Automatic arrival time detection for earthquakes based on logarithmic transformation

Earthquake Early Warning System (EEWS) greatly affects diminishing the mischief impacts coming about because of earthquakes, for example, human demise, atomic spillage, tainting of water, and properties harm. In this paper, we proposed a novel approach to detect the start of the earthquakes which is the main module in the EEWS. Our proposed algorithm based on dividing the seismic event into two parts noise and seismic signal. This target can be achieved using the segmentation techniques. In segmentation algorithm, we separated the seismic noise from the seismic signal and set the edge between those two categories as the onset time. We propose to use LOG transformation as a segmentation tool because its advantages in reducing the skew of the input data and use a hard decision threshold to detect the onset time. The proposed algorithm is simple and has high accuracy on picking the onset time of the earthquake. Our algorithm achieved an onset picking accuracy of 90.1 % with a standard deviation error of 0.10 seconds for 407 seismic field waveforms. Also, the proposed algorithm is hardware friendly, and a simple implementation is presented in this paper for it on a cheap FPGA kit. The implemented algorithm is compatible with the on-site and network approaches for implementing the EEWS.