That was my attempt to quantify just how this past summer shaped up against previous ones, since the perception was that this summer was not so fantastic. My results found this summer was just a bit below average.
But that was simply using average temperatures and number of summer days it rained.
David Obelcz asked for the raw data and ran with it to take it one step further. David used a weighting system that factored in the actual temperature -- giving graduated points or penalties depending on how hot or cold it got, and giving extra weight to weather that occurred on weekends and holidays.
Here's his methodology:"High Temperature: The ideal temperature range that got the most weighting was from 70 to 84 degrees for a high temperature. From 85 to 94 degrees there was a slight penalty, and for 95 and above there was a stronger penalty. From 60 to 69 there was a penalty and from 50 to 59 there was a strong penalty. There wasn’t a high temperature below 50 degrees for the model period."
"Low Temperature: The ideal low temperature range was from 60 to 69 degrees, above or below that range there was a weighting factor to penalize the range. Temperatures below 45 degrees gets a very heavy weighting penalty, temperatures below 40 (only happened on a couple of days) got an even heavier weighting."
"Rainfall: If there was a trace of rain specifically, then there was a minor penalty. It is hard to determine from this data set if the trace of range comes from say morning fog/drizzle that burns off or a cloudy gray day. The penalty grows the more rainfall that happened, with the maximum penalty happening after 1/2 inch of rainfall or more. I cut it off at 1/2 inch because that is pretty rare in a 24 hour period in the summer, and again there is no way to tell in the dataset if this happened during a rare heavy thunderstorm, or from a late/early season winter style storm."
"Days of the Week: Neutral range was Monday to Thursday. Friday got a little extra weight, as the weekend for most starts at 5 PM on Friday. Saturday got the heaviest weighting bonus (or in the case of poor weather it became a penalty) and Sunday got weighted between Saturday and Sunday."
"Months: June got neutral weighting, July got increased weighting August got the highest weighting and September got a slight weighting bonus."
"Holidays: June 21, 4th of July and Labor Day got a major weighting bonus. These were major days, but the model isn’t setup as such that a great 4th of July would hide an otherwise bad summer."
"Sample Dates: The window of “summer” weather I rated was from June 1 to September 20 to 22 (depending on when the autumnal equinox happens). This period is done from 1948 to 2008."
Note that my original index only spanned from June 1 through August 31. This factors in most of September as well.
"The model I built creates a score considering all of the above factors, with each day getting a score. The low in theory would be around 2.00 (sorry, didn’t run all possible combinations) and I did see one day that was over 40! Talk about a perfect Seattle summer day! The daily scores are added together for the summer season and totaled to give the result. One could argue that the wiggle from the 20th to 22nd at the end of September could impact the scores, but I don’t think that is very dramatic. I say that because in September the daily scores are pretty low, between 3 to 6, which I don’t think is a major factor."
MISC NOTES -- SUPREME IRONY ALERT!
His Issues and Observations: "Overall, the model gets better in more recent years, it may be possible that the scores get higher due to trends in global warming, which this model does not factor for (out of my meteorological ability). Also, the summer of 1948 is somewhat unfairly scored as we’re missing about a week of data."
Also of note, the summer of 2001 scored much better in his model than in my model, attributing to the fact that perhaps weekends were better than weekdays.
The best year was 1967 using this model, scoring an impressive 838.89. Only 1998, 1987, 1994, 1989, 1958, 2003 and 1967 scored over 800. The worst year by a landslide was 1954, scoring a 566.96. The average is 730.10, with 1975 being the closest to average summer.
You might also note that 2007 scored fairly well, when that summer was considered miserable by many people's standards as well. David explains:
"Now, I dug deeper into the 2007 numbers because darn it, the summer of 2007 just wasn’t that good, at least not the way I remember it and I know numbers don’t lie so then I got worried that maybe I coded my model wrong.
"One score jumped out on me on the daily data, one day in 2007 scored a 46.5, the highest score for a single day from 1948 to present. Translation, the perfect day in Seattle by every definition.
"That perfect day happened to fall on the 4th of July in 2007, which I vividly remember as being, well, perfect. Take away that absolutely perfect day on the 4th, and the score for the month plunges down to 727, close to the summer we just had (which also had an outstanding 4th of July)."
So there is your big dose of irony for the day. July 4th is statistically the wettest day in July, but July 4, 2007 might go down as the nicest day in Seattle's history. I sense we'll pay for that in spades someday :)
WHAT ABOUT THIS PAST SUMMER?
Adding in the very pleasant September has improved 2008's standings:
"The summer of 2008 scored a 729.06," David said. "The mean score for the period is 730.09 so actually, the summer of 2008 was about as average as you can get. What hurt 2008 from being an excellent summer was the wet period of 8/17 to 8/31 and the cooler weather we got at the end of this window, had that not happened I think this would have been a very high scoring summer..
A CHALLENGE TO ALL YOU OTHER WEATHER STAT GEEKS OUT THERE
David was kind enough to tackle this on, but do you want to come up with your own index?
Can do -- this link is a text file of the entire daily weather history of Sea-Tac Airport, courtesy of the Western Regional Climate Center in Reno, Nevada.
This spans from June 8, 1948 to June 30, 2008, but if you are going to do something with this, e-mail me and I'll get you the data between then and today.
(The data goes: Year, Month, Date, Time, High, Low, Average, Heating Degree Days, Cooling Degree Days, Daily rainfall (0.00T means "trace"), Daily Snowfall, and snow measured on ground.)
And if you come up with any similar type observations or indexes, let me know and I'll post it here!
Here is David's "Summer Satisfaction Index"