## 12 Years Of No Warming, Part 2 – Update on UAH, May 2009

Posted by The Diatribe Guy on May 29, 2009

UAH is basically tracking with RSS in the end conclusion of no warming for the last 12 years. It doesn’t go back quite as far as RSS – the UAH is truly at 12 years, going back to May 1997. RSS went back to March 1997. Since these are both satellite readings, they provide a nice check against the other. They are not necessarily in concert at all times. But the overall results are in line.

I won’t belabor the same old points as made in the RSS post. So, I’ll pretty much just get to things here:

Data source: UAH Data – link can be found on the right side of the page.

**Data Point:**

The April anomaly was 0.091 (in degrees Celsius. Heretofore, I’ll state as 9.1 – in hundredths of a degree Celsius). This is higher than last year by 7.6, and lower than last month by 11.5.

**Rank:**

The April anomaly ranks 15th of 31 Aprils (51st percentile) and 167th of 365 total anomaly readings (54th percentile).

**Average:**

The annual average is 12.1. This is the highest average since the year ending May 2008, and has increased each month since its low value of 3.9 in October 2008.

**Streaks:**

This is the sixth consecutive month of anomalies that are above the same month of the previous year. Unless there is a significant drop in May’s anomaly, this will likely be extended, since May of 2008 was -18.3 and June 2008 was -11.4.

**Trends:**

I have discussed the limitations of looking at a 30 year trend in the past. This is because it is apparent that cycles affecting temperature range over 60-100 years or more. I have also discussed the limitations of applying linear trends to the temperature charts. Primarily, in the UAH chart, a step is apparent around 1998. Applying separate trend lines to pre-1998 and post-1998 show two fairly flat lines with a step up at that point. This limitation is why I normally provide the charts of how the linear trend line slope changes. It provides a perfect example of why the linear trend cannot be used to extrapolate forward – it is constantly changing and cycling.

Other than the 60-month slopes, I didn’t provide that this time. So just take the linear trends presented below for what they are: a current snapshot of best-fit trend line. I understand the limitations, and so should the reader and user of such data.

*The overall UAH picture:*

*12-Year Flat Trend Line:*

We can extend a trend line back to May 1997 and show a best-fit trend line with no warming. Yes, it’s true that 1998 had a super El Nino. While that certainly acts to help explain one reason for the line, one could also argue that if you eliminate that event, subsequent temperatures may well have been cooler. You can’t simply remove that one event without making subsequent adjustments to the points that follow. Our atmosphere simply doesn’t react like that. But, in any case, the data is the data. If things are truly warming as people say, eventually they will lose credibility on the 1998 El Nino point, because even despite that we should see temperatures eventually exceed that level.

*60-month trend lines:*

The 60-month trend line is -0.20434. This is the rate of change in 0.01 C per month. This corresponds to a rate of cooling of 2.45 degrees per Century. I won’t continue to make this point, but it bears repeating once more that this represents a current rate based on 5 years of data, and that this rate changes. I certainly wouldn’t extrapolate the 5-year trend rate to a Century. But then, I wouldn’t extrapolate the 10, 15, 20, 25, or 30 year rate like that either. But, it’s the way we measure these things, so I’ll play along.

This is lower (more negative) than last month, after an increase in the previous few months. If the next four anomalies come in at the current annual average of 12.1, this slope will decrease briskly, reaching -0.42. This would be its steepest decline in quite some time, even below the levels of last year. Even at an anomaly of 20 over the next four months, it will exceed last year’s steepness (-0.369). That is because the anomalies at this point in 2004 were single digits or negative in May – Aug. So, unless things really heat up, we’re likely going to see a sharp downturn in the trend line over the next few months before moderating once again.

To show how the slope has changed over time, I present the following chart. We can expect the value to dip below the previous recent low in just a few months.

*120-month Trend:*

The current 120-month trend is +0.0965, corresponding to a rate of 1.16 degrees warming per century. This has declined for the second straight month after a period of increase. The subsequent months should show a series of declines. The 10-year trend would hit zero or negative by the end of the year with an average anomaly at or below 13.15. Whether or not this happens, it is nearly certain that we will see a sharp decrease in the slope, nonetheless. Even an average anomaly of 40 for the rest of the year woould still show a slight decrease in slope by year’s end.

*180-month Trend:*

Current slope is +0.0908, or 1.09 degrees per century. This is the lowest 180-month slope value since the period ending December 1997. By year’s end, this will likely decline to around 0.06.

*240-month Trend:*

The 240-month trend line has a slope of 0.139327, or a rate of warming of 1.67 degrees per century. This is currently at its lowest value since the period ending January 2002. I expect this to decline to a level below 0.125 by year’s end.

*300-month Trend:*

This is the steepest of the trend lines, at 0.142776, or 1.71 degrees warming per Century. It has declined, though, and is at it’s lowest point right now since the period ending January 2006. Even with anomalies above the current average for the rest of the year, it’s likely that we’ll see this slope dip to below 0.13 by year’s end.

*360-month Trends:*

No graph here, since it’s basically the same as the overall chart. The slope is 0.104705, or 1.26 degrees warming per century. It’s the lowest value of the 6 observed data points.

## jeroen said

I have a question.

By which amount is the R2 = showing a trend. I learned from other websites that it has to be bigger than R2 = 0.33

## The Diatribe Guy said

There are two issues: (1) what is the best-fit trend line, and (2) how robust is the best-fit trend line?

This creates different arguments. The best-fit trend line is what it is. I always show the r-squared on the trend line because I want full disclosure on what it is I’m presenting, warts and all. So, a low r-squared shows a trend, but the lower the r-squared, the less robust the trend is.

Basically, what a low r-squared means is that you could tilt the line up or down, run your least-squares formulas, and the results wouldn’t be necessarily all that far off from the result of the best-fit line. The longer your trend is, the higher your r-squared value would tend to be. Obviously, the lower the volatility or spread in the data, the higher, as well.

But there are also fallacies that accompany this. For example, the r-squared on the 300-month trend line is above 0.33. This suggests a good fit.

But does that mean the line is a good one? Do any of us believe that temperatures simply increased linearly over time? No, the trend lines change all the time through cyclical variations. In addition, you can greatly improve fit by splitting the data in to pre-1998 and post-1998 segments and fitting two separate, shorter trend lines.

All these things are simply ways of looking at the data. I often show charts of how slopes are changing over time to demonstrate the limitations of trend lines and drawing any long-term conclusions from them.