In reality, it is very difficult to say with certainty whether an earthquake has occurred or not from using only one sensor. For example, a person walking nearby a sensor will register same response as a small seismic shaking. This is where geospatial distribution of sensor array comes into play. An intelligent algorithm at the server can look at all the picks being registered by sensor array and decide if it is an earthquake. This is plausible because all the sensors in sufficiently large geo-spatial area won’t pick at once unless a significant ground motion occurs across all of them.
If such an event were to occur, then it could be said with very high certainty it must be a seismic event. The next step should be to determine hypocenter in real time, calculate arrival time of waves at the populated areas, and issue the warning.
In theory, if S wave and P wave arrival times at three geospatial points are known a correctly, Hypocenter can be estimated with high certainty using the method of triangulation. [http://www.csulb.edu/~rodrigue/geog558/labs/epicenter.html].
However, as stated earlier, it is difficult to distinguish S wave from P Wave in data stream generated by CSN Sensors, due to their cost effective nature. CSN however currently employs a Bayesian Technique to find most likely Hypocenter with accuracy of [X] Km. But this can be improved several fold by using past data to construct and validate new model. This project works on such a model, called Brute Force Hypocenter Estimation.
No comments:
Post a Comment