The reliability of a natural hazard systemM. Khaleghy Rad and S.G. Evans Reliability of any system is defined as the probability of survival of the system over a period of time. In the natural disaster context, this means the probability of zero losses due to a natural hazard event. We examine the reliability of natural hazard systems using a definition of risk in which risk is considered as annual loss due to a (hazardous) event (e.g. earthquakes, tsunamis) following our recently developed risk=hazard*(1/resistance*exposure) equation, where hazard gives the probability of an event and exposure characterizes the exposed object to the event. Here, resistance conditions the response of the affected area to an event. The inverse of resistance is equivalent to the widely-used definition of vulnerability of a system. Here we use resistance as the main parameter to identify reliability, since no loss occurs when a system is completely resistant. We use disaster data for losses due to an event to calculate resistance. Then we estimate the distribution of resistance from which we evaluate reliability as its complementary cumulative distribution function. We apply this concept and methodology to an example of earthquake events. All earthquake records for the period 1973-2010 in the NEIC (National Earthquake Information Center) catalog of earthquakes are analyzed. In this case life loss is the measure of risk due to earthquake disasters in this period, for events with the annual probability of exceeding M5.5 per year (M5.5 as a threshold that causes life loss according to our data). Then resistance is calculated in the risk equation by taking exposure as the yearly global population, which we use a proxy to the exposure term in the risk equation. Furthermore, the probability distribution of all recorded earthquakes in a year is best fitted to Poisson distribution. Using the probability of hazard in a particular year, the resistance value of that year is calculated. The resistance distribution is best fitted to the well-known log-normal distribution. Finally, knowing the resistance distribution, we calculate the reliability of exceeding the estimated resistance of year 2011 considering different life loss scenarios due to earthquake disasters. Results show a low global reliability in 2011 towards earthquake disasters. Our study also provides new quantitative criteria for natural hazard risk assessment, which is absent in the other risk assessment methods particularly in the FN-graph based approach. |
