M8 EEW Analysis

Estimated CISN EEW Performance for M8 Event

1. Calculate the P and S wave travel times to CISN stations for M8 event
   1. Pick the P arrival and calculate the time of peak amplitude for each site using M8 seismograms
   2. Create a table comparing calculated P and S and picked P and calculated time of peak
2. Create table of all Virtual Seismologist (VS), OnSite, and Elarms reporting stations from 1/2009 until present using CISN contributed data
3. Create table of all VS, OnSite, and Elarms forecast stations for same time period
4. Create mapping from reported station IDS to station locations (lat/lon)

---

attached the two station lists for southern and northern California. For calculation of the warning time you could use:

t_warn = S-wave@user - P-wave@sensor - delays

where S-wave@user = 1/vs * sqrt(epidist_user2 + depth2) P-wave@sensor = 1/vp * sqrt(epidist_sensor2 + depth2) delays = 3 seconds + 2 seconds [=algorithm time of on-site alg. + typical telemetry/processing delays at up-graded CISN stations]

As you write, you can of course use the simulated P- and S- wave arrivals rather than constant vp (~6.1 km/s) and vs (~3.55 km/s) values.

In principle, one sensor is sufficient to have an uncertainty of +-0.6 (standard deviation) for small to moderate earthquakes. However, I am quite skeptical that we will have a similar performance for large earthquakes (I am currently working on a magnitude-dependent error assessment) due to the saturation effects resulting from using a limited time window of 3 seconds and the point-source approximation (true for all 3 algorithms). The CISN EEW group has started to do some research on finite faults for consideration of large earthquakes, but this work is far from being ready to be included into the CISN ShakeAlert system.

Do you plan to show static maps for the warning times or movies? For the movies it might be nice to display a countdown of remaining time until S-wave (or peak shaking) arrival -similar to what Brad did for ShakeOut.

---