I was playing with so simple linear regressions more to aggravate some of the minions than to really accomplish anything. I am pretty positive there is a fairly larger range of undecipherable climate noise. Some might call it chaos, but any simple control system has a range around a set point. If you want to reduce that range you are looking at big bucks and potential instability. It is simpler and more cost effective to "live" with a reasonable amount of fluctuation unless you have a very temperamental process. Having worked with "novel" new control schemes like PDI feed forward, I have witness some spectacular initial failures. Most of these types of control systems are viable now with more and faster computing power, but for simple things like HVAC control at the time they were a huge waste of money. The average person cannot sense a temperature change of less than 2 F and there is no "ideal" temperature range for all people so a half degree to degree of "slop" for indoor air temperature control is perfectly acceptable. Even a larger range of humidity control is acceptable and with proper system sizing, pretty much takes care of itself.

There are cases where more precision is required and the cost is justified, but proper load sizing as staging can reduced the complexity required for even those cases. Simple isn't a bad thing.

What I did just for grins is a simple linear regression of GISS global with and uncertainty channel. I just eyeballed the fit and was pretty happy with 0.5 C which happens to be 1.6 sigma for the GISS monthly data.

A little light smoothing, 13 month moving average, provides a one sigma, 0.29 C eyeball fit. Since only five years of the 134 years total make it outside the channel, that is about a 95% uncertainty range in spite of the one sigma notation which implies about 68% uncertainty. So in my opinion, with about a 0.3 C range global climate is pretty stable. You can get the same fit with climate model runs with about +/- 0.35 C, mainly because of the volcanic forcing misses. Smooth those out and you can get a more respectable error range for the high dollar estimates.

I then spliced the global with the Oppo et al. 2009 reconstruction, but had to kick the range back up to 1.6 sigma to get a fit back to about 1780. After throwing in a pre-Little ice Age error range of +/- 0.29 C, I have a basic model of expected climate. The difference in the slope of the dark green LIA recovery and the blue instrumental is roughly "potential" Anthropogenic Climate Change. If you used the green regression and added CO2 forcing with a 0.8C "sensitivity" you could fit the two curve pretty well. Of course, the Oppo 2009 and instrumental have different smoothing and picking the slope of LIA recovery is a bit of a guess, but it is in the ball park.

Since I am fairly certain that tropical convection and cloud cover trigger around 28C to produce significant negative feedback, the intersect of the linear regression and pre-LIA range are a pretty good illustration of what I expect climate to do. That would be crab sideways with a fluctuation of around +/-0.5 C degrees. So this is my prediction of the month.

If you are an ENSO fan, the projection would be 26.9 C +/- 0.5 C degrees. Since some believe that ENSO, PDO and AMO are the sources of all variability, that might make them happy. Personally, for global climate I would watch the tropics in general not any region in particular. I have done a number of correlations between various tropical bands and "global" temperatures and the 30S-30N band is in the 70% and 80% range since it has over 50% of the total ocean area.

I haven't figured out a real "sciency" way to fit the sea level rise into the picture, but roughly matching instrumental, SLR and Oppo 2009 looks like this. Since there is more than just thermal expansion involved with SLR, a convincing fit with rational error bars will be quite a challenge. I do have some thoughts on the subject, but that will take some real work which is against my current religion.

So there is the title of the post, "circumstantial" evidence of a secular trend related to LIA recovery.

## No comments:

## Post a Comment