Picking the arrival of a seismic wave on a record

Reading a seismogram and marking the instant the P wave arrives, then the S wave buried in the coda of the P, was analyst work, and the S arrival was the part that separated a good analyst from an average one. Networks now record continuously on thousands of stations, which is already more than any group of people could read, so no human score was ever set as the benchmark. The machine was trained to imitate analysts and now does the picking for entire catalogues, including the small events nobody had time to look at. Not ceded: the review. The United States Geological Survey marks a solution automatic until a seismologist has signed it, and for a significant earthquake the location, depth and magnitude are still checked by a person before they are used.