Hello.

This year I decided to make a seasonal "outlook" of my own. I guess its a statistical forecast, because it is based on reanalysis. This is basically just a composite or a blend of the winters in the past. I used certain winters and blended them together. I have chosen the years and months, based on the overall global setup and similarities in the oceanic and atmospheric features. The graphics were made on October 29, when I finished my analysis process and first published on November 1st on a weather forum in my country. So lets say I have 1 month "lead time" before the start of the forecasted period. This is my first attempt of "statistical" forecasting, so I don't really expect much success to be honest, but I need to test it, so i decided to make a "public test", by posting it on different weather forums.

Some interpretation is required. This is basically an average picture, like an ensemble mean, so it has its respected deviation and variance. But just the average or "middle" picture will suffice for my first attempt. I posted this on a forum in my country, so the text on the graphic is in my language (Slovenian). But some words, like the names of the months are similar, so I think you will understand what period is represented on the graphic.

I am only aiming at the DJF 500mb geopotential height anomaly for the verification in March, when the forecasted period is over.

Now the graphics. Keep in mind that this is unaltered in any way. And it was made in the end of October, when none of these features were present in the seasonal models. At least not in this magnitude. I also have to point out, that this graphics only represent a large scale view.



Corresponding precipitation anomaly:


And temperature anomaly:


And I will also add the December 500mb height anomaly, made on the same date with the same system.


And this was CFSv2 when I made these graphics.


I first posted these images on a weather forum in my country, on November 1st. And all the graphics were made before that. Here is the link to that topic to prove it. Basically so no one can say that I "cheated", by looking at current models (which differ anyway) and intentionally making a similar picture.

Best regards.