HOW I DO THE 5 WEEK FORECAST.

I've been doing this method since the early 1980's.  There are surprising similarities to the LRC.  The main difference is the LRC (Lezak Recurring Cycle) is believed to only set up once a year around October through November and lasts for 10-12 months. 

I believe that may be possible, however I reset my theory about 4 times a year generally about the seasonal change because I believe events can occur to alter the pattern.

Around the week before Thanksgiving, I look at the weather for the previous three months and go back in time and try to find a year that most closely matches that pattern.  When I'm really lucky, I'll find more than one year and I'll combine the two. 

For LRC fans, the idea of the cycle is unique to every year, but it doesn't mean a similar pattern hasen't occured some other year.  That's why if I can find two other similar years, that only supports my theory.  Whatever those years are, the LRC will have already run it's course and that year too, cycles.  It gives me an advantage in that I can already see what should happen rather than guess on "x" amount of days in a cycle.

I reset my page about Thanksgiving, Easter, July 4th and Labor Day.  Sometimes I don't have to change anything and sometimes it changes radically.  Last year for example, I stuck with the same pattern from the 4th of July all the way to Thanksgiving and didn't have to change much of anything until that pattern died in mid November and that pattern won me the KSHB Summer Temperature Forecast in 2009 [and note that I haven't entered any contests since ].

I then put that pattern on my visual page.  Here's where it may seem complicated.

Usually twice a day, morning and afternoon, I look at the NWS forecast and run all the GFS and NAM maps usually between 4 and 4:30 AM.  The first 5 days are usually straight from the NWS and if my pattern was correct, I would have had to change very little.  From 5-15 days out, I may change the precip chances for any given day based on the GFS and NAM maps. 

I use 4 different GFS maps and average the precentage probabilities for each day for both precip and temperatures.  Let's say for a given day they average to 21% precip.  Let's say on that day, I already had a 19% precip.  I will raise my 19% to 20%.  Sometimes the GFS maps average on a given day to 90% where I only had 20%.  Even then, I'll only raise my figure to 21%.  I don't always trust the GFS maps, I only use them as a guide.  They are the road so to speak, but I'm the driver.

Now, within 5 days out, I'll split the difference (if needed).  Say 4 days out I had 30 for a high and 19 for a low, but the NWS has 35 for a high and 15 for a low, then I'll post 32 and 17.  Same with the precip chances. 

GaryB