How Often are Aurora Borealis Visible in Seattle?

A friend and I were sitting on my balcony the other night, enjoying a wonderful 10-year old Lowland malt and talking about whatever came to mind, when we started talking about aurora borealis.

This is the highest latitude I've ever lived at, and I wondered:  how frequently could I expect to see the aurora borealis in Seattle?  I've only seen it once before, while driving a high school girlfriend home in the dark countryside near Boiling Springs, PA, and I'm overdue for seeing this wonder of the night sky again.

First, I learned an interesting (and disappointing) fact while perusing the NOAA website:  Seattle is actually at nearly the same magnetic latitude as Pittsburgh, so I'm only slightly more likely to see aurorae in Seattle as I was in Pittsburgh!  A map of magnetic latitude can be found on the NOAA website, reproduced here since the US government is not allowed to hold copyright.

Magnetic latitude lines corresponding to areas likely to see aurorae at certain geomagnetic storm strengths.
Image reproduced from NOAA.

So, both Seattle and Pittsburgh may see aurorae during a geomagnetic storm with a K-index of 7.  How often does this happen?  I downloaded 11 years of K-index data (one solar cycle) from the NOAA website, and produced a histogram of frequencies of K-indices.  The K-indices are averaged over three hour periods, for eight readings per day.  My data spanned from May 13, 2001 to May 10, 2012.  A simple bit of R code, then, produced the distribution of K-indices over these 11 years:

data <- read.table("~/Desktop/spidr_1336629415966_0.txt")



graph <- hist(as.numeric(data[data[,3] > -1,3]), 

              probability=TRUE, 

              right=FALSE, 

              breaks=0:10, 

              xlab="K-index", 

              main="Histogram of K-indices")



Frequency of three-hour periods with various K-indices from May 13, 2001 to May 10, 2012

The histogram above shows that the K index is 0 through 3 during about 90% of 3-hour periods, and periods of strong solar activity where aurorae might be seen in the continential US become increasingly rare.  To answer my question, only 1 in 245 3-hour periods had a K-index of 7 or higher, corresponding to approximately one per month.  Since this has to occur in a six-hour window around midnight for me to be able to see the aurorae, then assuming each event is independent of the last then if it were clear out then the northern lights could appear at night in Seattle about once every four months.  The assumption of independence is obviously wrong, since if the last three hours were calm I certainly expect it to remain that way, but I really just wanted an estimate.  A more sophisticated choice would be to model K-indices over time as a Markov process, or even to model solar flares themselves as self-organized critical behavior.   
If you happen to live near the red line, with Kp = 9, then frequency of events that strong at night is about once every 3 years.   Hope it's not overcast!  For those of you even further South: ice cores indicate that an event as strong as the Carrington Event, when aurorae were reportedly visible at the Carribean, occurs about once every 500 years!  

Incidentally, I found a page on the NOAA's website where you can see predictions for upcoming days, so if the NOAA is predicting a strong geomagnetic storm I can plan ahead to be away from city lights and improve my chances of seeing the northern lights again.