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

Monday, August 16, 2010

An Introduction To ENSO, AMO, and PDO -- Part 2

I’ve moved to WordPress.  This post can now be found at An Introduction To ENSO, AMO, and PDO — Part 2
An Introduction To ENSO, AMO, and PDO – Part 1 provided a detailed description the El Niño-Southern Oscillation (ENSO). This post, Part 2, discusses the Atlantic Multidecadal Oscillation (AMO), its impact on Northern Hemisphere surface temperatures, and a disagreement on the cause. The Pacific Decadal Oscillation (PDO) is discussed in An Introduction To ENSO, AMO, and PDO -- Part 3.

The NOAA Earth System Research Laboratory (ESRL) Atlantic Multidecadal Oscillation webpage refers readers to the Wikipedia Atlantic Multidecadal Oscillation webpage, so we’ll start there. Wikipedia initially defines the AMO as, “The Atlantic multidecadal oscillation (AMO) is a mode of variability occurring in the North Atlantic Ocean and which has its principal expression in the sea surface temperature (SST) field.” In other words, it’s a variation in the sea surface temperatures of the North Atlantic Ocean.

Figure 1 compares Global and North Atlantic Sea Surface Temperature anomalies from January 1870 to May 2010. The data has been smoothed with a 37-month filter to reduce the signal noise. The Sea Surface Temperature anomalies of the North Atlantic appear to exaggerate the rises and falls of the global data. Smoothing both datasets with a 121-month filter, Figure 2, helps to show the extra variability of the North Atlantic.

Figure 1
Figure 2

Wikipedia continues, “While there is some support for this mode in models and in historical observations, controversy exists with regard to its amplitude, and in particular, the attribution of sea surface temperatures in the tropical Atlantic in areas important for hurricane development.” Hurricane development will not be discussed in this post. And the phrase “some support” does not display a high level of confidence.

Back at the ESRL Atlantic Multidecadal Oscillation webpage, they describe how they calculate the Atlantic Multidecadal Oscillation (AMO) data. Basically, they detrend the sea surface temperature anomalies of the North Atlantic. To detrend the North Atlantic Sea Surface Temperature anomalies, the monthly values of the linear trend are subtracted from the North Atlantic SST anomalies. Refer to Figure 3.
Figure 3

Note the ESRL webpage also provides the AMO data smoothed with a 121-month filter, so my use of a filter of that length is not unusual.

The ESRL uses Kaplan SST data for their AMO dataset. In this post, I’ve used HADISST data because it is available through the KNMI Climate Explorer from 1870 to present. The Kaplan data there ends in 2003, and, therefore, I would not have been able to create many of the graphs in this post that run to May 2010 using it. There are some minor differences between the Kaplan and HADISST-based presentation of the AMO, Figure 4, but they will have no effect on this post.
Figure 4

Wikipedia further defines the AMO: “The AMO was identified in 2001 by Goldenberg et al and named the ‘Atlantic multidecadal mode’.”

So the AMO has been studied for less than a decade.

Wikipedia continues, “The AMO signal is usually defined from the patterns of SST variability in the North Atlantic once any linear trend has been removed. This detrending is intended to remove the influence of greenhouse gas-induced global warming from the analysis.

The detrending has been discussed. The assumption in the second of those two sentences is that anthropogenic greenhouse gases have a measurable effect on Sea Surface Temperatures. Big assumption, considering that longwave radiation can only penetrate the top few millimeters of the ocean surface.

Wikipedia further states, “However, if the global warming signal is significantly non-linear in time (i.e. not just a smooth increase), variations in the forced signal will leak into the AMO definition. Consequently, correlations with the AMO index may alias effects of global warming.”

Let’s discuss the linear versus non-linear global warming signals. Global and North Atlantic SST anomalies were compared in Figures 1 and 2. In Figure 5, the difference between the two datasets is also illustrated, using the 121-month filter, with the Global SST anomalies being subtracted from the SST anomalies of the North Atlantic. The global signal is non-linear, but the difference between the Global SST anomalies and North Atlantic data (the green curve) shows multidecadal variability.
Figure 5

Figure 6 compares the AMO and the difference between the Global and North Atlantic SST anomalies. The two signals show similar variability over the term of the data, but the changes in the AMO data have higher amplitudes. The difference between the Global and North Atlantic data also flattens from approximately 1975 to 1990, indicating that the global and North Atlantic SST anomalies are changing at the same rate during that period. This is not captured by the AMO data.
Figure 6

RealClimate defines the Atlantic Multidecadal Oscillation (“AMO”) as, “A multidecadal (50-80 year timescale) pattern of North Atlantic ocean-atmosphere variability whose existence has been argued for based on statistical analyses of observational and proxy climate data, and coupled Atmosphere-Ocean General Circulation Model (“AOGCM”) simulations. This pattern is believed to describe some of the observed early 20th century (1920s-1930s) high-latitude Northern Hemisphere warming and some, but not all, of the high-latitude warming observed in the late 20th century. The term was introduced in a summary by Kerr (2000) of a study by Delworth and Mann (2000).”

“[S]ome, but not all” is not very helpful. Let’s start by comparing the SST anomalies of the North Atlantic to the Global SST anomalies that have had the North Atlantic data removed, Figure 7. To remove the North Atlantic SST anomalies, we’ll assume the surface area of the North Atlantic represents 15% of the global oceans. The data starts in 1975 to capture the “warming observed in the late 20th century”. As illustrated, the linear trend of the North Atlantic SST anomalies from January 1975 to May 2010 is approximately 3.4 times the linear trend of the remaining global oceans. In other words, the additional rise of the North Atlantic SST anomalies caused by the AMO represents a significant portion of the rise in global temperatures.
Figure 7

RealClimate limits their discussion to the high latitudes of the Northern Hemisphere. But let’s examine the Land Plus Ocean Temperature anomaly data for the entire Northern Hemisphere (0-90N). It should provide a good comparison with the North Atlantic Sea Surface Temperature anomaly data. The GISTEMP LOTI dataset will be used. That’s the GISTEMP land plus ocean surface temperature data with 1200km radius smoothing. Refer to Figure 8, which compares those two datasets from 1975 to present. Combined GISTEMP land plus sea surface temperature anomalies mimic the sea surface temperature anomaly variations. The combined land plus ocean dataset exaggerates the variations in the North Atlantic Sea Surface Temperature data, but the linear trend of the combined land plus sea surface temperature data is only (approximately) 20% higher than the trend of the North Atlantic SST anomalies.
Figure 8

I’ve used the GISS combined land and sea surface data in this example because the Atlantic Multidecadal Oscillation FAQ webpage of the NOAA Atlantic Oceanographic and Meteorological Laboratory (AOML) Physical Oceanography Division (PhOD) illustrates a correlation between the AMO and large portions of the Pacific Ocean. That correlation map is shown in Figure 9. I’ve provided only the lower half of the illustration from their webpage linked here:

Under the question “How much of the Atlantic are we talking about?” NOAA AOML writes, “Most of the Atlantic between the equator and Greenland changes in unison. Some area of the North Pacific also seem to be affected.”
Figure 9

Wikipedia notes, “In models, AMO-like variability is associated with small changes in the North Atlantic branch of the Thermohaline Circulation, however historical oceanic observations are not sufficient to associate the derived AMO index to present day circulation anomalies.”

That is a curiously written sentence. It is open to numerous interpretations, and that is not helpful, especially for a technical resource. Are they implying that additional forcings from anthropogenic greenhouse gases would be “sufficient to associate the derived AMO index to present day circulation anomalies”? They don’t state that. Or does it mean the models can’t explain all of the variability because causes and effects are not well understood?

Wikipedia provides a description of Thermohaline Circulation. There’s no reason to repeat it for this post, since it fails to provide a detailed description of the impact of Thermohaline Circulation on the AMO.

Back to the Wikipedia statement: As noted above, they write, “In models, AMO-like variability is associated with small changes in the North Atlantic branch of the Thermohaline Circulation.”

In an earlier post, Atlantic Meridional Overturning Circulation Data, I illustrated what appears to be a rough correlation between ENSO and the North Atlantic surface and subsurface flow based on a reconstruction of Atlantic Meridional Overturning Circulation at 26N. Refer to Figure 10 (which is Figure 6 in the post Atlantic Meridional Overturning Circulation Data). Note that the AMOC data was inverted (multiplied by -1) in Figure 9 to show how the AMOC flow appears to slow in response to El Niño events.
Figure 10

So ENSO events appear to be capable of impacting the rate at which North Atlantic Ocean currents transport water northward as part of Thermohaline Circulation, yet I have not found a paper that discusses this.

Foltz and McPhaden (2008) discussed the interaction between Sahel precipitation, Saharan dust, downward shortwave radiation (visible light from the sun), and their impact on Sea Surface Temperatures of the North Atlantic (and the AMO). Link to Foltz and McPhaden (2008) “Trends in Saharan dust and tropical Atlantic climate during 1980–2006”:

Foltz and McPhaden write in their Abstract, “Trends in tropical Atlantic sea surface temperature (SST), Sahel rainfall, and Saharan dust are investigated during 1980–2006. This period is characterized by a significant increase in tropical North Atlantic SST and the transition from a negative to a positive phase of the Atlantic multidecadal oscillation (AMO). It is found that dust concentrations over western Africa and the tropical North Atlantic Ocean decreased significantly between 1980 and 2006 in association with an increase in Sahel rainfall. The decrease in dust in the tropical North Atlantic tended to increase the surface radiative heat flux by 0.7 W/m^2 which, if unbalanced, would lead to an increase in SST of 3 deg C. Coupled models significantly underestimate the amplitude of the AMO in the tropical North Atlantic possibly because they do not account for changes in Saharan dust concentration.”

In other words, studies that fail to account for the multiple interactions between North Atlantic Sea Surface Temperature (AMO), Sahel precipitation, Saharan Dust, and Downward Shortwave Radiation may not be able to properly account for the rise in North Atlantic Sea Surface Temperature.


Figure 11 is a comparison graph of North Atlantic and scaled NINO3.4 SST anomalies from January 1975 to May 2010. Note that the NINO3.4 SST anomaly data were scaled by a factor of 0.15 and they were shifted back in time (lagged) so that the changes in NINO3.4 SST anomalies align with the response by the North Atlantic SST anomalies. It is obvious that North Atlantic SST anomalies rise and fall in response to ENSO events. North Atlantic Sea Surface Temperature anomalies vary due to the changes in atmospheric circulation caused by the ENSO events. This is discussed in more detail in An Introduction To ENSO, AMO, and PDO – Part 1.

Unfortunately, due to the differences in the slopes of the two curves in Figure 11, some divergences are difficult to see.
Figure 11

Figure 12 is a comparison graph similar to Figure 11, but the North Atlantic SST anomalies have been detrended in Figure 12. The similarities and differences between the variations in NINO3.4 and North Atlantic SST anomalies are more obvious. North Atlantic SST anomalies respond to some ENSO events but not others. At times the North Atlantic SST anomalies exaggerate an ENSO signal, at times (especially during La Niña events) it fails to respond fully to the ENSO event, and at times the North Atlantic SST anomalies can be out of synch with NINO3.4 SST anomalies.
Figure 12

In Figures 13, 14, and 15, I’ve added three sets of notes to Figure 12. The differences between the scaled NINO3.4 and North Atlantic SST anomalies from 1982 to about 1986 and from 1991 to approximately 1996 are highlighted in Figure 13. These divergences are explained by the explosive volcanic eruptions of El Chichon and Mount Pinatubo.
Figure 13

In Figure 14, there are a number of periods circled. Prior to 1976, the North Atlantic SST anomalies do not appear to have dropped significantly in response to the 1973/74/75/76 La Niña. But immediately after, from 1979 to 1981, there is a significant rise in North Atlantic SST anomalies that appears to be an exaggerated response to a minor warming in the NINO3.4 region—a warming that is not strong enough to register as an El Niño. Then, after 2002, the year-to-year variations in North Atlantic SST anomalies are out of synch with the NINO3.4 SST anomalies. Why? These divergences are possibly the result of sea level pressure variations, which can have a strong impact on Sea Surface Temperatures.
Figure 14

Figure 15 highlights the two periods when North Atlantic SST anomalies failed to respond fully to La Niña events. If there are epochs when North Atlantic SST anomalies rise in response to El Niño events, but fail to respond fully to La Niña events, North Atlantic SST anomalies will increase. And with the rise in North Atlantic SST anomalies comes the corresponding rise in Northern Hemisphere land surface temperatures and, based on the correlation map shown above in Figure 9, a corresponding rise in North Pacific SST anomalies.
Figure 15

Figure 16 is a graph taken from the post Reproducing Global Temperature Anomalies With Natural Forcings. Basically, in that post I showed how the underlying curve of Global Land and Sea Surface Temperature anomalies can be reproduced using a simple integral (a scaled running total) of NINO3.4 SST anomalies. The assumption is that the oceans integrate the effects of El Niño and La Niña events.

The curiosity: An AMO signal was not needed to reproduce the global temperature anomaly curve. Does this imply that the AMO is an aftereffect of ENSO in the North Atlantic--that the North Atlantic integrates ENSO?
Figure 16

And for those wondering if I had cherry-picked the Hadley Centre data for the post Reproducing Global Temperature Anomalies With Natural Forcings, the post also illustrates the ability to reproduce the GISS and NCDC global temperature anomaly curves. Refer to Figures 15 and 16 in that post.

A final note about the AMO: Many bloggers will write that the AMO has been positive since 1995. They also imply the AMO contributes to the rise in global temperatures only after that year. And they conclude it will be 30 years after 1995, assuming a 60-year AMO cycle, before the AMO “turns negative” again. But the actual basis for the additional contribution of the North Atlantic is the fact that the North Atlantic SST anomalies are rising faster than global SST anomalies, not that the AMO is positive. Refer again to the green curve in Figure 5.

Note: The warm and cold phases of the AMO are, however, used by climate scientists to explain shifts in North American precipitation patterns.

Looking back at all of the graphs of North Atlantic SST anomalies with 13-month and 37-month filters, it is difficult to tell if the North Atlantic SST anomalies and AMO have recently peaked. The North Atlantic SST anomaly data is simply too volatile. However, there is a dataset that represents, in part, the temperature of the top 700 meters of the oceans, and that dataset is much more stable. It is the National Oceanographic Data Center (NODC) Ocean Heat Content (OHC) data. Refer to the NODC Global Ocean Heat Content webpage. Breaking the oceans down into the individual ocean basins, Figure 17, reveals that the drop in the North Atlantic Ocean Heat Content since 2004/05 is the major cause of the recent decline in Global Ocean Heat Content. Is this an indication that the AMO has recently peaked?
Figure 17

The Atlantic Multidecadal Oscillation is a recently discovered mode of Sea Surface Temperature variability for a significant portion of the global oceans. Climate studies provide different causes for the additional strength of the changes in North Atlantic SST anomalies: some blame the Atlantic branch of Thermohaline Circulation, while another discusses the multiple interactions between Saharan dust, Sahel precipitation, solar radiation, and Atlantic Sea Surface Temperature. While cause may be debatable, its impact on Northern Hemisphere sea surface and land surface temperature is clear.

HADISST data is available through the KNMI Climate Explorer:


HR said...


I had a similar thought about the detrending process as you. I'm not sure if the re is a subtle difference.
I can see why detrending is a useful tool for scientists in helping to identify oscillations. But the process itself does put into ones mind the idea that oscillation and trend are unrelated things. Is there any evidence that the oscillation have to have the same magnitude. I'm going to pre-empty your answer by saying probably the data is too short to answer that. But do you have any other thoughts?

Bob Tisdale said...

HR: You asked, "Is there any evidence that the oscillation have to have the same magnitude."

With the chaotic influence of volcanic eruptions and the semi/quasi periodic nature of ENSO, it doesn't seem likely the frequency and magnitude of the AMO would ever be constant.

Another point to consider: detrending the data is useful if the trend in the underlying global data is in one direction. But if the global temperatures were to drop (for example) for 150 years due to some other cycle, then the detrended AMO really wouldn't be useful. Wouldn't we then need to rely on the residual (North Atlantic SST anomalies minus Global SST anomalies) or some other method of presenting the AMO?


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.