Loading...
Please wait, while we are loading the content...
Similar Documents
Uncertainty on coastal flood risk calculations, and how to deal with it in coastal management: case of the Belgian coastal zone
| Content Provider | Semantic Scholar |
|---|---|
| Author | Verwaest, Toon Vanneuville, Wouter Peeters, Patrik Mertens, Tina Wolf, Peter De |
| Copyright Year | 2007 |
| Abstract | A coastal flood risk calculation estimates the damage by coastal flooding for a certain time horizon. Five different sources of uncertainty can be distinguished: unpredictability of the weather; uncertainty on the extreme value probability distribution of storm surges; unknown future values of economic growth rate, population growth rate, sea level rise rate and discount rate; limited knowledge of the behaviour of the coastal system; limited amount of measurements about the state of the coastal system. From a preliminary analysis for the Belgian coastal zone it is clear that the combined effect of these different sources of uncertainty results in a very large uncertainty on the calculated risk, namely a sigma of a factor more than 10. Some important sources of uncertainty are impossible to decrease substantially by doing research or measurements. Therefore the only option for coastal management is to deal with these large uncertainties. It is suggested to use calculation results relatively, namely to compare scenarios of coastal management in order to determine which scenario can best use an available budget for investment. Also it is concluded that risk calculation results would best be compared as ratios between scenarios (in %), not as differences (in euro/year). 1. The Belgian coastal zone The Belgian coastal zone is part of the North West European low-lying coastal areas along the Southern North Sea, with a length of 65 km. In Belgium this area has an average width of 20 km and is located an average of 2 m below the surge level of an annual storm. The natural sea defences are sandy beaches and dunes. However, hard defence structures have replaced the dunes almost everywhere in the coastal towns and ports, and hence representing approximately two thirds of the Belgian coastal defence line. The Belgian standard of coastal protection is to be safe against a surge level with a return period of 1000 years. At present it is investigated if and how this standard could be modified using risk calculations. 2. Sources of uncertainty on coastal flood risk calculations A coastal flood risk calculation aims at estimating the damage by coastal flooding for a certain time horizon. It is important to distinguish the different sources of uncertainty which influence a prediction of damage for a certain time horizon, for a given coastal zone. Five different sources of uncertainty are being distinguished in this paper: 1) the unpredictability of the weather 2) the uncertainty on the extreme value probability distribution of extreme storm surge events 3) unknown future values of economic growth rate, population growth rate, sea level rise rate and discount rate, for the time horizon under consideration 4) the limited knowledge of the behaviour of the coastal system during a coastal flooding event 5) the limited amount of measurements about the actual state of the coastal system. In the following sections a method will be presented to estimate the impact of these five sources of uncertainty on the damage, with a preliminary application for coastal flood risk assessment of the Belgian coastal zone. 3. Uncertainty caused by the unpredictability of the weather The predictability of the weather is very limited (~a few days), compared to the time horizon T considered when doing coastal flooding risk calculations (~100 years). The chance of occurrence of a coastal flooding is very small (for the Belgian coast ~1/10000 years) compared to the time horizon considered in actual coastal zone management (~100 years). So for the Belgian coast in 99 out of 100 possible futures there is no coastal flooding damage during the time horizon under consideration. Coastal flooding damage in a specified time horizon T is the result of a Poisson process. Weather systems change each few days. For every independent weather system there is a chance of occurrence of an extreme storm surge that results in coastal flooding. Because the considered time horizon T (~100 years) is much smaller than the return period R of coastal flooding (~10000 years) the expected damage E(D) can be calculated as E(D)=T/R×S in which S is the damage in case of a coastal flooding. For clarity of the arguments the damage due to coastal flooding is simplified to a constant value S. In other words, the relation between damage S and return period R is simplified as a step-function. This binary approach is to be generalised for a more realistic case in which the damage S is an increasing function of the return period R, but this is out of the scope of this paper. So in this paper the expected damage (euro) and risk (euro/year) are defined by the following equations (3.1) and (3.2). S R T D E ⋅ = ) ( (3.1) R S T D E risk = = ) ( (3.2) The variation around the expected value for damage during the time horizon considered can be expressed by the coefficient of variation valid for a Poisson process μ = R/ T. For the Belgian coast typical values are R ~10000 years and T ~100 years, so μ ~10. This means a very large uncertainty on the damage, namely a factor ~10. 4. Uncertainty on the extreme value probability distribution of extreme storm surges The extreme value probability distribution is essentially the result of an extrapolation of storm surge events recorded during the past decades/century. For the Belgian coast almost 100 years of reliable storm surge measurements are available. However, such a dataset remains very limited when one has to determine storm surge levels of extreme events with return periods of ~10000 years. Because of the importance of extreme storm surge levels in coastal management, in Belgium several detailed statistical studies were carried out in previous years. Extreme value probability distributions were determined, and also the uncertainties on the distributions. The results of Probabilitas (1999) are shown in Figure 1. 525 550 575 600 625 650 675 700 725 750 775 800 825 850 875 900 1 10 10 |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://www.vliz.be/imisdocs/publications/141414.pdf |
| Alternate Webpage(s) | http://www.vliz.be/imisdocs/publications/ocrd/141414.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
