DEVELOPMENT OF ALGORITHM FOR EARTHQUAKE EARLY WARNING SYSTEM

Abstract

Part of North-Western Himalaya, Which typically comprised of, Jammu and Kashmir, Laddakh, Himachal Pradesh, and Uttarakhand, particularly Uttarakhand, has been identified by several eminent seismologists as a gap area where large size earthquake( greater than magnitude 7) is expected. The region has witnessed two earthquakes having a magnitude greater than 6 in the last 25 years ( 28th March 1999 Chamoli earthquake and 20th October 1991 Uttarkashi earthquake ), both having an epicenter in the Garhwal region of Uttarakhand. Several studies have shown that a future large earthquake in Central Himalayas can generate severe ground motion in the National capital region of Delhi, which is about 300 km from this expected source. Several densely populated cities and villages having a total population of several million are located between Delhi and Uttarakhand. These towns/cities will be severely affected by a large earthquake having an epicenter in this region. Using advancements in communication technology and real-time seismology, a project to have an earthquake early warning system for Northern India is under progress at the Indian Institute of Technology, Roorkee. In this project, a dense network of 84 accelerometers has been installed, covering an area of about 100 km x 40 km in the Garhwal region. This network has an intermediate station to station distance of less than 10km, and all the sensors are streaming data, which is being processed in real-time at the central server stationed at Roorkee. The cost of the sensor becomes one of the most important parameters on which the viability of an Earthquake Early Warning (EEW) system depends since a dense network of the sensor is required for an effective EEW system. Thus in this study, a detailed analysis is also performed to find out the optimum dynamic range of sensor that is required so that the most relevant information, EEW parameter, and strong motion parameters could be retrieved from the obtained ground motion data. In the study, it is observed that for all relevant EEW parameters, a sensor with ADC with the 16-Bit resolution is quite sufficient, and most of the required information could be retrieved from such data. This thesis describes details of the network, sensors, present status of development, performance of instrumentation during recent events and processing details. The thesis also provides details of the theoretical and numerical analysis done for the ADC dynamic range varying from 24-Bit to 10-Bit and its impact on the most relevant EEW parameter and strong motion parameters. A new approach for P-phase picking has also been developed in this study. The parameter used is ‘Damage Intensity’ which has been widely used in Japan for intensity estimation for UrEDAS. One of the problems faced during the testing of the EEW system was to eliminate far-field earthquakes. By far-field events are those events that have their epicenter at a large distance from the networked region. During testing, one such event from the Hindukush region was picked by the software. Thus it was a concern that such events should not be picked or, if picked, should be ignored by the EEW system. Using this approach, it was found that far-field earthquakes could be eliminated and were not picked if this approach is used.