Beating the (usual) spread
Posted July 3, 2012
The "spread" in question here is the standard deviation of high temperatures for this time of year, which we exceeded, and then some, with the wave of very high temperatures that set in over the area late last week and through the weekend. You're no doubt used to hearing us mention a "normal" temperature that we compare the observed or forecast values against from time to time, but there is also a "standard deviation" that gives an idea of what range of temperatures we historically remain within most of the time. Around here, that spread varies notably with the seasons - it's about +/- 2 degrees in mid winter, and we're all used to the idea that day to day or year to year on the same date, we can see very wide swings in our temperature values that time of year.
It's a little different in late June and early July, when we frequently see high temperatures run pretty close to the long-term averages of upper 80s to low 90s, with the result that the standard deviation is only about +/- 6 degrees. On Friday, Saturday and Sunday however, we ran 15, 15 and 13 degrees above normal, well outside that typical variability, and of course it was enough to set some new records for the Raleigh area in the process. On Friday, the observed high of 105 matched the all time high for the area, and also set a new record for June 29th, breaking the old mark of 101 set back in 1945. Saturday brought a high of 105 again, breaking the old record of 102 from 1959, and on Sunday the high of 103 easily topped the previous record of 99 from 2005.
We caught a "break" on Monday with a high of 94, no record there, but may warm back up a little in the next few days, putting is striking distance of another record or two, perhaps. Today, for instance, we're expecting a high around 97. The record for today is only 98, though, so if we're off a degree or two on the cool side with our forecast, that could possibly be tied or broken.
Beyond that, there is some variability among models, but at least some indicate temps in the upper 90s to around 100 at times right into the weekend ahead, with some indications that by Tuesday or so of next week we may finally settle back down to a stretch of days with highs in the more "normal" range, perhaps mid 80s to low 90s then. That is illustrated in the top graph of the image shown here, which is a projection of temperatures every 6 hours from the North American Ensemble Forecast System (NAEFS). That system makes many runs from several different computer models, assigning slight "perturbations" to the starting points for the models in an attempt to mimic some of the uncertainties associated with erroneous or missing temperature, pressure and wind observations that creep into the initial analyses. The result is a spread of possible outcomes indicated by the "box and whiskers" you see on the plot. The yellow box contains the middle 50% of solutions, while the whiskers give an indication of what the most extreme (highest and lowest) model results are. Inside the box, there is also a bold line that shows what the median result was. This is often a reasonably good starting point for a longer-range forecast, but you can get some idea of how much or little confidence to have in the results by how big the spread of solutions around that median turns out to be.
For this particular case, you'll see that out through the weekend, the afternoon temperatures (at 2 PM, so the high for the day may be a little warmer than that value) are quite warm right through the weekend, a bit cooler on Monday, and then settle in notably lower through the remainder of next week - IF that holds up, we'd be seeing highs in the mid 80s to low 90s, much more in the "typical" range for this time of year than we're seeing at the moment. Here's hoping!