There has been a lot of discussion about my recent post titled 4 of 4 Global Metrics Show Agreement in Trends. I had posted that in response to his post saying that 3 of 4 global metrics show nearly flat temperature anomaly in the last decade. I had previously looked at how the mean global monthly temperature tracked well between the four major metrics. I’ve included that figure in this post too because it shows how remarkably well these four temperature anomalies track each other. Thus, when I read that he found the GISS trend to be different than the other 3, I was skeptical.

The major difference between these global temperature metrics is that GISS has a lower anomaly during the big 1998 El Nino event. This may be because GISS attempts to account for polar temperatures and the other groups do not. HadCRUT, RSS, and UAH do not have polar data, and thus any temperature they report may be biased towards temperature fluctuations in the tropics and mid-latitudes. When there is a positive ENSO event in the tropical Pacific, these three groups may tend to over-estimate its influence on the global temperature.

This plot shows how the current linear temperature trend varies with the number of years used to make that trend. There is a dip at around 10 years. This is exactly during the El Nino event in 1998. We see that the GISS trend does not drop as much as the other three, perhaps because of the way they handle the polar regions. However, it is still obviously being influenced by the warm tropical waters. In the context of global climate change (global warming), this fluctuation in the temperatures is unimportant; we don’t care how temperatures vary on the ENSO time-scale (3-7 years).

Another Temperature Correction

To look at how to correct the temperature for this, I first looked at the SOI (southern oscillation index) to get an idea how long this ENSO event was in duration. In the top panel of the plot below, the black line is the SOI, there is a redish link directly over-laid on the SOI during the time period I considered to be the El Nino event. To determine the duration, I first took a running Gaussian filter to this section of the data (with a full width-half max of about 3 months). This is the dark blue line in the plot. I then subtracted the smoothed line from the original line to get the light pink line. Sorry about the color. It looked better on my screen.

The temperature correction was computed by using the “shape” of the smoothed SOI during the El Nino event. Smoothing was again done on the edges to make the corrections align to the original temperature data. The result can be seen in the bottom panel in the above figure. The very light pink is the original temperature during the El Nino event, the navy blue being the correction, and the orange represents the final corrected temperature data.

This has implications on the temperature trends longer than 9 years. Using the same analysis as used before, I calculated the current temperature trend as a function of number of years used. Before, we saw a sharp dip in the temperature trend because of the positive temperature spike of El Nino when 10 years of data were used to calculate the trends. After the anomalous ENSO event is corrected, there should no longer be a dip.

Here we have the same plot as the second figure, but I’m just looking at temperatures from GISS. The black line represents the original GISTemp trends, and the cyan line represents the corrected temperature trends. It is fairly obvious when the El Nino correction starts - when about 9 years are used to calculate the trend. The corrected temperature trend is fairly flat when using at least 10 years in the calculation.

Conclusions

Because of the effects of ENSO “noise”, temperature trends calculated over short time intervals may be biased. This is especially true if the ENSO events occur near the edges of the intervals over which the trend is calculated. For the current date, a strong El Nino occured at the start of the ten year trend and we are currently in a strong La Nina. The combination of these effects strongly biases these calculations towards lower trends.

(No Ratings Yet)

 Loading ...

Related Posts:

Trackback URI |