Long-Term Monthly NINO 3.4 SST Data

The majority of ENSO SST-based data available online is anomaly data that has been normalized, standardized, smoothed, filtered, or combined with other variables. I went searching for and found what I believe to be unadulterated NINO 3.4 SST data. It’s available from NCDC.

Main Time Series Page:
http://www.cdc.noaa.gov/Pressure/Timeseries/

NINO 3.4 Time Series Page:
http://www.cdc.noaa.gov/Pressure/Timeseries/Nino34/

NINO 3.4 SST Data from 1871 to Present:
http://www.cdc.noaa.gov/Pressure/Timeseries/Data/nino34.long.data

There is nothing in their description to indicate anything other than raw data based on the reconstructions or composites noted.

Figure 1a illustrates the raw monthly NINO3.4 SST data for a little more than 137 years. There’s a minor increase in temperature, using the linear trend line as reference. Replacing the linear with a 6th order polynomial trend, NINO3.4 temperature has a slow underlying oscillation and it peaked in the early 1990s. Refer to Figure 1b. Only a 3rd order polynomial trend shows a minor upswing toward the end of the data series, not illustrated.


http://i25.tinypic.com/2588hhe.jpg
Figure 1a: NINO3.4 SST - Monthly - January 1871 to March 2008 – Linear Trend

http://i32.tinypic.com/4js010.jpg
Figure 1b: NINO3.4 SST - Monthly - January 1871 to March 2008 – Polynomial Trend


What also stands out in the monthly NINO3.4 SST data is the increase in the range of extremes in later years, the higher amplitude. But if NINO3.4 reflects changes in solar irradiance, doesn’t that seem to agree with the increasing amplitude in the solar cycle. Refer to Figure 2.

http://i29.tinypic.com/25tjkmo.jpg
Figure 2: Solar Irradiance – Lean + ACRIM - 1880 to 2007


For those who don’t know where the NINO 3.4 region is located, I’ve provided a map with it highlighted in Figure 3. I’ve always found it remarkable that a 1 to 2 degree temperature swing in the SST of that relatively small section of the planet, and in the other adjoining NINO areas, can create havoc with global temperature and precipitation.

http://i26.tinypic.com/25ibqlh.jpg
Figure 3: NINO 3.4 Region

The annual maximums, minimums, and average NINO3.4 data for each year from 1871 to 2007 are shown in Figure 4.

http://i29.tinypic.com/fp1h5c.jpg
Figure 4: NINO3.4 SST Annual Maximum-Minimum-Average 1871 to 2007

Based on the linear trend in Figure 5, annual average NINO3.4 SST rose less than 0.3 deg C over 136 years. Other than that the average temperature graph appears similar to most representations of NINO3.4 temperature.

http://i32.tinypic.com/9u7zpx.jpg
Figure 5: NINO3.4 SST Annual Average 1871 to 2007

Again using the trend lines as reference, annual maximum NINO3.4 SST, Figure 6, rose while annual minimum NINO3.4 SST, Figure 7, was flat.

http://i32.tinypic.com/2h3shw2.jpg
Figure 6: NINO3.4 SST Annual Maximum 1871 to 2007

http://i28.tinypic.com/ng3thc.jpg
Figure 7: NINO3.4 SST Annual Minimum 1871 to 2007

This of course creates an increase in the difference between annual temperature extremes, as illustrated in Figure 8.

http://i27.tinypic.com/nno8iw.jpg
Figure 8: NINO3.4 SST Maximum minus Minimum 1871 to 2007

I hesitated before posting the next graph, because correlation might be in the eye of the beholder. In other words, I may be reading too much into what appears to be a possible influence of solar irradiance on Minimum NINO3.4 temperature. Figure 9. To scale the TSI, I used the following equation.

Figure 9 TSI Scale = (TSI-1351.8)*1.8

Someone with filtering capabilities may be able to extract a better visual.

http://i27.tinypic.com/2a98mtc.jpg
Figure 9: NINO3.4 SST Minimum vs TSI Scaled – Lean + ACRIM 1871 to 2007

A Fresk Look At NCDC Absolute - Data Source Notes

Update: I was advised yesterday, May 16th, that JunkScience corrected the errors described in the following.

In the text of the first part of this series, I wrote that I originally downloaded the .csv files from JunkScience. I have not noted that I had subsequently downloaded the NCDC monthly anomaly data found here:
http://www.ncdc.noaa.gov/oa/climate/research/anomalies/anomalies.html#anomalies


Then followed the NCDC directions for creating the absolute data:
http://www.ncdc.noaa.gov/oa/climate/research/anomalies/anomalies.html#means


There are no differences in the global land and sea and the global sea data if you use the JunkScience spreadsheets or create your own, but there are insignificant differences in the January and February mean land temperatures. This is apparently caused by an NCDC change in base monthly mean temperature data for January and February (two of those twelve monthly numbers). Since these differences in the monthly mean land temperatures effect the data for each year, and since they are so small in comparison to the annual variations in mean land temperatures, there is no problem using the JunkScience spreadsheets for simple blog evaluations such as these. The graph resolution can’t pick up the differences. However, if you’re going to process the data using the multitude of available filters, you would be wise to follow the NCDC directions and create your own absolute global temperature data.

That way you won’t have to go back and update your work halfway through your analysis or, worse, after you’ve created all the graphs and uploaded them!

This morning I notified JunkScience of their need to update the Land data. I was later notified they will post a note on the webpage and correct the problem shortly.

A Fresh Look at NCDC Absolute Part 3

Please refer to the notes on data source prior to downloading the JunkScience .csv files.
http://bobtisdale.blogspot.com/2008/05/fresk-look-at-ncdc-absolute-data-source.html


GLOBAL LAND TEMPERATURE


In my view, I saved the best for last.


Figure 3.1 illustrates the large span of annual global Land Surface Temperature (LST) and the significant changes in maximum and minimum values.


http://i32.tinypic.com/35a9ch4.jpg
Figure 3.1: NCDC Absolute Global Temperature – Land – Jan 1900 to Mar 2008




Figure 3.2 shows the maximum, minimum, and average readings of each calendar year from 1880 to 2007, with all the interconnecting data removed. The difference between the minimum and maximum values is so great it hides many of the highlights in those variables.

http://i30.tinypic.com/104s3e8.jpg
Figure 3.2: NCDC Absolute Annual Global LST – Maximum, Minimum, Average – 1880 to 2007

The average absolute annual global LST, Figure 3.3, differs little from the standard anomaly representations of LST.

http://i32.tinypic.com/2rwu835.jpg
Figure 3.3: NCDC Absolute Annual Global LST – Average – 1880 to 2007

Like Figure 1.8 in the first part of this series, the curve of the annual minimum LST data, Figure 3.4, is much more linear; that is, the multiple positive and negative trends are much less pronounced, and they may have shifted. Again, using the linear trend as reference, it’s relatively straight from 1890 to 2007, with a dip between 1960 and 1980.

http://i25.tinypic.com/1jl1m0.jpg
Figure 3.4: NCDC Absolute Annual Global LST – Minimum – 1880 to 2007

Again, that dip in Annual Minimum Global LST seems to correlate with the 1960 to 1980 drop in TSI associated with Solar Cycle 20. Refer to Figure 3.5.

http://i25.tinypic.com/2n6vgxu.jpg
Figure 3.5: TSI and Maximum-to Maximum Trends

The Maximum Global LST data holds such a surprise that I first want to illustrate the data from 1880 to 1997. Refer to Figure 3.6a and Figure 3.6b. Nothing out of the ordinary in either graph.

http://i32.tinypic.com/30t2ouq.jpg
Figure 3.6a: NCDC Absolute Annual Global LST – Maximum – 1880 to 1997


http://i29.tinypic.com/2j10jsm.jpg
Figure 3.6b: NCDC Absolute Annual Global LST – Maximum with Polynomial Trend – 1880 to 1997

Now, let’s add the 1998 through 2007 data. Refer to Figure 3.6c and Figure 3.6d. Look at that step change. I got a little carried away with the red trend lines in Figure 3.6c, but the transition in 1998 is visible in Figure 3.6d. I’ve marked 1997 with a red dot. Based on the linear trend lines, maximum annual global LST jumped approximately 0.6 deg C after the 1997/98 El Nino and stayed there. Hopefully, the current La Nina will draw that back down. Time will tell. Refer also to “Let’s Not Forget The Other Influences” that follows, prior to the closing.

http://i28.tinypic.com/11jp09d.jpg
Figure 3.6c: NCDC Absolute Annual Global LST – Maximum – 1880 to 2007

http://i26.tinypic.com/2j5ym11.jpg
Figure 3.6d: NCDC Absolute Annual Global LST – Maximum with Polynomial Trend – 1880 to 2007

The long-term effects of the 1997/98 El Nino are also visible in Figure 3.7.

http://i27.tinypic.com/11tx3dz.jpg
Figure 3.7: UAH MSU Temperature Anomalies 12/78 to 2/08 – Global, Northern Hemisphere, North Pole

The difference between global LST annual extremes (maximum minus minimum) decreased over time. Refer to Figure 3.8. This is no surprise.

http://i32.tinypic.com/vzke1h.jpg
Figure 3.8: NCDC Absolute Annual Global LST – Maximum minus Minimum – 1880 to 2007

Unexpected: the greatest rate of change occurred from 1890 to 1920, where minimum temperature rose considerably faster than maximum, and that from 1980 to 2000, the rates of change were almost identical. To create that graph, I shifted both maximum and minimum data until their polynomial trends nearly overlapped in the late part of the 20th Century. Refer to Figure 3.9.

http://i30.tinypic.com/zl94lf.jpg
Figure 3.9: NCDC Absolute Annual Global LST Annual Maximums and Minimums Shifted for Comparison of Polynomial Trends – 1880 to 2007

RETURNING TO THE LONG-TERM EFFECTS OF THE 97/98 EL NINO

Figures 3.10a through 3.10d further illustrate the long-term effects on LST of the 1997/98 El Nino. Figure 3.10a compares average LST of the 10 years before, after, and including 1997/98. Figure 3.10b illustrates the difference in the LST average of those two decades. Since the 1997/98 El Nino years could impact the averages, Figures 3.10c and 3.10d repeat the illustrations with the years 1997 and 98 removed.

http://i27.tinypic.com/2epui42.jpg
Figure 3.10a: Average Monthly LST of 1988-97 and 1998-07 – Includes El Nino Years of 1997/98
http://i25.tinypic.com/n4xo40.jpg
Figure 3.10b: Difference Between Monthly LST Averages of 1988-97 and 1998-07

http://i30.tinypic.com/xe343p.jpg
Figure 3.10c: Average Monthly LST of 1988-96 and 1999-07 – Excludes El Nino Years of 1997/98

http://i32.tinypic.com/skzu2r.jpg
Figure 3.10d: Difference Between Monthly LST Averages of 1988-96 and 1999-07

In Figures 3.11a through 11d, I replaced LST with SST to see if there was an oceanic step response to the 97/98 El Nino. It’s there, and if we assume there are no other heat sources for sea surface temperature, big assumption (Refer to “Let’s Not Forget The Other Influences” that follows), it’s on the order of 0.16 deg C.

http://i28.tinypic.com/2whl3yr.jpg
Figure 3.11a: Average Monthly SST of 1988-97 and 1998-07 – Includes El Nino Years of 1997/98

http://i32.tinypic.com/2dmfudy.jpg
Figure 3.11b: Difference Between Monthly SST Averages of 1988-97 and 1998-07


http://i29.tinypic.com/29wph5l.jpg
Figure 3.11c: Average Monthly SST of 1988-96 and 1999-07 – Excludes El Nino Years of 1997/98


http://i29.tinypic.com/121s8sl.jpg
Figure 3.11d: Difference Between Monthly SST Averages of 1988-96 and 1999-07

LET’S NOT FORGET THE OTHER INFLUENCES

Scaffeta and West in their March 2008 opinion titled “Is Climate Sensitive to Solar Variablility?”, published in “Physics Today”, produced Figure 3.12. This illustrates that more than a 0.1 deg C portion of the rise from 1997 to 2002 could result from solar irradiance. However, solar peaked in 2002, while maximum LST have remained high.

http://i30.tinypic.com/2w65c2e.jpg
Figure 3.12: Global Surface Temperature Anomaly and Phenomenological Solar Signatures – ACRIM (Red) vs PMOD (Blue) –Scaffeta and West (2008)

The Atlantic Multidecadal Oscillation rose from the mid-70s to 2005, but it too has been dropping since. The Pacific Decadal Oscillation (PDO) and its basin-wide kin the Interdecadal Pacific Oscillation (IPO) have been decreasing spasmodically since 1997.

In addition to the 1997/98 El Nino, there have also been three El Nino episodes since: 02/03, 04/05, 06/07. These were originally countered by the La Nina episodes 98/99, 99/00, 00/01 prior to the significant La Nina we are presently experiencing.

Regardless, these variables cannot explain that large step change in maximum LST without the 97/98 el Nino.

CLOSING

The preceding three-part evaluation of NCDC Absolute Global Temperature Data exposed relationships and natural effects hidden by anomaly data. I hope, and I’m sure, this will lead to other finds that further detail the major impacts of natural causes of climate change. I’ve limited this investigation to graphics available with my very basic working knowledge of EXCEL. Those with more advanced statistical devices, such as Hodrick-Prescott (HP) filters and filtering processes available through WoodforTrees http://www.woodfortrees.org/ , should be able to prompt more.