Uncertainty On Shaking Map Prediction And Its Application

Destructive earthquakes often cause heavy casualties and property losses, posing a great disaster to human society. Because of the extreme complexity involved in the earthquake process, reliable earthquake prediction is not currently possible. In order to reduce earthquake damage, seismic fortification and earthquake emergency research must be strengthened. As a tool for earthquake response, shaking map can portray the extent of potentially damaging shaking following an earthquake, providing distinctly important information for earthquake relief.The output of shaking map involves many aspects, such as observations (from macroseismograph or seismograph), ground motion prediction for phantom station, interpolation method, acquisition of shaking intensity (instrumental intensity), site effect and so on. Each of them can cause uncertainty in some ways. For this reason, the ShakeMap(?) system implemented by United States Geological Survey releases uncertainty map besides instrumental intensity, peak ground acceleration, peak ground velocity, and response spectrum. Since shaking map research in China is in its infancy, there is less study on its uncertainty.If there is a macroseismograph, the ground motion acceleration, which is closely related to structure damage, can be obtained directly. If there is a seismograph, the ground motion acceleration can be estimated from the velocity record and there is uncertainty in the process. If there is no seismic station, the ground motion is just estimated from Ground Motion Prediction Equation (GMPE), which brings uncertainty due to GMPE itself. Lack of certain relationship between intensity and ground motion parameters, there is uncertainty when changing ground motion parameters into intensity. These uncertainties are analyzed in this dissertation and the results are used in rapid earthquake damage assessment in the end. Also, as another application of shaking map, the heavy damage index is built to judge which county will suffer heavy damage after an earthquake.The main work and results are as follows:1. A new Ground Motion Prediction Equation is gained based on NGA strong motion data, which includes site term and Joyner-Boore Distance. The new GMPE has not been converted for China Mainland and there are several reasons to do this:(1) NGA collects strong motion data from shallow crustal earthquake all over the world, not confined to one region.(2) Due to high uncertainty of intensity, the transform of GMPE from one region to another through intensity will result in increasing uncertainty.(3) The new GMPE adopts Joyner-Boore distance, while intensity prediction equation often uses epicentral distance.2. We propose a method to compute the probability of shaking intensity for counties in seismic area by means of the stochastic variable ε in GMPE. Specifically, we build the logarithmic normal distribution about peak ground acceleration, using the estimated value and the standard deviation of the GMPE, to calculate the probability of every possible shaking intensity and the probability exceeding seismic fortification intensity for counties in seismic area. It is thought that the intensity displayed in a probability way is much more reasonable.3. Using the data recorded by both Hi-net and KiK-net from282earthquakes with JMA magnitude greater than5and focal depth less than100km, we study the distribution of peak acceleration ratio and acceleration response spectrum ratio between Hi-net velocity differential and KiK-net observed acceleration, as well as the ratios in relation to magnitude and focal distance. Also, we obtained statistical relation in peak value between Hi-net velocity and KiK-net acceleration, as well as statistical relations in peak value and acceleration response spectrum between Hi-net velocity differential and KiK-net acceleration. It turns out that the differential of digital velocity record is different from observed acceleration in both amplitude and spectrum, which therefore cannot be directly used as acceleration.4. We build a shaking map model which considers uncertainty. And it is employed to produce the shaking map for the Lushan earthquake, which conforms well to the result of USGS ShakeMap.5. Using the method of computing the probability of shaking intensity for counties, we compute the probability of every possible shaking intensity for counties in Lushan area. The intensites and their probabilities are then used to estimate economic losses and casualties. Due to the vulnerability model, the economic losses, in spite of the same scale, are lower than the actual economic losses and the estimation of deaths is higher than the true number. 6. The heavy damage index, which is useful for earthquake emergency, is built based on the method of computing the probability of shaking intensity for counties to judge which county will suffer heavy damage after an earthquake. Besides, the heavy damage index for counties, under scenario earthquakes in North China, can be a damage database providing improtant information for future similar earthquakes.