Likelihood of an 8.0

We prepare — or choose not to — based on our perceived risk.
For that reason alone, understanding the risk is key to 
Surviving Cascadia.

We often hear there is a “37% chance of a major Cascadia Subduction Zone (CSZ) earthquake in the next 50 years”. The probability is derived from a log-normal distribution, like that shown on the left of this image.

To learn more about the 37% chance and to play around with the calculations to see how they change over time, visit Step 2 on Surviving Cascadia’s Timing The Next Big One page. This page, however, exists because the Log-Normal formula is not the only way to evaluate the risk.

Where the Dates of Past Events Come From

Regarding the information provided throughout the page below:

Dates for earthquakes T1 through T18 can be found in Table 10 (shown here on the right), page 97, (109/184) of the United States Geological Survey (USGS) research paper Turbidite Event History—Methods and Implications for Holocene Paleoseismicity of the Cascadia Subduction Zone. Event 19, listed at ~10,200 cal yr B.P. can be found on page 32 (44/184) of the same paper.

These are the 43 events used to calculate the “37% chance of an earthquake in 50 years”, which was mentioned above. But again, this page isn’t diving into that probability.

Statement from 2010:

“Perhaps more striking than the probability numbers is that we can now [as of 2010] say that we have already gone longer without an earthquake than 75 percent of the known times between earthquakes in the last 10,000 years… And 50 years from now [in 2060], that number will rise to 85 percent.”

Oregon State University: Odds are 1-in-3 that a huge quake will hit Northwest in next 50 years

Statement from 2012:

“Failure analysis suggests that by the year 2060, Cascadia will have exceeded… 85 percent of recurrence intervals for the southern margin.”

USGS research paper Turbidite Event History—Methods and Implications for Holocene Paleoseismicity of the Cascadia Subduction Zone (pg 2 (14/184)). If you have questions about what is meant by the “southern margin”, details are located on Surviving Cascadia’s Size of The Next Big One.

To fully grasp what this failure analysis means, it helps to have a visual. Take a look at this ten thousand-year historical bar graph for Cascadia. 

The horizontal blue line in the graph above lies at 360 years — the amount of time that will have passed between the last major CSZ quake in 1700 and the year 2060 (the year mentioned in the USGS paper) if an earthquake doesn’t occur by then. Intervals longer than 360 years are shown in red, while those that are shorter are in green.

Only 6 intervals stretch above the blue line… and they’re kind of clumped together. More on that in a minute.

The math supporting the USGS quote above:
42 intervals – 6 red intervals = 36 green intervals, 36/42 = 85.71%.

Of note: Interval #1 shows the amount of time the PNW will have gone without a CSZ event by the year 2060, so it isn’t an actual interval with two “bookend” earthquakes. This fact will hold true for interval #1 in the graphs below, as well. Therefore there have been 42 intervals (between 43 earthquakes) during the past 10,272 years.

Okay, but that’s in 2060, which is still 37 years away. What does the current data look like?

Failure Analysis for 2023

This bar graph drops the horizontal blue line to the current number of years the PNW has gone without a major CSZ earthquake. As before, intervals longer than 323 years are shown in red. There still aren’t many.

The current data (based on 10,272 years of history):
42 intervals – 8 red intervals = 34 green intervals, 34/42 = 80.95 ˜ 81% … as of today, meaning…

81% of the time, the fault has not had to wait 323 years for the strain to break it.

Let’s Play a Game

If you knew that 81% of the time, your toddler was asleep by 8:00 pm, would you expect her to be up until 9:00 pm tonight? Probably not.

Knowing that 81% of the time, our fault has broken prior to reaching its 323rd year… should we expect the fault to hold for another 50 years? 20 years?

Now, One Step Further

Remember earlier when I said, “More on that in a minute”? This is that minute.

Placing a yellow border around intervals 31 through 43 helps illustrate how the largest intervals are grouped toward the oldest chunk of the 10,272-year history.

How does the failure analysis look if only intervals 1-30 are taken into consideration? Let’s zoom in on the most recent 6,031 years of history on the CSZ.

There have been 29 intervals between the 30 earthquakes during the most recent 6,031 years. In that time, only 2 intervals (shown in red) have been longer than interval #1 (which, again, isn’t an actual interval, but rather our current time without incident). The two red lines don’t stretch far above!

The math supporting the current data (based on 6,031 years of history):
29 – 2 = 27, 27/29 = 93.1%

93% of the time, the fault has not had to wait 323 years for the strain to break it.

Let’s Play That Game Again

If you knew 93% of the time your toddler was asleep by 8:00 pm, would you expect her to be up until 8:10? Maybe. 9:00 pm? Nope. Just nope. It could happen. But you probably wouldn’t expect it to.

To be clear, if we only use the most recent 6,031 years of data and project out to 2060, as was done in the USGS paper… not a single interval would reach that high.

In fact, we only have 22 years until the PNW passes the longest interval during this time frame. Twenty-Two years…

Should we really be evaluating risk based on a 50-year probability model (37% chance) whose percentage has hardly changed in the past 50 years, will hardly change over the next 50 years, and which currently shows that our risk has been DECREASING since 2017 despite the continued building of stress on the fault? (Details available in the Bottom Line section of Surviving Cascadia’s Timing the Next Big One).

How would your preparedness journey look if you based likelihood on the log-normal 37% chance? How would it look if you based it on the failure analysis? It matters.

Additional Caution Regarding Intervals

In 2021, Forbes wrote a piece that stated, “The average recurrence time for a major earthquake along the CSZ is 300 to 500 years”. Scientific American, the Berkeley Seismology Lab, the City of Seattle, the USGS, and smaller organizations like KOMO News, a subsidiary of the Sinclair Broadcast Group, quote that same 300 to 500-year range.

When we hear there is an average recurrence time of every 300 to 500 years, and then we hear that “only” 323 years have passed since the last big event, it springs forth an image like the one below, because, surely, we are still toward the beginning of the normal range… right? Looking at this, it can feel like there is probably no real risk any time soon.

As you have probably guessed, based on the previous information provided on this page, this isn’t a good representation of the intervals. In that way, the statement provides a false sense of security.

Only 7 of the 42 intervals ranged between 300 and 500 years…
27 of the 42 intervals ranged between 100 & 300 years.

Check out this representation. Each rectangle represents one interval. The number of years between its two surrounding earthquakes is written inside. For example, the rectangle in the bottom right represents 508 years that stretched between one earthquake and the next. These intervals are displayed by ascending values. Any intervals that lasted fewer than 100 years are stacked in the first row, and so on. The rectangle in blue is technically not an interval. It represents the number of years the PNW has currently gone without a CSZ earthquake.

So is the PNW at “the beginning” of the normal recurrence intervals? Nope. If we take this a step further and look at intervals over the most recent 6,031 years like we did above, we get the following.

Just like the bar graphs above, we can see that only two intervals were longer than our current amount of time without an event, and they really weren’t that long.


The Oregon Seismic Safety and Policy Advisory Council (OSSPAC) has some good resources on its page. In accordance with Oregon Public Law, all of OSSPAC’s meetings are open to the public, and materials are posted on the website. Under OSSPAC’s Presentations and Testimony section is a link, Lifeline Resilience Teams Overview – Steve Robinson Submission (2019-09-10). It states:

Wait, the earthquakes have occurred every 202 years over the past 6,000 years?

How did they get that number? Looking at the list of ages below, if we are strictly looking at how many earthquakes have occurred in the most recent 6,000 years (as bulleted above), that includes events up to T10f — 29 earthquakes. (For the record, the bar charts above extended to event T11, just slightly above the 6,000-year cutoff that we’re examining here.) The age of T10f is estimated at 5,845 years. 5,845/29 = 201.55… Reminder, we have currently gone 323 years.

A New Perspective

So why do the oldest 4,000 years, and the most recent 6,000 years, look so different on the bar graphs? This EOS image is one possible explanation.

Graphic A, the steady earthquake cycle, is similar to the CSZ’s behavior in the oldest 4,240 years (with regards to the 10,272-year timeframe). This pattern flows from a magnitude 9.0 to a single, smaller 8.0 range, then back to a 9.0.

Graphic B, earthquake clustering, resembles what the CSZ has experienced over the last 6,030 years (more magnitude 8 earthquakes coming between the 9s).

Here is what a couple of professionals in the field have to say about the matter.

website 4.JPG
website 3.JPG

When Did The M8s vs M9s Occur?

Looking at the chart below, it is again clear that the most recent history on the right side looks ‘busier’ than the left, with more 8.0s (short lines) between the 9.0s (long lines). Occasionally, there aren’t any 8s between 9s. Look at the two most recent megathrust earthquakes, for example. Still, those two earthquakes came fairly close together with only 232 years between.

The short bars above represent partial ruptures, while the tall bars represent full-margin ruptures.

So Are We Doomed?

Referencing the 2011 Tōhoku earthquake and tsunami in the OPB documentary Unprepared, Oregon State University Professor Chris Goldfinger says:

“Japan is a great example of how an earthquake like this, as bad as it is, is survivable. And to just throw up your hands and say ‘we’re all gonna die’, is wrong.

We’re not doomed. Nor are we totally safe. Denial isn’t an effective preparedness strategy. A Cascadia Subduction Zone megaquake is a very real threat to our region. It’s critically important to understand the risk of that threat.

Click Here to return to the top of the page.

%d bloggers like this: