nice interesting one.
This one covers Rebound/long duration Mercalli X events (r/lx), similar to the 1964 event in Alaska and the 2004 event of the S. Andaman backarc basin.
The book rightly states that Cascadia is a backarc basin and is therefore spreading with runout as a result greater then the recompression. The problem is that unlike classical backarc basins, the basin & range is driven by actively spreading hotspots in the region of the backarc backfault.
The frequency 500-650 years is about right but this is the general frequency for the backarc basin spread in the vector of spreading from the Panguitch hotspot & Basin & range (NW) opposed to the Juan de Fuca (~E).
Problem is that the Basin & range backarc basin is driven by hotspot, Panguitch as previously mentioned), Raton-Clayton. Colorado, Yellowstone, and Wells Grey-Clearwater(Canada). Runout @ the backarc basin also varies from classical arc (~300 years) through to the hotspots (over 5000 years), and the vectors of runout are radial to the hotspots opposed to the motion of the Juan de Fuca plate (which by the way is also hotspot driven (Axial).
The last was a runout driven WNW. and ran out all the way to Colorado with the result that the backarc basin is therefore rather quiet, and the biggest runout was in Oregon.
Biggest problem is that the Wells grey-Clearwater hotspot & Garibaldi range SW. vector is currently also in the latest stage or recompression with magma erupting @ the hotspot volcano in recent history. This as a result is likely to overprint the Cascadia zone with a SW. directed event centred on the Garibaldi range and extending all the way through Cascadia. The generally dtated "megaquake" generally covers an r/lx which covers more than one arc, which is the likely scenario in the next event. Working out when such an event occurs is rather difficult as the previous runout must be known in order to accurately calculate the recompression interval time(!).
As the runout in a backarc basin is variable and may typically cover arc, backarc fore, backarc main, backarc rift, backarc rear or backarc backfault (or anywhere in between these classical features), this runout distance can only be calculated by gaining an accurate handle on the runup of the tsunami produced by the event, and in a slow crawl situation lasting over a day, thie tsunami height and runout would always be an underestimate.
In the last event 1700AD the tsunami which struck Japan with a WNW vector runout produced a 7 wave train lasting for 6 hours. The possibility of slow crawl runout lasting for a day or more in the runout centre in Oregon is quite high.
Also note that the last event decompressed the backarc basin from Oregon through to the Colorado hotspot. This also decompressed the continental crust in a slow spread from Colorado right across the Continental USA with as a result 3 major rift event intraplate quakes in New Madrid (1812-13) and Charleston (Carolina coast) in the late 18th century. all in line in a great circle between Cascadia (Oregon) and the Colorado hotspot.
Time to have a close look at NW Vancouver island,(forearc region of the arc adjacent to Silverthrone volcano), that one will be the best clue to the timing of the next event, as it covers exclusively the Garibaldi range , rather than the Juxtaposed events in Cascadia alone.
Also note that the Gorda arc (North California) is also a backarc basin with the backarc rift zone in Owens Valley california, so there are 3 juxtaposed backarc basin in the region adding to the confusion with vector of runout & timing.
Have a nice day: Ag