## May 2008 Update on Global Temperature, Part 2

Posted by The Diatribe Guy on May 15, 2008

Here are a few of the charts that paint a better picture of the current trend lines in the data. As one responder notes to me, and correctly so, there is only limited value in looking at linear trends when it comes to temperature. The trends themselves are constantly changing, which is what I have discussed numerous times when explaining my predictive methodology. Given that they are constantly changing, there really is not, in fact, a linear trend. I think it’s most accurate to say that there is a very long-term trend that can be approximated by the overall historical linear trend (although, in truth, that is also not linear. The long-term trends cycle as well. But these centueis-long and millennium-long cycles can be shown fairly well with a linear trend. In fact, if I fit different polynomials, I think the result is worse because even 128 years of data is too small of a segment to get a good read on the overriding underlying long-term curve). Now, on top of that long-term, *generally* linear trend, we have other cycles that appear. I call them cycles, and not trends, because I do not believe the last 30 years was any long-term trend, just as the 30 years before that was, etc.

This can really be seen fairly clearly from the overall historical chart. I think it is legitimate to say there is a natural increasing temperature, but at a fairly benign rate of 0.56 degrees Celsius per Century. This will not go on into perpetuity, but it is likely to continue forward in our lifetimes. Perhaps the rate is slowing a tad, or increasing a tad, but it will not accelerate or decelerate on a dime. Just my opinion.

The long-term trend is shown below. The R-squared is actually fairly good. In fact, it is even better than the 30-year trend we hear so much about (not shown, that r-squared is around 0.5):

So, if I just said that only the long-term trend is the real trend, and all the other “trends” are really cycles, then what’s the deal with the other trend lines? Well, the trend lines do give us a good indication regarding the direction we’re moving in, at least in the short-term. I think there is definite value in presenting the data in terms of a linear trend, if for no other reason than purposes of understandability. But one of the things I want to stress is that my use of the trend lines goes beyond a simple projection of the current line into the future. There is little basis in that. What I do is watch how that line is changing from period to period, and then I incorporate a model that looks for weighting patterns in those changes to project the future trend line changes. From there, then, I back into the anomaly. So, when I present the following trends, it is simply the latest trend. The more important trend line, in my opinion, is the trend of the slopes, which tells us how those slopes are changing over time. (In my model I go one step further and calculate the rate at which the trend of slopes is changing).

The first chart is really just a point of interest. How far back into the data can we go and still calculate a cooling trend line? While the r-squared fit is weak, it is still the best fit. The longer this goes on, the more diffficult it is to say that we are continuing to warm. That graph is below:

As you can see, the current cooling trend can be traced as far back as March 2001, suing the GISS data. I may note here that other temperature measures show a cooling trend as far back as 1998.

From there, I just selected a couple of the more recent trend lines to demonstrate, in particular, what has occurred with the changing slopes of those trend lines as time has passed. Starting with the 60-month trend line:

This shows the trend fit for the last 60 months for the raw anomaly data. The r-squared is only 0.007, but there is a negative slope of -.0596 (which means -.0596 change in the anomaly each month. A unit anomaly = .01 degrees Celsius. Thus, the current rate of cooling 0.007 degrees Celsius per year, or 0.70 degrees per century. Of course, extending such a trend out for a Century would be silly, but it at least demonstrates the magnitude of the *current* trend.

More importantly, though, is how that line has changed over the last 4-5 years:

Here we see that the trend line had a slope just a few years ago of nearly 0.6. With a couple gyrations, the slopes of the 60-month trend lines have changed quite substantially to the current level. This, too, will reverse at some point. My model hopes to anticipate such reversals (though I still need more weighting points, more than likely).

Next, we take a look at the 120-month trend line:

OK, so now this tells us that the best fit trend line for the 120-month period is positive. In fact, the slope of 0.1357 tells us that temperature is changing by 0.016 degrees Celsius per year, or 1.6 degrees per Century. But are we really warming? Well, maybe… but taking a look at the way the 120-month trend line has changed, let’s look at the trend in the slope calculations:

We see a very consistent and clear downward trend in the 10-year slopes, dating back to 2002, when the peak slope was at 0.336 (which would be the equivalent of 4 degrees Celsius warming per Century… now we see where some of these wild figures come from).

At the very least, what this tells us is that warming trends have slowed. It may not have ceased, although based on many projections of the upcoming solar cycle activity, the Pacific Decadal Oscillation flip and a similar flip in the Atlantic, it does not seem as if we’ll see a major upturn in the slopes any time soon.

Referring back to the first chart, what is a simple logical expectation? Well, let’s see… temps generally above trend line (except for one major bliparound 1895) from 1880-1904 or so… Below for a few years until 1925 or so… Above until the late 1940s… Below until mid-80s… Above until now. The most recent months have seen some anomalies below the trend line. It can only be expected that we are pulling back to the line, and at some point will likely dip below.

Which kind of sucks, actually. I’m not a big cheerleader for cold weather. But I guess if it cools off some rash polcy-making, it may be worth it. Let’s just not go rooting for one of those Minimums like the Little Ice Age.

## Alex Llewelyn said

Really interesting – maybe you could do a graph to show how the trend has changed over the whole period and changes in how much points differ from the trend.

## Diatribical Idiot said

I can work on those charts. Thanks for the suggestion. I have a few other things in the works as well, including the continuing refinement of the predictor model.