Performance of Network Based EEW Systems in Marmara Region; VS, ElarmS-2 and PRESTo

Abstract:

Several earthquake early warning algorithms are in operation in various countries. Kandilli Observatory and Earthquake Engineering Institute (KOERI) implemented and recently running three different EEWS within KOERI network in Marmara Region. Due to high probability of an earthquake with M_w > 7 in the next 30 years in the Marmara Sea, there is an urgent need to evaluate existing EEWS performance and to find out how reliable an EEW is in terms of calculating time, location, magnitude of an earthquake in the region. Here we assessed performance of Virtual Seismologist (VS), Earthquake Alarm System-2 (ElarmS-2) and PRobabilistic and Evolutionary early warning SysTem (PRESTo) which are developed at Caltech, UC Berkeley and University of Naples respectively and are being monitored for over six months. We choose four representative earthquakes that are detected by three algorithm and evaluated their performances. We found each algorithm has its own strengths and weakness in calculating source parameters. Even though, algorithms are not optimized, cities around the Marmara Sea could get up to 50 seconds warning time with existing infrastructure. First alert could be given in 12 seconds after origin time and blind-zone could get minimum 40 km radius. The main reason of delays on detecting earthquake is packet size of data (7 seconds in average). We confirm that SeisComp3 platform requires additional time to locate earthquakes and it needs to be optimized for EEWS. This study could help algorithm developer of three EEWS to optimize their systems in other seismically active region and different earthquake environment. We believe results derived from this study will provide precious information for both habitants in Marmara region and geo-scientist.