Advances in flood risk management under uncertainty
- First Online:
- Cite this article as:
- Levy, J.K. & Hall, J. Stoch Environ Res Ris Assess (2005) 19: 375. doi:10.1007/s00477-005-0005-6
- 404 Views
Flood events have an enormous impact on human well-being, jeopardizing important social development goals such as addressing poverty, ensuring adequate food, water, and sanitation, and protecting the environment. Accordingly, floods are among the most devastating of all natural disasters. Direct losses from floods include drownings and injuries as well as damage to infrastructure and property, agricultural production, and sites of historical and cultural value. Indirect health problems often arise, such as water-borne infections, exposure to chemical pollutants released into flood waters, and vector-borne diseases.
In many regions of the world, flooding is a frequent, widespread, and increasing natural hazard. For example, in recent years, water levels in England (Ouse River) and the European Continent (Elbe River) rose to their highest levels in recorded history, breaking previous records set in 1625 and 1845, respectively. Globally, the number of “great floods” (floods with discharges exceeding 100-year levels from basins larger than 200,000 km2) is on the rise. From 1990–1999, floods killed approximately 100,000 people and affected over 1.4 billion. Society is expected to become increasingly vulnerable to flood risks. For example, about 20% of the world’s population now lives in coastal ecosystems, which range from either 100 m elevation inshore or 100 km from the coastline. Moreover, the population density of coastal ecosystem zone (175 people per km2) is the highest among all ecosystems.
Statistically significant increases in global land precipitation have been observed over the twentieth century, such as wetter winters in Europe and wetter summers in the Asia monsoon region. The Inter-governmental Panel on Climate Change and global climate models predict larger inter-year and intra-year variations in precipitation and changing climatic conditions: in the past century, average global temperatures have risen about 0.6°C since 1850 and global sea levels have increased by 10–20 cm. In addition to climate change and rising sea levels, the extent and impact of flood events is also increasing due to human activities, including land-use changes (such as the alteration of vegetation along rivers and coastlines), growing populations (and denser concentration of people and infrastructure in flood-prone areas), and development in headwaters (altering natural hydrologic balance). The concomitant effect of these changes has led to a greater, more rapid runoff. For example, flooding on the Mississippi and Missouri rivers have substantially increased over the past century. Economic losses are also on the rise. The great US western flood of 1993 affected over 1600 km in the Mississippi and Missouri rivers and caused up to $16 billion US dollars in damages. The 1998 Yangze flood was exceptionally devastating, leading to over $30 billion US dollars in economic losses and more than 3,600 deaths.
Many flood-related activities have been undertaken around the world to reduce flood threats and damages. The United Nations (UN) declared the last decade of the twentieth century to be the International Decade for Natural Disaster Reduction. In the following years, the UN World Disaster Reduction Campaigns and other international disaster frameworks have attempted to heighten community awareness of flood risks and to improve societal resilience and adaptive capacity. In addition, advances in computing power and remote sensing technologies have provided for the cost-effective, continuous, and synoptic coverage of many hydrologic variables (including rainfall, snow melt, water level, ocean waves, and sea-surface height). However, flood risk management continues to be fraught with uncertainty, particularly with respect to hydro-climatic and socio-economic variables. For example, hydrological models often have significant uncertainties in the input (including forcing) data and the model itself (including model structure, parameters, states, and boundary conditions). Moreover, establishing a relationship between the statistical maxima of rainfall and runoff is problematic as the precision of disaggregated floodplain models must be tempered by the irreducible uncertainty of climate system dynamics and causative rainfall.
In order to improve flood risk planning and management under uncertainty a rich array of analytic techniques and decision support systems are herein put forth. The results of this special issue should be especially useful for organizations concerned with flood mitigation measures and disaster monitoring, including UN organizations, research institutions, government agencies, and other stakeholders. For example, insights from this issue can be used to meet the rising expectations of society for more timely and reliable heavy rainfall warnings.
There are a number of reasons why modeling and managing uncertainty are essential for sustainable flood risk management. They are as follows: 1. Extreme flood events are, by definition, rare, often requiring the use of synthetic data. 2. Translating the physical hydrological system into a numerical model involves simplification. This in turn leads to model (conceptual representation) uncertainty, particularly when the structural flood model is inadequate or ambiguous. 3. Predictive uncertainty arises in many contexts, such as extrapolating flood models to predict inundation behavior under conditions of variability and change. 4. There are limitations in measuring hydrologic forcing variables (such as precipitation and temperature) due to inadequate spatio-temporal sampling densities. 5. Significant uncertainty may arise due to insufficient understanding of socio-economic watershed characteristics (risk attitudes, infrastructure costs, etc.) and an inability to determine the ‘best’ model parameter set combination (based on available variables). The accurate parameterization of some hydrologic and meteorological processes is particularly difficult (cloud formation, soil moisture dynamics, land-surface interactions, etc.).
Model identification for hydrologic forecasting under uncertainty
Sampling based flood risk analysis for fluvial dike systems
Spatial–temporal rainfall modeling for flood risk estimation
Seasonal effects of extreme surges
Weather and seasonal climate prediction for flood planning in the Yangtze River Basin
Multiple criteria decision making and decision support systems for flood risk management
The first paper discusses state-of-the-art hydrological model identification, and outlines desirable features for an identification framework under uncertainty. The authors note that the existing methods for the identification of hydrological forecasting models have failed to adequately capture the specific nature of these models as well as the uncertainties present in the modeling process. They discuss the history and current state of model identification, as well as opportunities and challenges for obtaining a “general framework” of model identification. Moreover, desirable features for an identification framework under uncertainty are discussed and current gaps in the literature are identified.
The second paper uses an adaptive risk-based sampling technique to analyze dike system reliability and to estimate flood damage. The risk associated with dike failure is a complex function of loading, dike(s) properties, floodplain topography, the geographical location and the type of assets in the floodplain. The proposed methodology provides an efficient means of assessing the flood risk of a complex dike system. The authors deal with two important uncertainties in probabilistic terms: river flow at the upstream boundary of the hydrodynamic model and the failure of dike sections. The authors contribute to broad-scale (strategic) dike system management by efficiently sampling the risk response-surface to reduce the computational burden of analyzing all system states over the loading space.
The third paper summarizes recent developments in the stochastic modeling of single site and spatial rainfall for flood risk estimation. The authors also present a range of valuable methodologies for fitting and selecting single site models based on Poisson cluster processes. Poisson process-based methods are developed to simulate rainfall in continuous space and time (for input into rainfall-runoff models). Moreover, Poisson cluster single site models are used to disaggregate temporal sequences with simple spatial generalization.
The effect of seasonality upon extreme value analyses of sea levels is investigated in the fourth paper. The authors note that strong seasonal effects have been previously identified in the eastern coastline of the UK. The construction and inference of extreme value models are discussed for processes that involve seasonality. The authors provide a point process representation of extreme value behavior, using simulation based techniques in a Bayesian inferencing framework. Finally, an estimated model for seasonal surge is combined with tide records to develop extreme sea level estimates that take into account seasonal variation in both surge and tidal processes.
The fifth paper describes the use of numerical weather and climate models for predicting severe rainfall anomalies over the Yangtze River Basin from several days to several months in advance. Such predictions are extremely valuable, allowing time for proactive flood protection measures to be taken. Specifically, the authors apply a dynamical climate predication system in order to predict rainfall and improve flood planning and management in the Yangtze River Basin. Ensemble prediction with dynamically conditioned perturbations is employed in order to reduce the uncertainty associated with seasonal climate prediction.
Finally, the sixth paper uses advances from the fields of Multiple Criteria Decision Making (MCDM) and Decision Support Systems (DSS) to improve flood risk planning and management. Flood DSS are customized, interactive computing environments that integrate models/analytical tools, databases, and graphical user interfaces. A DSS architecture is presented that integrates the latest advances in remote sensing, GIS, hydrologic models, real-time flood information systems, and MCDM under uncertainty in order to compare, select or rank flood management alternatives. In particular, the author applies DSS to a recent flood management problem in the mid-reaches of the Yangtze River. The author shows how MCDM and DSS can facilitate the elicitation of flood preferences and improve the coordination among flood agencies, organizations and affected citizens.
Until the 1960s, flood mitigation measures in the US, Canada and other Western countries were dominated by the state promotion of large-scale engineering projects, such as dams, dikes and diversions. For example, since the nineteenth century, flood control works by the US Army Corps of Engineers have been used in an attempt to regulate large rivers such as the Mississippi. This emphasis on building flood embankments, constructing flood relief channels, and sometimes a series of flood control dams constitutes the structural (so called “hard”) approach to managing hydrologic systems. From this perspective, floods are “controlled” and defenses are prepared in order to “protect” floods from interfering with human activity.
Although providing some benefits, the effectiveness of the structural approach has been called into question: flood damages continue to rise (climbing to approximately $6 billion per year in the US alone) and floods have become more severe (Mekong, Mississippi, Yangtze, etc.). There are a number of soft measures for flood control such as developing flood hazard maps, identifying flood evacuation procedures, relocating and flood-proofing structures in the floodplain, and altering land management practices. The papers of this special issue present a range of techniques that can be used to enhance both the hard and soft flood management approaches.
While the US continues to rely heavily on structural flood control measures, many countries around the world (particularly in Europe) have shifted towards a flood management policy of “living with floods”, focusing on society’s ability to co-exist with floods, rather than be protected from them. In particular, after the severe Rhine River flooding of 1993 and 1995, the Dutch government has adopted a flood control policy of “more room for rivers” with an emphasis on establishing new storage and conveyance space. In this spirit, six European governments created the Meuse River High Water Action plan which provides “longer storage and more release” for rivers. The Netherlands alone has allocated three billion US dollars for a broad array of levee alternatives.
Fortunately, these flood risk management efforts have delivered impressive results. For example, in the state of Baden-Wurttemberg, Germany, the addition of over 200 m3 of floodplain storage has reduced flood stages to 1950 levels. The scientific advances found in this collection of papers complements the recent success of flood risk management in Europe and elsewhere. Specifically, this issue provides theoretical insights and practical examples to reduce the impact of extreme flood events under uncertainty. By doing so, we strive to promote a more holistic, sustainable relationship between society and the environment.