My climate forecasts, from Sep.2017

They were made by analysing 20 years of montly global temperature,
and it forecasts 15 months from September 2017, marked by the yellow line.
The most likely forecast is the one on top, while the other ones are successively less likely.

As you can see, the forecasts are different, yet build on the same data.
This is the unpredictability of the future. The laws of physics allow many possible futures, and these are some likely ones.

I got the temperature data at the beginning of September, and the calculations took about a month, so I presented them at the beginning of October.

My methods dig very deep to find patterns.
The variability in the forecasts show that there are not much pattern to find, in these data.

And this lack of patterns has been reported by many researchers of climate. They say it is close to a random walk. That it is chaotic. Generally fairly unpredictable. And their analyses seem quite sound to me.

So, I decided to test this for myself, using totally different methods for finding patterns.

And yes, there really are not much pattern at all. Quite similar to random walks.

5 month later, Feb.2018:

The prediction actually predicted!
That I did not expect, because it looked too random. The red lines in the background are the real new global temperatures. My black lines follow the current winter spike fairly well, with the downslope. The first month under the yellow line was weirdly off, but the other 4 were quite good.

March 2018:

Yep, predicted this month too. 5 months predicted in advance now! But I really do not expect this to continue.

You may contact me politely:
kim.oyhus@gmail.com

Changelog:

2017.10.2
Put it on the Web.