data cannot replace empirical findings on the ground, it offers an excellent option for obtaining spatial overviews and precise georeferenced positions and most importantly, combining such information with empirical, local and non-spatially derived data. Many empirical parameters will be necessary to gather and add to test our assumptions in future research.
References
Adger WN (2006) Vulnerability. Glob Environ Change 16:268–281
Albuquerque JP, Herfort B, Brenning A, Zipf A (2015) A geographic approach for combining social media and authoritative data towards identifying useful information for disaster management. Int J Geogr Inf Sci. doi:10.1080/13658816.2014.996567
Anderson MB, Woodrow PJ (1998) Rising from the ashes: development strategies in times of disaster. Lynne Rienner, Boulder
Aubrecht C, Meier P, Taubenböck H (2015) Speeding up the clock in remote sensing: identifying the ‘black spots’ in exposure dynamics by capitalizing on the full spectrum of joint high spatial and temporal resolution. Nat Hazards. doi:10.1007/s11069-015-1857-9
BBK (2013) Abschätzung der Verwundbarkeit von Bevölkerung und Kritischen Infrastrukturen gegenüber Hitzewellen und Starkregen. Bonn
Birkmann J (2013) Measuring vulnerability to promote disaster-resilient societies: conceptual frameworks and definitions. In: Measuring vulnerability to natural hazards: towards disaster resilient societies, 2nd edn. United Nations University Press, Tokyo, pp 9–54
Birkmann J, Krings S, Renaud F (2008) Assessment of vulnerability to floods at local level with a special focus on human-environmental systems and critical infrastructures. In: Simonovic SP, Bourget PG, Blanchard SF (eds) Proceedings of the Proceedings of the 4th International Symposium on Flood Defence 6–8 May 2008 Managing Flood Risk, Reliability and Vulnerability; 2008: Institute for Catastrophic Loss Reduction, Toronto
Bruneau M, Chang SE, Eguchi RT, Lee GC, O’Rourke TD, Reinhorn AM, Shinozuka M, Tierney K, Wallace WA, von Winterfeld D (2003) A framework to quantitatively assess and enhance the seismic resilience of communities. Earthq Spectra 19:733–752
Cutter SL, Boruff BJ, Shirley WL (2003) Social vulnerability to environmental hazards. Soc Sci Q 84:242–261
Cutter SL, Barnes L, Berry M, Burton C, Evans E, Tate E, Webb J (2008) A place-based model for understanding community resilience. Glob Environ Change 18:598–606
Davidson RA, Shah HC (1997) An urban earthquake disaster risk index, stanford digital repository. In: Blume JA (ed) Earthquake Engineering Center Technical Report 121. http://purl.stanford.edu/zy159jm6182
de Moel H, Jongman B, Kreibich H, Merz B, Penning-Rowsell E, Ward PJ (2015) Flood risk assessments at different spatial scales. Mitig Adapt Strateg Glob Change 20:865–890
de Sherbinin A (2014) Mapping the unmeasurable? Spatial analysis of vulnerability to climate change and climate variability. University of Twente, Enschede
Depietri Y, Welle T, Renaud FG (2013) Social vulnerability assessment of the Cologne urban area (Germany) to heat waves: links to ecosystem services. Int J Disaster Risk Reduct 12(6):98–117
Engel H (1997) The flood events of 1993/1994 and 1995 in the Rhine River basin. In: Leavesley GH (ed) Destructive water: water-caused natural disasters, their abatement and control, vol 239. IAHS, Wallingford
European Commission (2008) COUNCIL DIRECTIVE 2008/114/EC of 8 December 2008 on the Identification and Designation of European Critical Infrastructures and the Assessment of the Need to Improve their Protection, Official Journal of the European Union. 23.12.2008. L 345/75-82. In
Fekete A (2011) Common criteria for the assessment of critical infrastructures. Int J Disaster Risk Sci 2:15–24
Fekete A (2012) Spatial disaster vulnerability and risk assessments: challenges in their quality and acceptance. Nat Hazards 61:1161–1178
Fekete A, Lauwe P, Geier W (2012) Risk management goals and identification of critical infrastructures. Int J Crit Infrastruct 8:336–353
Fekete A, Hufschmidt G, Kruse S (2014) Benefits and challenges of resilience and vulnerability for disaster risk management. Int J Disaster Risk Sci 5:3–20
Fekete A, Tzavella K, Armas I, Binner J, Garschagen M, Giupponi C, Mojtahed V, Pettita M, Schneiderbauer S, Serre D (2015) Critical data source; tool or even infrastructure? Challenges of geographic information systems and remote sensing for disaster risk governance. ISPRS Int J Geo Inf 4(4):1848–1869. doi:10.3390/ijgi4041848
FMIG - Federal Ministry of the Interior of Germany (2009) National Strategy for Critical Infrastructure Protection (CIP Strategy). Berlin
Fohringer J, Dransch D, Kreibich H, Schröter K (2015) Social media as an information source for rapid flood inundation mapping. Nat Hazards Earth Syst Sci 15:2725–2738
Fraser S, Wood N, Johnston D, Leonard G, Greening P, Rossetto T (2014) Variable population exposure and distributed travel speeds in least-cost Tsunami evacuation modelling. Nat Hazards Earth Syst Sci 14:2975–2991. doi:10.5194/nhess-14-2975-2014
GFOCD (2010) Estimation of vulnerability of municipalities towards floods (Abschätzung der Verwundbarkeit gegenüber Hochwasserereignissen auf kommunaler Ebene). Federal Office for Civil Protection and Disaster Assistance, Bonn
Glover F, Laguna M (1997) Tabu search. Kluwer Academic Publishers. http://dl.acm.org/citation.cfm?id=549765
Goodchild MF (2007) Citizens as sensors: the world of volunteered geography. Geojournal 1–12
Goodchild MF, Glennon JA (2010) Crowdsourcing geographic information for disaster response: a research frontier. Int J Digital Earth 3(3):231–241
Grünthal G, Thieken AH, Schwarz J, Radtke KS, Smolka A, Merz B (2006) Comparative risk assessments for the city of Cologne: storms, floods, earthquakes. Nat Hazards 38:21–44
Herfort B, de Albuquerque JP, Schelhorn S-J, Zipf A (2014) Exploring the geographical relations between social media and flood phenomena to improve situational awareness. In: Connecting a digital Europe through location and place. Springer International Publishing, Cham, pp 55–71
Holland JH (1975) Adaptation in natural and artificial systems. The University of Michigan Press, Ann Arbor. 2nd edn. (1992). MIT Press, Boston
Horita FEA, Degrossi LC, Gomes de Assis LF, Zipf A, de Albuquerque JP (2013) The use of volunteered geographic information (VGI) and crowdsourcing in disaster management: a systematic literature review. In: Proceedings of the 19th Americas Conference on Information Systems (AMCIS), p 1
Hu Y, Janowicz K (2016) Prioritizing road network connectivity information for disaster response. Retrieved 2 March 2016, http://geog.ucsb.edu/~jano/paper9_EMGIS
IKSR (2001) Rheinatlas. Koblenz
ISO-International Organization for Standardization (2009) ISO/IEC 31010:2009. Risk management—risk assessment techniques. In: Geneva
Jenelius E, Petersen T, Mattsson L-G (2006) Importance and exposure in road network vulnerability analysis. Transp Res Part A Policy Pract 8(40):537–560
Karadimas NV, Doukas N, Kolokathi M, Defteraiou G (2008) Routing optimization heuristics algorithms for urban solid waste transportation management. WSEAS Trans Comput 7(12):2022–2031
Keller S, Atzl A (2014) Mapping natural hazard impacts on road infrastructure: the extreme precipitation in Baden-Württemberg, Germany, June 2013. Int J Disaster Risk Sci 5:227–241
Kern J, Zisgen J (2014) “I like Hochwasser”. Eine stichprobenhafte Untersuchung der Nutzung von Facebook während des Hochwassers 2013 in Deutschland. BBK Bevölkerungsschutz 3:17–19
Kirkpatrick S, Gelatt CD, Vecchi MP (1983) Optimization by simulated annealing. Science 220(4598):671–680. http://www.jstor.org/stable/1690046. Accessed 22 Mar 2016
Kuenzer C, Klein I, Ullmann T, Georgiou EF, Baumhauer R, Dech S (2015) Remote sensing of river delta inundation: exploiting the potential of coarse spatial resolution, temporally-dense MODIS time series. Remote Sens 7:8516–8542
Kussul N, Shelestov A, Skakun S (2011) Flood monitoring from SAR data. In: Use of satellite and in-situ data to improve sustainability, Springer, Dordrecht, pp 19–29
MEA (2003) Ecosystems and human well-being. A framework for assessment
Middleton SE, Middleton L, Modafferi S (2014) Real-time crisis mapping of natural disasters using social media. IEEE Intell Syst 29(2):9–17
Moteff J (2005) Risk management and critical infrastructure protection: assessing, integrating, and managing threats, vulnerabilities and consequences. Washington DC
Munich RE (2015) Natural catastrophes 2014 Analyses, assessments, positions 2015 issue. Munich
Ohder C, Sticher B (2013) Approaches towards a community-oriented and enabling disaster management model. In: Crisis management—Emergency planning—Civil Protection Festschrift on the occasion of 60 years training in civil protection offered by partners, friends and employees of the Federal Office of Civil Protection and Disaster Assistance. Berlin: Duncker & Humblot
Ohder C, Röpcke J, Sticher B, Geißler S, Schweer B (2014) Relief Needs and Willingness to Help in the Event of Long-term Power Blackout. Results of a citizen survey in three Berlin districts
Pala O, Wilson D, Bent R, Linger S, Arnold J (2014) Accuracy of Service Area Estimation Methods Used for Critical Infrastructure Recovery. In: Critical Infrastructure Protection VIII: 8th IFIP WG 1110 International Conference, ICCIP 2014, Arlington, VA, USA, March 17–19, 2014, Revised Selected Papers. Berlin, Heidelberg: Springer Berlin Heidelberg. pp 173–191
Parker M (2001) Planning land information technology research project: efficient recycling collection routing in Pictou County. http://www.cogs.ns.ca/planning/projects/plt20014/images/research.pdf. Accessed 22 Feb 2016
Rajabifard A, Coleman D (2012) Towards spatial enablement and beyond. In: Rajabifard A, Coleman D (eds) Spatially enabling government, industry and citizens: research and development perspectives. GSDI Association Press, Needham, pp 9–22
Resch B (2013) People as sensors and collective sensing-contextual observations complementing geo- sensor network measurements. In: Krisp JM (ed) Progress in location-based services. Springer, pp 391–406. doi:10.1007/978-3-642-34203-5_22
Rhine Atlas 2015 Koblenz; [cited 10.March 2016]. http://www.iksr.org
Rinaldi SM, Peerenboom JP, Kelly TK (2001) Identifying, understanding, and analyzing critical infrastructure interdependencies. IEEE Control Syst Mag 21:11–25
Rufat S, Tate E, Burton CG, Maroof AS (2015) Social vulnerability to floods: review of case studies and implications for measurement. Int J Disaster Risk Reduct 4:470–486
Stevens D (2007) Geospatial information, sustainable development and decision making in developing countries. In: Geoinformation for development. Herbert Wichmann Verlag, Heidelberg
Taubenböck H, Esch T (2011) Remote Sensing: An Effective Data Source for Urban Monitoring. 01.06.2014
Taubenböck H, Strunz G (2013) Special issue: remote sensing contributing to mapping earthquake vulnerability and effects. Nat Hazards 68(1)
Taubenböck H, Geiß C (2014) Vulnerability and resilience research: a critical perspective. Int J Disaster Risk Sci 5:86–87
Turner BL, Kasperson RE, Matson PA, McCarthy JJ, Corell RW, Christensen L, Eckley N, Kasperson JX, Luers A, Martello ML et al (2003) A framework for vulnerability analysis in sustainability science. Proc Natl Acad Sci USA 100:1–6
UNISDR (2012) Making Cities Resilient Report 2012. My city is getting ready ! A global snapshot of how local governments reduce disaster risk. Geneva
United Nations (2015) Sendai Framework for Disaster Risk Reduction 2015–2030. Geneva
US DHS (2013) NIPP 2013: Partnering for Critical Infrastructure Security and Resilience. Washington DC
US Government (1996) The President’s Commission on Critical Infrastructure Protection (PCCIP), executive order 13010. Washington DC
Ward PJ, Jongman B, Salamon P, Simpson A, Bates P, De Groeve T, Muis S, de Perez EC, Rudari R, Trigg MA et al (2015) Usefulness and limitations of global flood risk models. Nat Clim Change 5:712–715
Welle T, Birkmann J (2015) The world risk index: an approach to assess risk and vulnerability on a global scale. J Extreme Events 02:1550003
Winn MT (2014) A road network shortest path analysis: applying time-varying travel-time costs for emergency response vehicle routing, Davis County, Utah. http://www.nwmissouri.edu/library/theses/2014/WinnMichael.pdf. Accessed 7 Mar 2016
Wurm M, Taubenböck H, Krings S, Birkmann J, Roth A, Dech S Derivation of population distribution for vulnerability assessment in flood-prone German cities using multisensoral remote sensing data 2009. Int Soc Opt Photon
Wurm M, Taubenbock H, Roth A, Dech S (2009) Urban structuring using multisensoral remote sensing data: By the example of the German cities Cologne and Dresden 2009: IEEE
Acknowledgements
We would like to thank the experts for their readiness for the interviews and provision of information. We would like to thank Jessica Bussing and Lukas Edbauer for assistance in certain data search and compilation. We are grateful to the guest editors for inviting us to this paper and thus inspiring us for the work conducted.
Author information
Authors and Affiliations
Corresponding author
Additional information
Special Issue in Natural Hazards: ‘Remote sensing for multiscale mapping of elements at risk’ by Hannes Taubenböck and Christian Geiß.
Rights and permissions
About this article
Cite this article
Fekete, A., Tzavella, K. & Baumhauer, R. Spatial exposure aspects contributing to vulnerability and resilience assessments of urban critical infrastructure in a flood and blackout context. Nat Hazards 86 (Suppl 1), 151–176 (2017). https://doi.org/10.1007/s11069-016-2720-3
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11069-016-2720-3