NHERI-SimCenter
Question about Ground Motion Generation in San Francisco Bay Area Case #330
-
|
Dear Authors: I am investigating the San Francisco Bay Area Case published by SimCenter When doing this, I found the following description in Section 1.3 "The ground motions are assigned to buildings using a nearest-neighbor search algorithm, where the four nearest grid points are identified for each building and a set of 25 seismographs are assigned by weighted random sampling of the set of time histories from the nearest grid points. The weight of each grid point is inversely proportional to its squared distance from the building" Considering that the ground motion on each grid point is fixed, and the distances among the building and gird points are also known. How could the 25 seismographs be generated? In other words, how does the "random" means in the " weighted random sampling" process? Is it some adjustment based on the basic weight, which is inversely proportional to its squared distances? Thanks for your time in advance for answering my question. |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
Hello, The 25 seismographs are generated by randomly selecting 25 seismographs from the four nearest grid points. For a given building in the region, the four nearest grid points are first identified. We then randomly selected the grid points 25 times following a multinomial distribution with the weights determined using the inverse of distance from the building to the grid points. This is a method to "probabilistically interpolate" the seismogram, as "arithmetic averaging" does not work for time sequences, such as seismograms. Using this method, some seismograms are applied to the building multiple times. This is fine because the building parameters are different in each simulation run. Overall, this is a Monte Carlo method to quantify the uncertainty in seismogram interpolation and building modeling. Hope this is clear. |
Beta Was this translation helpful? Give feedback.
-
|
Hi Jinyan, Got it. Thanks a lot for your clear explanation! Yongjia Xu |
Beta Was this translation helpful? Give feedback.
Hello,
The 25 seismographs are generated by randomly selecting 25 seismographs from the four nearest grid points. For a given building in the region, the four nearest grid points are first identified. We then randomly selected the grid points 25 times following a multinomial distribution with the weights determined using the inverse of distance from the building to the grid points.
This is a method to "probabilistically interpolate" the seismogram, as "arithmetic averaging" does not work for time sequences, such as seismograms. Using this method, some seismograms are applied to the building multiple times. This is fine because the building parameters are different in each simulation run. …