Ecole Normale Superieure, Paris,France

PRESTo (PRobabilistic and Evolutionary early warning SysTem) is a software platform for regional earthquake early warning that integrates recently developed algorithms for real-time earthquake location and magnitude estimation into a highly configurable and easily portable package. The system is under active experimentation in Southern Italy on the Irpinia Seismic Network (ISNet), which is deployed in a seismogenic area that is expected to produce a large earthquake within the next 20 years. In this paper we describe the architecture of the system and test its performances using both small earthquakes (M<3) recorded at the ISNet, and a large event recorded in Japan, through a simulation mode. The results show that, when a dense seismic network is deployed in the fault area, PRESTo can produce reliable estimates of earthquake location and size within 5-6 seconds from the event origin. Each estimate is provided as a probability density function, with an uncertainty that typically decreases with time: a stable solution is generally reached within 10 seconds from the origin. Thanks to its fully probabilistic approach, PRESTo can be a powerful tool for end-users in addressing the trade-off problem of whether and when to init safety measures. The software makes use of widespread standards for real-time data input and output, and can be finely tuned to easily adapt it to different networks and seismogenic regions.