I’ve moved to WordPress: http://bobtisdale.wordpress.com/

Thursday, August 28, 2008

NINO3.4 SST (Not Anomaly) - Part 3

The highlights of the first two installments of this series of posts about NINO3.4 SST (not NINO3.4 anomaly) data are:
1. There is a negative trend from 1854 to 2007 in the annual minimum NINO3.4 SST. All other SST and LST data sets I’ve checked so far have positive trends.
2. In the graphs of the temperature differences between NINO3.4 SST and hemispheric and global SST and in the graphs of the temperature differences between NINO3.4 and global LST and global combine surface temperature, there are underlying oscillations with a time span of approximately 80 years.
3. In recent years, there are also shorter time-span oscillations that mimic the solar cycle. These will now be discussed.


In part 2 of this series, I noted the oscillation in the short-term graphs of NINO3.4 SST data. These oscillations appeared to have a frequency that mimicked that of the 11-year cycle of Total Solar Irradiance (TSI). Refer to Figure 1 & 2.

Figure 2

I’ve added monthly Sunspot data (I haven’t been able to find long-term monthly TSI data) and expanded the time frame for the comparative graphs, Figures 3 and 4. In both figures, the Sunspot data has been scaled with a multiplier of 0.015 and shifted by 25 to bring it close to the same range. I have made no other changes to the data in Figure 3. Referring to the last three cycles, there does appear to be a correlation, though the lag between Sunspot Number and NINO3.4 SST changes. Prior to then, the correlation evaporates through most solar cycles, periodically lining up again. In Figure 4, I’ve shifted the Sunspot data by 48 months to bring it into line with the NINO3.4 data, but the result is no better. I’ve also added volcanic aerosols, simulated by the Mean Optical Thickness data of the Sato Index, to see if there was any cause and effect there. No help.
Figure 3

Figure 4


Does this mean that NINO3.4 SSTs are not driven by solar irradiance? No. But, on the other hand, the possible correlation between the two data sets for the last three solar cycles does not prove that it is. It simply infers that there might be a link and that it’s something that needs to be investigated further. It could be that NINO3.4 SSTs are also influenced by other variables (very likely) and that the influence of these other variables changes with time (possible). Or it could also be that NINO3.4 SST is responding to changes in “apparent” solar irradiance of the surrounding Pacific Ocean, with “apparent” solar irradiance also taking into consideration the variations in solar irradiance reaching the ocean surface caused by volcanic eruptions, cloud cover, etc. Or it could be that there are different oceanic time lags that are reflected in the NINO3.4 SSTs. Or there might have been a change in SST sampling methods that impacted the earlier results, causing the correlation to be dampened. (NINO3.4 data before the opening of the Panama Canal in the 1910s have a much lower sampling number than after its opening, and there have been many recent papers about the effects of different sampling methods--insulated versus non-insulated buckets versus engine intake--on SST data.) Or it could be a combination of the above and other unnamed factors.


Sea Surface Temperature Data is Smith and Reynolds Extended Reconstructed SST (ERSST.v2) available through the NOAA National Operational Model Archive & Distribution System (NOMADS).

SATO Index Data is available at:

The Monthly Sunspot Data is available here:

Tuesday, August 26, 2008

NINO3.4 SST (Not Anomaly) - Part 2


In the first post of this three-part series that examines NINO3.4 SST (not NINO3.4 anomaly) data, I illustrated the differences between the NINO3.4 SSTs and the SST anomalies and showed an unusual aspect of NINO3.4 SST data (other than the magnitude of its variations) that differentiates it from other SST data sets. That notable difference in NINO3.4 data is a negative trend from 1854 to 2007 in the annual minimum NINO3.4 SST. In other words, the annual minimum NINO3.4 SST has decreased over the period of the instrument temperature record. In contrast, of the many data sets I’ve checked, the minimum annual temperatures of all SST or LST data sets have increased over the past 150+ years.

In this the second post of the series, I illustrate the temperature differences (Delta T) between NINO3.4 SST data and SST data of the northern and southern hemispheres and between NINO3.4 SST and global land surface and global combined (land plus sea) surface temperatures. There are also underlying oscillations that are exposed by the Delta T graphs. These deserve more research and further explanation.


Figures 1 and 2 are noisy graphs of the temperature difference (Delta T) between NINO3.4 SSTs and Northern and Southern Hemisphere SSTs. Both graphs are of monthly SST, and in both graphs, the data has not been filtered. Figure 1 is the longer-term data, covering the period of January 1854 to July 2008. In Figure 2, the short-term data extends from January 1978 to July 2008. They’re provided to show the magnitude of the annual changes in Delta T.

Figure 1

Figure 2

In Figure 3, the Delta T between NINO3.4 and Northern and Southern Hemisphere SSTs has been smoothed with a 12-month running-average filter. There is very little apparent difference in the Delta T between NINO3.4 SST and the SST of the two hemispheres when the data is smoothed. The differences are there. Northern and Southern Hemisphere SST data sets are significantly different. It’s just that the magnitudes of the annual variations in Delta T mask those differences.
Figure 3


In Figures 4 and 5, I’ve isolated the Delta Ts between NINO3.4 SST and the Northern and the Southern Hemisphere SSTs. If you were to open both illustrations in separate windows and switch between the two, you’d note that there are minor differences between the data sets. What does stand out, however, especially in the NINO3.4 to Northern Hemisphere Delta T data, is an underlying oscillation with a length of approximately 80 years. I’ve been plotting temperature data for a number of years and I’ve never run across such an obvious oscillation of that length. There also appear to be short-term variations in trend, but it was the long-term oscillation that caught my eye. The only natural oscillation I know of with a frequency of that length is the Gleissberg Cycle, a solar cycle. But I know so little about the Gleissberg Cycle I would not even venture to comment. Also, what shows in the graph could simply be an underlying oscillation with a length that approximates that of the Gleissberg Cycle. The oscillation is apparent and noteworthy, though.
Figure 4

Figure 5

To determine if that long-term oscillation is only a product of the Delta T between NINO3.4 SST and hemispheric SST, I’ve plotted the Delta T of NINO3.4 SST and Absolute Land Surface Temperature in Figure 6 and plotted the Delta T of NINO3.4 SST and Absolute Combined (Land plus Sea) Surface Temperature in Figure 7. Both data sets are smoothed with 12-month filters. Also note that the start year for the NCDC absolute temperature data is 1880, so Figures 6 and 7 have shorter time spans than the other graphs in this post. The underlying long-term oscillation is present in those Delta T data sets as well. I’ve added a polynomial trend to Figure 7 to emphasize it.
Figure 6

Figure 7

One more data set to check: The last graph, Figure 8, is a long-term graph of the Delta T between NINO3.4 SST and Global SST. It contains a poly trend, also for emphasis. And yes, the underlying oscillation shows up in it.
Figure 8


Other than the fact that there appears to be an underlying oscillation in the calculated Delta T data sets, only one thing can be said and it’s obvious: During periods when the Delta T between NINO3.4 SST and the reference temperatures is increasing, NINO3.4 SSTs are rising faster than the reference temperature, and vice versa when the Delta T is decreasing.


Please scroll up to Figure 2. There appears to be an oscillation with an 11-year length that’s visible in the short-term Delta T graph. It’s also visible in the short term NINO3.4 SST data from the first post of this series, which I’ve duplicated here as Figure 9.
Figure 9

Don’t get your hopes up about what it will show. The correlation between NINO3.4 SST and TSI is reasonably good for the past three solar cycles, but prior to that someone with better analytical capabilities would be required to extract the noise. This assumes there is, in fact, a correlation earlier in the two data sets.


I’ve used the polynomial trends in this post simply to confirm and emphasize what I was seeing. They don’t work as well as I would like in drawing out the trends. In Figure 8, the poly trend appears to minimize the amplitude, and as you will note in Figure 9, the poly trend can also exaggerate parts of the trends. I would be happier with a different means of filtering or trending, but since I use the polynomial trends so rarely, they will suffice.


Sea Surface Temperature Data is Smith and Reynolds Extended Reconstructed SST (ERSST.v2) available through the NOAA National Operational Model Archive & Distribution System (NOMADS).

NCDC Absolute Temperature data was created from the data supplied here:

Sunday, August 24, 2008

NINO3.4 SST (Not Anomaly) - Part 1

This is the first post of a three-part series that examines NINO3.4 SST (not NINO3.4 anomaly) data. In it, I’ll illustrate the subtle differences between the SST and the SST anomaly data and show unusual aspects of NINO3.4 SST data (other than the magnitude of its variations) that differentiate it from other SST data sets.


Figures 1 and 2 illustrate long- and short-term NINO3.4 SST data. The long-term data spans the period of January 1854 to July 2008, and the short-term data, January 1978 to July 2008. Due to the volume of data in the long-term graph, Figure 1, it’s difficult to visualize any difference between it and NINO3.4 anomaly data. The short-term data, however, shows a slightly greater period of elevated temperatures during the 82/83, the 86/87/88, and the 97/98 El Nino events.

Figure 2

Figures 3 and 4 are comparative graphs of NINO3.4 SST and NINO3.4 SST anomaly data. To bring the SST data down to a similar scale, I shifted it 26.521 deg C, which was the SST value for January 1854. You’ll have to use the Tinypic links because I had to decrease the weight of the anomaly curve to assure it wasn’t obscuring the SST curve. As shown in the long-term graph, Figure 3, the SST data presents the monthly cycles over the course of a year, while the anomaly data is suppressed by its comparison to a monthly mean. In the short-term graph, Figure 4, the SSTs for the April 1992 and May 1993 are shown to be comparable to the SSTs in January 1983 and November 1997, which are the peak SSTs of the significant 82/83 and 97/98 El Nino events. Using the ONI index for reference, May 1993 is not considered an El Nino month, though its SST is similar to that of the “super” El Ninos. In the second part of this series, I’ll determine the temperature differences between NINO3.4 SST and other variables to illustrate the importance, or lack thereof, of elevated NINO3.4 SSTs like that of May 1993.
Figure 3

Figure 4


Figure 5 illustrates the annual Maximum, Average, and Minimum NINO3.4 SST from 1854 to 2007. The unusual aspect of the NINO3.4 SST data is that the trend in the minimum annual SSTs is negative. Why is that unusual? In all other data sets I’ve investigated so far, such as minimum global and hemispheric SSTs and minimum global Land Surface Temperature (LST), they all have positive trends.
Figure 5

Why would NINO3.4 minimum annual SSTs have dropped when minimum temperatures for all other data sets rose? Sorry, you won’t find the answer here.


The next two graphs provide additional views of the differences between minimum and maximum annual NINO3.4 SST. I’ll offer no additional input other than to explain the graphs.

To better illustrate the differences in trends between the NINO3.4 annual maximum and minimum SSTs, I zeroed both data sets at their 1854 values in Figure 6.
Figure 6

The last graph, Figure 7, shows the difference between NINO3.4 annual maximum and minimum SST.
Figure 7


In Part 2 of this series, I’ll examine the temperature differences between NINO3.4 SST and other SST data and between NINO3.4 SST and absolute global combined (land plus sea surface) temperatures.


Sea Surface Temperature Data is Smith and Reynolds Extended Reconstructed SST (ERSST.v2) available through the NOAA National Operational Model Archive & Distribution System (NOMADS).

Sunday, August 17, 2008

Tropical Atlantic, Indian, and Pacific Ocean SSTs

Figures 1 and 2 are comparative graphs of Tropical (20S to 20N) and Global SSTs. Figure 1 is of long-term data, smoothed with an 85-month filter. Figure 2 is of raw data from January 1978 to present. Referring to Figure 1, between its early peak in the 1860s to its minimum near 1910, Tropical SSTs dropped approximately 0.42 deg C. Then from trough (near 1910) to peak (about 2003), Tropical SSTs rose approximately 0.77 deg C. Figure 2 shows that, excluding the rise and fall of the 97/98 El Nino, Tropical SSTs have been dropping since 2003, though the decrease is erratic.

Figure 1

Figure 2

In Figure 3, another comparison, the Tropics have been divided into the Atlantic (70W to 15E), Indian (35 to 100E), East Pacific (70W to 180), and West Pacific (100E to 180). The greatest variation occurs in the Atlantic and the least variation in SST is in West Pacific. The magnitude of the changes in the Atlantic skews the perspective of other data sets.
Figure 3

Figure 4 illustrates the short-term oscillations the SSTs for the Tropical East Pacific that are attributable to ENSO and the long-term oscillations that result from thermohaline circulation/meridional overturning circulation. This curve is typical of NINO data that’s been smoothed with an 85-month filter. It reveals an overall increase in East Pacific SSTs that peaked in 1995, two years prior to the major El Nino of 97/98. The rise in Tropical East Pacific SST is difficult to determine due to the variability. Adding a 6th order polynomial curve to the data (Figure 5) illustrates the underlying trends over the term. Suffice it to say, without the ENSO variations, the overall rise in Tropical East Pacific SSTs would be less than the rise in Tropical SST for the globe of 0.77 deg shown in Figure 1.
Figure 4

Figure 5

The long-term Tropical West Pacific SSTs are shown in Figure 6. Its dip from the late 19th to the early 20th centuries preceded the dip in the Tropical SST for the globe but the rebound lagged, waiting until the late 1920s before stating to rise. That rise from the 1920s to approximately 2003 is just under that of the data for the Tropical SST for the globe shown in Figure 1. The recent downturn in Tropical West Pacific SST is visible even with the 85-month filter. Refer to Figure 7 for the short-term raw data.
Figure 6

Figure 7

The Tropical Indian Ocean SST anomalies are illustrated in Figure 8. Its rise of approximately 0.87 deg C exceeded the rise in Tropical SST for the globe of 0.77 deg C.

Off Topic Note: The dip in temperature from 1942 to 1960 does not appear to be a result of “bucket adjustments”; it looks more like a combined effect of the multiple-year El Nino in the early 40s and a THC/MOC signal. Refer again to Figures 4 and 6.
Figure 8

The greatest long-term variation in tropical SSTs is present in the Atlantic, with a whopping 1.16 deg C rise from 1905 to present. Refer to Figure 9. Although it’s over a slightly different time frame, that’s over 50% greater than the rise in the Tropical SST for the globe (1.16 deg C/0.77 deg C = 1.506). It surely is the driver of the Tropical SST data for the globe. Let’s examine it further to see what it’s comprised of.
Figure 9

Comparing the Northern and Southern Tropical SSTs for the Atlantic, Figure 10, the Southern data shows the greater variation of approximately 1.27 deg C, where the Northern data shows a 1.08 deg C increase.
Figure 10

In Figure 11, the North Atlantic and the Tropical North Atlantic SST comparison illustrates that Tropical North Atlantic SSTs are a function or blend of North Atlantic SSTs and Tropical South Atlantic SSTs.
Figure 11

Comparing the South Atlantic SSTs and the Tropical South Atlantic SSTs illustrates something entirely different. Refer to Figure 12. The variation in South Atlantic SSTs is less than that of the Tropical South Atlantic. Since the Tropical South Atlantic SSTs should also be effected by the Tropical North Atlantic SSTs, and since the variations in Tropical South Atlantic SSTs are greater than the Tropical North Atlantic SSTs, Figure 10, I would have expected that whatever is driving the Tropical South Atlantic SSTs would have a greater variation than the Tropical South Atlantic. But the South Atlantic does not.
Figure 12

Instead of segmenting the South Atlantic data more, let’s say into mid-latitude data, I first looked at the upwelling area that runs along the West Coast of Southern Africa. The Benguela Current that runs along the West Coast of Africa should transport the upwelling of water there into the Tropical South Atlantic. See Figure 13.
Figure 13

Figure 14 illustrates the SSTs of the Tropical South Atlantic and the upwelling area off the West Coast of Africa (0 to 35S, 0 to 20E), what I have identified as the South Atlantic Upwelling SST data. It has a substantially greater SST variation, highlighted by the 1.0 deg C plunge from 1900 to 1905. The upwelling area then appears likely to be the driver of the sizable variation in the Tropical South Atlantic SST.
Figure 14

To confirm this, I isolated the Tropical and the upwelling area data from the remainder of the South Atlantic, using the coordinates 20 to 60S and 0 to 70W and compared the remaining South Atlantic data to the Tropical South Atlantic data. Refer to Figure 15. The shallower decline in SSTs during the late 19th to early 20th centuries and the plateau reached in the 1970s indicate that Tropical South Atlantic is not impacted greatly by that area. The Eastern South Atlantic upwelling area off the African coast does drive Tropical South Atlantic SSTs along with Tropical North Atlantic SSTS.
Figure 15

Note that the plateau in the cyan colored curve (Mid-Latitude South Atlantic Excluding the Upwelling Area) in Figure 15 is due to the influence of the Southern Ocean.


Sea Surface Temperature Data is Smith and Reynolds Extended Reconstructed SST (ERSST.v2) available through the NOAA National Operational Model Archive & Distribution System (NOMADS).

Saturday, August 16, 2008

30+ Years of Ocean SSTs – January 1978 to July 2008

The following are graphs of SST anomalies from January 1978 to July 2008 for the:
North Atlantic
South Atlantic
North Pacific
South Pacific
Indian Ocean
Arctic Ocean
Southern Ocean

I found the following to be highlights:

- The absence of an immediate major reaction in the North Atlantic to the 97/98 El Nino.

- The flatness of the South Atlantic curve.

- The decline in North Pacific SST since 2005, though the PDO is said to have just recently "shifted".

- The relative flatness of the South Pacific curve before the 97/98 El Nino and then the rise and fall afterwards.

- The saw-toothed step changes in Indian Ocean SST after the 82/83, 86/87/88, and 97/98 El Ninos.

- The 1990 peak in Arctic SSTs (not 2007 or 2008).

- The continuous decline in Southern Ocean SSTs since the mid 80s.

I haven’t yet discovered what day of the month the SST data on NOMADS are updated, but obviously, it’s by the 16th. I will attempt to revise them monthly.

North Atlantic (0 to 75N, 70W to 10E)

South Atlantic (0 to 60S, 70W to 20E)

North Pacific (0 to 65N, 90 to 180W) & (0 to 65N, 100 to 180E)

South Pacific (0 to 60S, 70 to 180W) & (0 to 60S, 145 to 180E)

Indian Ocean (30N to 60S, 20 to 145E)

Arctic Ocean (65 to 90N)

Southern Ocean (60 to 90S)


Sea Surface Temperature Data is Smith and Reynolds Extended Reconstructed SST (ERSST.v2) available through the NOAA National Operational Model Archive & Distribution System (NOMADS).

Thursday, August 14, 2008

The Barents and Bering Seas


It seems that each time I see an SST anomaly map there’s a hot spot north of Scandinavia in the Barents Sea. Refer to Figure 1.

Figure 1

Figure 2 illustrates the Barents Sea SST anomalies from January 1854 to May 2008, smoothed with an 85-month filter. With the possible exception of ENSO, the amplitude and frequency of the SST modulations are greater than any other SST data set I’ve downloaded and graphed so far, especially with what is essentially a 7-year filter. Is the decrease in amplitude before 1930 a function of the smoothing they apply in periods of reduced samples? Or does the increase in amplitude after 1930 result from a reduction in Arctic sea ice?
Figure 2

Figures 3 and 4 show a decrease in summer, spring, and annual Arctic ice extent since the 1950s. For the winter and autumn sea ice extent, the declines began in the 1970s and 80s. (For those concerned about the “hockey stick” appearance of these graphs, sea ice extent is limited my North Atlantic and North Pacific Ocean currents. These graphs do not illustrate ice thickness.) Assuming that some of the sea ice loss was in and around the Barents Sea, would less sea ice there have allowed larger variations in SSTs?
Figure 3

Figure 4

Is this confirmed by the fact that in recent years, the Arctic Oscillation (AO) appears to have had more of an impact? Refer to Figures 5 and 6. Like the Barents Sea SST anomalies, the AO has changed its amplitude after 1950 as well. And it appears as though, as time progresses, Barents Sea SST responds more to a change in the AO. Is this also a function of continued sea ice loss, or did other influences, such as ENSO and the AMO, cause the difference? (I will not be comparing ENSO and AMO in this post.)
Figure 5

Figure 6


On the Pacific Ocean side of the Bering Strait lies the Bering Sea. Figure 7 illustrates its SST anomalies smoothed with an 85-month filter. The amplitude of the oscillation is remarkable.
Figure 7

As shown in Figure 8, the Bering Sea SST anomalies follow the general trends of the North Pacific.
Figure 8

But they also contain a varying portion of ENSO. Refer to Figure 9.
Figure 9


AMO Data (1899 to 2002)

AMO (1950 to 2007)


Data is Smith and Reynolds Extended Reconstructed SST (ERSST.v2) available through the NOAA National Operational Model Archive & Distribution System (NOMADS).


Tips are now being accepted.

Comment Policy, SST Posts, and Notes

Comments that are political in nature or that have nothing to do with the post will be deleted.
The Smith and Reynolds SST Posts DOES NOT LIST ALL SST POSTS. I stopped using ERSST.v2 data for SST when NOAA deleted it from NOMADS early in 2009.

Please use the search feature in the upper left-hand corner of the page for posts on specific subjects.
NOTE: I’ve discovered that some of the links to older posts provide blank pages. While it’s possible to access that post by scrolling through the history, that’s time consuming. There’s a quick fix for the problem, so if you run into an absent post, please advise me. Thanks.
If you use the graphs, please cite or link to the address of the blog post or this website.