The 2014/15 El Niño – Part 5 – The Relationship Between the PDO and ENSO

UPDATE: I’ve added NOAA’s description of the PDO index toward the end of the post, before the closing.

# # #

During the earlier discussions of the upcoming El Niño event, the topic of the Pacific Decadal Oscillation or PDO was raised a number of times. Apparently, it’s time again to discuss what the PDO does and does not represent. I’m trying a couple of new approaches this time. Hopefully, it will be more suitable for those new to the topic. The intents of the post are to dispel many of the misunderstandings about the PDO and to correct the misinformation about the relationship of the PDO with El Niño and La Niña events.

THE PDO IS A USEFUL INDEX

This post is not intended to suggest, in any way, that the Pacific Decadal Oscillation Index is not a useful dataset. It was born as a tool for the Pacific fishing industry. I also understand that the PDO index is used by some meteorologists in forecasting weather. With that in mind…

WHAT THE PDO DATA DO AND DO NOT REPRESENT

The PDO index represents the spatial pattern of the sea surface temperature anomalies in the extratropical North Pacific (20N-65N)…not the sea surface temperature anomalies themselves. A strong positive PDO index value indicates the sea surface temperature anomalies of the eastern extratropical North Pacific are warmer than the western and central portions, which is a spatial pattern created by El Niño events. On the other hand, a strong negative PDO index value indicates the sea surface temperature anomalies of the western and central portions of the extratropical North Pacific are warmer than the eastern portion, and that’s a spatial pattern created by La Niña events.

Figure 1

Figure 1

Studies of salmon production (fishing) in the early 1990s noted abrupt changes in productivity that lasted for multidecadal periods. These changes in salmon production were tied to the climate of the North Pacific Ocean. The term “Pacific Decadal Oscillation” was coined in the mid-1990s as a result of that research.

The method for calculating the PDO index data was first presented in a 1997 paper by Zhang et al. ENSO-like Interdecadal Variability: 1900–93. In that paper the PDO is referred to as NP, for extratropical North Pacific poleward of 20N. There’s a very important quote from Zhang et al. (1997) and we’ll return to it later in the post. El Niños and La Niñas take place in the tropical Pacific, so climate scientists use the extratropical North Pacific for the PDO data to minimize the direct influence of El Niño and La Niña events on the PDO data.

The PDO index data is maintained by the Joint Institute for the Study of the Atmosphere and Ocean (JISAO). See their Pacific Decadal Oscillation (PDO) webpage, and their PDO Index Monthly Values webpage for data. When reviewing the JISAO PDO webpage, consider that the latest paper on that webpage is from the year 2000 and that the webpage is dated that same year. Much has changed in the understanding of the PDO since that time.

WARM AND COOL PHASES OF THE PDO

The use of the words “warm” and “cool” when discussing the phases of the Pacific Decadal Oscillation may lead to some confusion. “Warm” and “cool” are tied to the phase of the eastern portion of the North Pacific, and how the eastern portion relates to the western-central portion, not the actual surface temperature of the entire extratropical North Pacific. That is, when the PDO is in the “warm” phase, the eastern portion of the North Pacific (north of 20N) is warmer than the west-central portion, and vice versa for the “cool” phase.

Consider this when thinking of the PDO index data and its “warm” and “cool” phases. It’s a simple way to determine that the PDO does not relate to the surface temperatures of the North Pacific. A cooling of the sea surface temperature anomalies of the western-central portion of the North Pacific can cause the PDO index to increase, and a warming of the sea surface temperatures of the eastern North Pacific can also cause the PDO index to increase. In the first example, the surface temperatures of the entire North Pacific would have cooled, and in the second, they would have warmed, but the values of the PDO index increased in both examples. Further, as we’ll illustrate later in the post, the PDO index agrees better with the opposing variations western-central portion than it does with the alike-direction changes in the eastern portion.

SPATIAL PATTERNS CREATED BY EL NIÑO AND LA NIÑA EVENTS

Let’s discuss what the PDO represents in more detail: A La Niña event in the tropical Pacific typically creates a spatial pattern in the extratropical North Pacific where it’s cooler in the eastern portion than it is in the western and central portions. See the left-hand cell of Figure 2. The left-hand cell illustrates the average sea surface temperature anomalies of the extratropical North Pacific for the period of July 1999 to June 2000. That’s a full year of the multiyear 1998-01 La Niña. July to June is used because La Niñas normally peak in boreal winter. The dataset is NOAA’s satellite-enhanced Reynolds OI.v2 sea surface temperature data. Note how it’s cooler in the eastern North Pacific than it is in the western and central portions. That’s a textbook spatial pattern in the extratropical North Pacific caused by a La Niña. It’s also a classic “cool” PDO spatial pattern. An El Niño event creates the opposite spatial pattern, where it’s warmer in the eastern extratropical North Pacific and cooler in the western and central portions, and that also relates to a “warm” PDO spatial pattern.

Figure 2

Figure 2

It is often said that the PDO pattern is the dominant spatial pattern in the extratropical North Pacific, and that makes sense because the PDO pattern represents the El Niño- and La Nina-like pattern in the extratropical North Pacific…and…El Niños and La Niñas are the dominant mode of natural variability for the global oceans.

HOW THE PDO DATA ARE CALCULATED

The PDO data are not sea surface temperature data of the North Pacific. The PDO data, on the other hand, are determined from the sea surface temperature data there, using a statistical analysis called Principal Component Analysis. Note the distinction. Once again: the PDO data are determined from the sea surface temperature data of the North Pacific, but they do not represent the sea surface temperatures there. Now, I’m not intentionally trying to make this difficult to understand, but that’s the reality of the PDO index data. The PDO index data is the 1st Principal Component (PC1) of the sea surface temperature anomalies of the extratropical North Pacific (north of 20N)…after the global sea surface temperature anomalies are subtracted from the North Pacific sea surface temperature anomaly data in each 5-deg latitude by 5-deg longitude cell. One more step: The 1st Principal Component is then standardized (divided by its standard deviation) to create the PDO data. And in the case of the PDO, the standardization greatly amplifies the variability of the PDO, making the PDO index appear more significant than it really is. The scaling of the PDO will be confirmed later in the post.

So the PDO data are a very abstract form of the sea surface temperatures of the extratropical North Pacific.

It may be easiest to think of the PDO data in another way—as representing how closely the spatial pattern in the North Pacific at any point in time matches the spatial pattern created by La Niña and El Niño events. If the spatial pattern closely matches the La Niña pattern in Figure 2, then the PDO index value would be negative. The closer the match in the spatial pattern to one created by La Niña events, the greater the negative value. And the opposite holds true for the El Niño-related spatial pattern. The closer the resemblance to the El Niño pattern, the greater the positive PDO index value.

For the right-hand map in Figure 2, I used the Empirical Orthogonal Function analysis (EOF1) feature of the KNMI Climate Explorer, with the Reynolds OI.v2 sea surface temperature data. Empirical Orthogonal Function Analysis is the same as Principal Component Analysis but it also determines the associated spatial patterns. (Note: I took a shortcut with that map. Instead of removing the global data from each grid, I simply had the software in the KNMI Climate Explorer detrend the North Pacific sea surface temperature data. The results are close enough for this discussion.) So the map on the right in Figure 2 presents a classic cool PDO pattern, which would be represented by a negative PDO index value.

A zero PDO index value indicates that La Niña-related or El Niño-related patterns do not exist at that time.

ANOTHER WAY TO VIEW THE RELATIONSHIP BETWEEN THE PDO AND THE SEA SURFACE TEMPERATURES OF THE NORTH PACIFIC

Let’s consider that La Niña-related (cool-PDO) spatial pattern from Figure 2, which shows up in the PDO index as a negative value. The actual sea surface temperatures of the North Pacific can rise and fall, but just as long as that spatial pattern remained fixed (cooler in the east and warmer in the west and central North Pacific) the PDO index would remain at a constant negative value. While it’s not determined this way, sometimes it’s easier to think of the PDO as a temperature difference between the eastern extratropical North Pacific and the western and central portions.

Again, it’s very important to understand that the PDO index does not relate to the actual sea surface temperature anomalies of the North Pacific (north of 20N). The PDO index value only relates to the spatial pattern there.

We’ll confirm this again later in the post, using data.

WHERE THE PDO DATA COMES FROM

Note also that only the extratropical North Pacific is presented in Figures 1 and 2. The PDO data are derived from sea surface temperature data in that location and that location only. The PDO data do not represent the tropical Pacific or the ENSO events that take place in the tropical Pacific. The PDO data also do not represent the Pacific Ocean as a whole.

THE PDO SPATIAL PATTERN IS ALSO IMPACTED BY WINDS

Figure 3 is a time-series graph that presents the PDO index data and the Reynolds OI.v2 satellite-enhanced sea surface temperature anomalies for the NINO3.4 region. The NINO3.4 region sea surface temperature anomalies are a commonly used index for the timing, strength and duration of El Niño and La Niña events. The El Niños are the large upward spikes in the sea surface temperatures of the NINO3.4 region, and the La Niñas are the downward ones. The NINO3.4 data have not been standardized, but the PDO index has been. In the case of the PDO, the standardization multiplies its year-to-year variations. So the standardization greatly exaggerates the importance of the PDO index. The monthly PDO index data are quite variable (see the graph here) so both datasets have been smoothed with a 12-month running-mean filter in Figure 3.

Figure 3

Figure 3

You’ll note that the two datasets run in and out of agreement. That indicates that not every La Niña (El Niño) creates a classic La Nina-like (El Niño-like) spatial pattern in the extratropical North Pacific.

The reason that the PDO index does not mimic an ENSO index at all times is the spatial patterns of the sea surface temperatures in the North Pacific are also affected by the sea level pressures (and the related variations in wind patterns) there. (We’ll confirm this in the next section.) That is, the sea level pressures and the strength and direction of the winds (basically weather) in the extratropical North Pacific can suppress or enhance the El Niño-like or La Niña-like spatial patterns there. That causes the year-to-year and the multidecadal variations in the PDO index to be different than an ENSO index. The differences in the annual variability can be seen in Figure 3, and the differences in the multidecadal variations between the PDO index and our ENSO index can be seen in Figure 4.

Figure 4

Figure 4

Figure 4 compares a long-term ENSO index (NINO3.4 sea surface temperature anomaly data based on the HADISST dataset) to the long-term monthly PDO index. Both datasets have been smoothed with 61-month filters to suppress the year-to-year variations and to highlight the differences in their multidecadal variations. (See the monthly long-term PDO and NINO3.4 data here.) It’s very obvious that ENSO and PDO data both show long-term variations. The smoothed NINO3.4 data indicate that there are decadal and multidecadal variations in the strengths, frequency and durations of El Niño and La Niña events. And the smoothed PDO index shows that there are decadal and multidecadal variations in the El Niño- and La Niña-like spatial patterns in the extratropical North Pacific. Because the sea level pressures and wind patterns in the extratropical Pacific also have decadal and multidecadal variations, the PDO index does not track the ENSO index over these timeframes.

The Japan Meteorological Agency (JMA) webpage Explanation of the Pacific Decadal Oscillation (PDO) also confirms this relationship between the PDO and the North Pacific sea surface temperatures…and the sea level pressures there. The JMA writes (note: the North Pacific Index is based on sea level pressure):

In the North Pacific, the atmosphere and the ocean display a trend of co-variance with aperiod [sic] of about 20 years. This variability is called the Pacific Decadal Oscillation (PDO).

Supposing that SSTs are lower than their normals in the central part of the North Pacific, they are likely to be higher than their normals both in the eastern part of the North Pacific and in the equatorial Pacific. This seesaw pattern varies slowly, and appears repeatedly with a period of about 20 years. In this case, the Aleutian Low and the jet stream in the upper troposphere tend to be strong. The North Pacific Index (NPI) is utilized as an indicator for the strength of the Aleutian Low.

We’ll discuss the North Pacific Index in the next section.

This long-term variability of the PDO and its relationship with El Niño and La Niña events in the equatorial Pacific is easier to see with multiyear filters…as shown above in Figure 4. On monthly and annual timescales, the variations in the PDO data can, at times, be remarkably similar to those of the ENSO index, as shown in Figure 3 above, and at times they can be quite different.

THE NORTH PACIFIC INDEX (NPI) AND ITS RELATIONSHIP WITH THE PDO

The North Pacific Index was introduced in the 1994 Trenberth and Hurrel paper Decadal Ocean-Atmosphere Variations in the Pacific. The UCAR North Pacific (NP) Index webpages include the following summary, which confirm our understandings so far of the PDO:

The North Pacific (NP) Index is the area-weighted sea level pressure over the region 30°N-65°N, 160°E-140°W. The NP index is defined to measure interannual to decadal variations in the atmospheric circulation. The dominant atmosphere-ocean relation in the north Pacific is one where atmospheric changes lead changes in sea surface temperatures by one to two months. However, strong ties exist with events in the tropical Pacific, with changes in tropical Pacific SSTs leading SSTs in the north Pacific by three months.

The monthly North Pacific Index data from UCAR are here. They are also available in anomaly form through the Monthly climate indices webpage at the KNMI Climate Explorer (through August 2012). Looking at the monthly North Pacific Index anomalies, the PDO index and the NINO3.4 sea surface temperature anomalies during the satellite era of sea surface temperatures, the relationships between the North Pacific Index data and the other two indices is hard to imagine. See Figure 5. Smoothing all three datasets with 12-month filters doesn’t seem to help much either, as shown in Figure 6.

Figure 5

Figure 5

# # # #

Figure 6

Figure 6

So 4 years ago, I presented a very simply way to show the relationship between the North Pacific Index data and the other two variables. (That earlier post is here.) I subtracted the PDO index data from standardized NINO3.4 sea surface temperature anomalies, and compared that difference to the North Pacific Index data, where the difference and the North Pacific Index data were both smoothed with multiyear filters. The long-term variations mimicked one another reasonably well over much of the term of the data. In Figure 7, for a change of pace, I used Kaplan SST-based NINO3.4 data and smoothed the data with 121-month filters.

Figure 7

Figure 7

The multidecadal variations in the North Pacific Index (sea level pressure) mimic the difference between the ENSO and PDO indices from the early 1920s to present. Considering how sparse the data are before the early 1920s, the divergence then is not unexpected.

Figure 8 presents two correlation maps for the satellite era of sea surface temperatures. The top map shows the correlation between the North Pacific Index and the sea surface temperature anomalies of the extratropical North Pacific. The North Pacific Index creates a weak PDO-like pattern. And in the bottom map is the correlation between NINO3.4 sea surface temperature anomalies (our ENSO index) and the sea surface temperature data for the same region. ENSO creates a strong PDO-like pattern.

Figure 8

Figure 8

I believe it’s safe to say that the difference between the long-term variations in the Pacific Decadal Oscillation and ENSO is related to the sea level pressure of the North Pacific as expressed by the North Pacific Index (and interrelated wind patterns), and that ENSO dominates over the sea level pressure on short-term timescales.

“THE PDO IS DEPENDENT UPON ENSO ON ALL TIMESCALES”

The heading is a quote from the concluding comments of the Newman et al. (2003) ENSO-Forced Variability of the Pacific Decadal Oscillation. In other words, ENSO drives the Pacific Decadal Oscillation, not the other way around.

Figure 9 is cell d of Figure 1 from Newman et al. (2003).

Figure 9

Figure 9

Their discussion of it in the text reads:

ENSO also leads the PDO index by a few months throughout the year (Fig. 1d), most notably in winter and summer. Simultaneous correlation is lowest in November– March, consistent with Mantua et al. (1997). The lag of maximum correlation ranges from two months in summer (r ~ 0.7) to as much as five months by late winter (r ~ 0.6). During winter and spring, ENSO leads the PDO for well over a year, consistent with reemergence of prior ENSO-forced PDO anomalies. Summer PDO appears to lead ENSO the following winter, but this could be an artifact of the strong persistence of ENSO from summer to winter (r = 0.8), combined with ENSO forcing of the PDO in both summer and winter. Note also that for intervals less than 1yr the lag autocorrelation of the PDO is low when the lag autocorrelation of ENSO (not shown) is also low, through the so-called spring persistence barrier (Torrence and Webster 1998).

This also agrees with the findings of the paper that first presented the PDO index: Zhang et al. (1997) ENSO-like Interdecadal Variability: 1900–93. Recall that Zhang et al. refer to the PDO as “NP”. For an ENSO index, they use the Cold Tongue Index (CT) in place of NINO3.4 SST anomalies. The Cold Tongue Index represents sea surface temperature anomalies of the region bordered by the coordinates of 6S-6N, 180-90W, where NINO3.4 region occupies 5S-5N, 170W-120W. In the bottom cell of Figure 7 of Zhang et al., shown here as Figure 10, they illustrate the cross-correlation functions between their ENSO index and the PDO. As illustrated in the original paper that calculated the PDO index, NP (PDO) lags CT (ENSO) by approximately 3 months.

Figure 10

Figure 10

Thus, my repeated statement in blog comments whenever the relationship between the PDO and ENSO is discussed: the PDO is an aftereffect of ENSO…that’s also impacted by the sea level pressure and related wind patterns of the North Pacific.

IF THE PDO LAGS ENSO, THEN THE PDO CANNOT DRIVE ENSO

There are posts and comments around the blogosphere that state something to the effect of “during a warm PDO (a multidecadal period when the PDO is positive), El Niño events are more frequent, and during a cool PDO (a multidecadal period when the PDO is negative), La Niña events are more frequent.” Expressed in that way, those comments lead people to believe that the PDO drives the frequencies, strengths and durations of El Niño and La Niña events. Because the PDO lags ENSO, the warm or cool phase of the PDO cannot drive the strength, frequency and duration of El Niño and La Niña events. It’s the other way around, “the PDO is dependent upon ENSO on all timescales”. The persons making the claims that the PDO is the driver of ENSO have cause and effect reversed, and they forget to account for the addition variability of the PDO caused by variations in sea level pressure and wind patterns.

HOWEVER

This does not mean that the sea surface temperatures of the North Pacific do not provide feedback to the tropical Pacific. Unfortunately, the PDO index does not represent the sea surface temperatures of the North Pacific. And the PDO index correlates poorly with the sea surface temperature anomalies of the eastern extratropical North Pacific.

THE PDO INDEX IS NOT AN ENSO INDEX

We cannot look at the PDO index and always determine if an El Niño or La Niña is occurring in the tropical Pacific. Sometimes, yes…but not always. The same holds true in the other direction: we cannot look at an ENSO index and assume the North Pacific will always have an El Niño-like spatial pattern during an El Niño, or a La Niña-like spatial pattern during a La Niña. Here are a few examples.

Figure 11 presents the monthly satellite-era comparison of the PDO index and NINO3.4 sea surface temperature anomalies (our ENSO index). In November 1994, the PDO index was showing a strong negative value (suggesting a La Niña), while the sea surface temperatures of the NINO3.4 region are indicating a moderate El Niño was taking place.

Figure 11

Figure 11

The sea surface temperature anomaly map of the tropical Pacific and the extratropical North Pacific in Figure 12 confirm both realities. The tropical Pacific is showing an El Niño event and the extratropical North Pacific is showing a “cool” PDO pattern, where it’s warmer in the western-central North Pacific than it is in the eastern North Pacific.

Figure 12

Figure 12

Let’s reverse the situation. A good example is a little more difficult to find, but we’ll settle on May 1996. See Figure 13, where I’ve highlighted that month in the time-series comparison. The PDO index peaks at a relatively high value in May 1996, while the ENSO index is showing ENSO neutral conditions, but leaning toward La Niña.

Figure 13

Figure 13

And we can confirm both realities with the sea surface temperature anomaly map for May 1996. See Figure 14.

Figure 14

Figure 14

CONFIRMATION – THE PDO INDEX DOES NOT REPRESENT THE SEA SURFACE TEMPERATURE ANOMALIES OF THE EXTRATROPICAL NORTH PACIFIC

Let’s confirm/reinforce early statements: that the PDO index data do not represent the sea surface temperature anomalies of the North Pacific. Figure 15 includes the sea surface temperature data for that region (20N-65N, 100E-100W). Also included in the graph is the PDO index data from JISAO, which have been scaled (multiplied by a factor of 0.2) to bring their variations down into line with the sea surface temperature data. You’ll also note that I’ve had Excel add trend lines to the graph. The trends show that sea surface temperature anomalies of the extratropical North Pacific have warmed over the past 32+ years, while the PDO data show a negative trend.

Figure 15

Figure 15

Also note the monthly variations in the sea surface temperature data and the PDO data. They also appear to oppose one another…but not always. The PDO data and the sea surface temperature anomalies do have an opposing relationship but the correlation is poor. The correlation coefficient is -0.52, which is pretty bad. A correlation coefficient of -1.0 indicates a perfect negative correlation.

Figure 16 presents a similar graph, but using the long-term PDO data from JISAO (January 1900 to February 2014) and the sea surface temperature anomalies of the extratropical North Pacific using the HADISST dataset. The PDO data also do not represent the sea surface temperatures of the North Pacific over the long term.

Figure 16

Figure 16

In Figure 17, I’ve detrended the sea surface temperature data and compared them to the scaled PDO data. The multidecadal variations of the PDO data also do not coincide with the multidecadal variations in the sea surface temperature anomalies of the North Pacific.

Figure 17

Figure 17

For more information about the multidecadal variations in sea surface temperatures, see the post Multidecadal Variations and Sea Surface Temperature Reconstructions.

CAN THE PDO DATA BE USED TO DETERMINE THE CONTRIBUTION OF THE NORTH PACIFIC SEA SURFACE TEMPERATURES TO GLOBAL WARMING?

It’s often noted that when the PDO is positive for multidecadal periods, global surface temperatures warm, and when the PDO is negative, global surface temperatures stop warming or cool a little. It’s is then assumed that the PDO has something to do with the warming or cooling of global surface temperatures. The problem: there is no mechanism through which the PDO can raise or lower global surface temperatures, because the PDO does not represent the surface temperatures of the extratropical North Pacific (where the PDO is derived).

(Note: Referring back to Figure 4, multidecadal variations in the strengths, frequencies and durations of El Niño and La Niña events are capable of raising and lowering global surface temperatures. Those processes exist. During multidecadal periods when El Niño events dominate, the tropical Pacific is releasing more heat than normal from the tropical Pacific to the atmosphere and redistributing more warm water than normal from the tropical Pacific to the adjoining ocean basins. The opposite holds true during multidecadal periods when La Niña events dominate. On the other hand, multidecadal variations in the spatial patterns in the North Pacific sea surface temperature anomalies, aka the PDO, are not capable of raising or lowering global sea surface temperatures.)

Figure 18 compares the sea surface temperature anomalies of the extratropical North Pacific with global sea surface temperature anomalies. The UKMO HADISST data are being presented. And the data have been smoothed with 61-month filters to help show the differences in the long-term variations. There are periods when the sea surface temperatures of the extratropical North Pacific run in parallel with the global surface temperature data. At those times, the sea surface temperatures are not adding to or suppressing the variations in the global data. There are periods when the sea surface temperatures of the extratropical North Pacific warm faster than the global data. At those times, the sea surface temperatures of the North Pacific (north of 20N) are adding to the warming of global sea surface temperatures. And, conversely, the sea surface temperatures of the extratropical North Pacific are suppressing the warming of global data when they are warming at a slower rate than the global data…and adding to the cooling when the surface of the North Pacific is cooling faster than they are globally.

Figure 18

Figure 18

Let’s subtract the global sea surface temperature data from the data for the extratropical North Pacific. We’ll call that difference the “Extratropical North Pacific Residual”. See Figure 19. The “Extratropical North Pacific Residual” illustrates the additional variations of the sea surface temperatures of the North Pacific (north of 20N) above and beyond the global sea surface temperatures. Also included in Figure 19 is the scaled PDO index. Both datasets have once again been smoothed with 61-month filters. It’s very obvious that the PDO data do not represent the additional variability of the North Pacific sea surface temperatures. In fact, over multidecadal timeframes, the PDO data can be inversely related to the extra variability of the sea surface temperatures of the extratropical North Pacific.

Figure 19

Figure 19

There’s another way to illustrate the inverse relationship between the PDO and surface temperatures, and that’s with maps that show the correlations of Northern Hemisphere surface temperatures with the PDO and with the sea surface temperatures of the extratropical North Pacific. See Figure 20. The top map shows that surface temperatures of most of the Northern Hemisphere are negatively correlated with the PDO index. That negative correlation means that when the PDO data increases, the surface temperatures for most of the Northern Hemisphere show cooling.

Figure 20

Figure 20

On the other hand, the surface temperatures for most of the Northern Hemisphere are positively correlated with the sea surface temperatures of the extratropical North Pacific. See the bottom map of Figure 20. That is, the surface temperatures for much of the Northern Hemisphere warm when the surface of the extratropical North Pacific warms, and it cools when the surface of the extratropical North Pacific cools.

So the answer to the question posed in the heading for this section is no. The PDO data cannot be used to determine the contribution of the North Pacific sea surface temperatures to global warming. The PDO data do not represent the sea surface temperatures of the extratropical North Pacific.

THE EASTERN OR WEST-CENTRAL PORTIONS OF THE NORTH PACIFIC: WHICH AGREES BEST WITH THE PDO INDEX?

Earlier we discussed how a “warm” or “cool” PDO index value appears to be tied to the sea surface temperature anomalies of the eastern extratropical North Pacific. That is, sea surface temperatures in the eastern extratropical North Pacific tend to be elevated when the PDO is “warm”, and depressed when the PDO is “cool”. And we discussed how the west-central portion of the extratropical North Pacific is “cool” when the PDO is “warm” and, conversely, the west-central portion is “warm” when the PDO is “cool”.

But, in terms of sea surface temperature anomalies, which of those two regions agrees better with the PDO index data: the eastern or the west-central? The use of the terms “warm” and “cool” draws our attention to the eastern extratropical North Pacific, but does it actually correlate well with the PDO?

To determine this, we’ll present the satellite era sea surface temperature data for those two portions of the extratropical North Pacific, and compare them to the PDO index. See the map in Figure 21. For this part of the discussion, we’ll call the west-central portion by the name that’s commonly used for that region: the Kuroshio-Oyashio Extension (KOE). (I’ve tipped my hand with the notes on the illustration.)

Figure 21

Figure 21

Let’s start with the eastern region. Figure 22 presents the sea surface temperature anomalies of the eastern extratropical North Pacific (20N-65N, 150W-100W). Also shown is the PDO index that has been arbitrarily scaled (scaling factor = 0.5). After the 1997/98 El Niño, the variations in the PDO and in the sea surface temperatures of the eastern region of the North Pacific (north of 20N) tend to agree with one another, but before that El Niño, there is little agreement. Overall, the two datasets correlate quite poorly since November 1981, with a correlation coefficient of 0.52.

Figure 22

Figure 22

The scaled PDO index is compared to the sea surface temperature anomalies of the Kuroshio-Oyashio Extension (west-central portion of extratropical North Pacific) in Figure 23. If not for that three-year stretch from 2004 through 2006, the variations in the PDO data would mirror the variations in the sea surface temperatures of the Kuroshio-Oyashio Extension. The correlation coefficient of those two datasets is -0.91, which is much better than the eastern region (though inverted).

Figure 23

Figure 23

That suggests that studies of the variability of the PDO during the satellite era and the relationship of the PDO with ENSO are, for the most part, studies of the variations in the sea surface temperatures of the Kuroshio-Oyashio Extension (not the eastern portion of the North Pacific), with the results inverted for the PDO.

THREE PDO INDICES ARE AVAILABLE THROUGH THE KNMI CLIMATE EXPLORER

The PDO index maintained by JISAO is the traditional PDO index. It’s used most often in climate-related studies. But the JISAO PDO index is derived from three sea surface temperature datasets, two of which are obsolete. The PDO data from JISAO are calculated from an obsolete version of the UKMO sea surface temperature data from January 1900 to December 1981 and from an obsolete version of the Reynolds OI sea surface temperature data from January 1982 through December 2001. JISAO uses the up-to-date Reynolds OI.v2 sea surface temperature data since January 2002. So the JISAO PDO data are a mix of very different sea surface temperature datasets.

The Monthly climate indices webpage at the KNMI Climate Explorer includes 3 PDO datasets: the JISAO data, and PDO data based on two other seas surface temperature datasets: NOAA’s ERSST.v3b data and UKMO’s HADSST3 data.

Looking at the three PDO datasets in monthly form, Figure 24, the three datasets appear to agree reasonably well. In fact, the ERSST.v3b and HADSST3 data correlate well with one another (correlation coefficient of 0.94). But the correlations are lower between the JISAO PDO data and the other two datasets (JISAO PDO correlation coefficients: 0.85 with ERSST.v3b data and 0.86 with HADSST3 data). With that in mind, if you were studying the PDO in close relation to a specific surface temperature dataset, it might be best to use the PDO data associated with the surface temperature data being studied.

Figure 24

Figure 24

Figure 25 presents the three PDO datasets smoothed with 121-month running-mean filters. They agree better in some periods than in others. The temporary upswing in the JISAO PDO data from the mid-1950s to about 1970 is not as great as the other two datasets. And the JISAO PDO does not dip as low as the other two in the early part of the 20th century.

Figure 25

Figure 25

A couple of things to note about Figures 24 and 25: On the data pages at the KNMI Climate Explorer for the ERSST.v3b- and HADSST3-based PDO data, KNMI lists the scaling factors applied to those datasets. (Thanks, Geert Jan.) For the ERSST.v3b-based PDO data, the scaling factor was -3.5, and the HADSST3-based PDO data was scaled by a factor of 3.0. I suspect the scaling factors are based on the standard deviations of both datasets…and that they have been rounded.

Note also that the ERSST.v3b data were scaled by a factor of -3.5. This indicates the ERSST.v3b-based PDO data were originally inverted…or the opposite sign of how the PDO data is normally presented. KNMI then multiplied the ERSST.v3b PDO data by a negative number in order to present them as we’re used to seeing them.

With the scaling factors, assuming the values are rounded, we can get a rough estimate of the actual variations of the PDO data in terms of sea surface temperatures in deg C.

THE PDO DATA IN DEG C

When I first began to study ocean processes, I believed (wrongly) that the Pacific Decadal Oscillation were El Niño-like and La Niña-like events taking place in the North Pacific. One of the influences on my initial misunderstanding was how the PDO data were presented. Overlooking the fact that I didn’t know how greatly the standardization exaggerated the PDO index, the graphs of the raw PDO data showed monthly variations in excess of -2.0 to +3.0, which were comparable to, if not larger than, the variations in the sea surface temperature of the equatorial Pacific. See Figure 26. The standardization of the JISAO PDO data may also have greatly exaggerated the importance of the PDO for other people. Hopefully, this presentation will put an end to that.

Figure 26

Figure 26

I later got a rough idea of the impact of the standardization of the PDO data (the magnitude of the scaling). (See the discussion of Figures 10 and 11 in the post here.) I used the North Pacific residual data (shown above smoothed in Figure 18), not the formula used by JISAO to determine the PDO. Based on the scaling factors determined by KNMI for the HADSST3- and ERSST.v3b-based PDO indices, my scaling factor was 1.5 to almost 2 times too high. My apologies. But I was on the right track.

So let’s look at the magnitudes of the variations of the HADSST3 and ERSST.v3b PDO indices in terms of deg C. To accomplish this, we’ll simply divide those PDO indices by the scaling factors listed by KNMI, leaving the ERSST.v3b PDO data inverted…and keeping in mind that the listed scaling factors were likely rounded. So we’re looking at rough approximations of the two PDO datasets. In Figure 27, the results are compared to NINO3.4 sea surface temperature anomalies, our ENSO index, as a reference. The sea surface temperature-based ENSO signal dwarfs the variations of the two PDO datasets.

Figure 27

Figure 27

That’s not to say that the sea surface temperatures of the North Pacific are not important. As we discussed above and illustrated in Figure 18, the multidecadal variations in the sea surface temperature anomalies can enhance or suppress the warming of global sea surface temperatures. And again, the PDO index data do not represent the sea surface temperatures of the North Pacific.

ENSO DRIVES THE TRADE WINDS OF THE EQUATORIAL PACIFIC, NOT THE PDO

I had a recent blog discussion about the PDO and trade winds. In it, I showed how the variability of the trade winds of the equatorial Pacific was more closely related to an ENSO index than they were to the PDO index. That really should have been expected, but I suspect the person I was having the discussion with was basing his/her understanding of the PDO on beliefs, not on data.

The role of the trade winds in El Niño and La Niña events and the interrelationship of the trade winds with the sea surface temperatures of the equatorial Pacific were discussed in a number of recent ENSO posts:

Here’s the body of my comment from the recent blog discussion. I’ve added Figure numbers to the text in brackets:

Because you somehow believe that the PDO influences the trade winds of the equatorial Pacific, I have an exercise for you. To simplify things, we’ll look at the satellite era of sea surface temperatures (Nov 1981 to Mar 2014). Take the weighted average of the NOAA trade wind index anomalies, basing the weighting on the longitudes included. The West Pacific Trade Wind Index anomalies are here, the Central Pacific Trade Wind Index anomalies here and West Pacific Trade Wind Index anomalies here. (The second of the three datasets on each page) After you’ve determined the weighted average of the trade wind indices, standardize it. Then, to your spreadsheet, add NINO3.4 sea surface temperature anomalies (Reynolds OI.v2) from the NOAA NOMADS website. We’ll use that as our ENSO index. Standardize it as well. Now add the JISAO PDO data. It’s already standardized so you can save yourself a step.

Now determine how well the weighted average of the trade wind data correlates with the PDO and with the ENSO index. The correlation coefficient of the trade wind data with the PDO is about -0.36. Not very good. The correlation coefficient of the trade wind data with the ENSO index is about -0.82. Much better.

Now plot them. The trade wind data are noisy, so if you like, smooth all three datasets with 12-month running-average filters. Your graph should look something like this [Figure 28]:

Figure 28

[Figure 28]

We can see that the trade wind indices are inversely related to the ENSO index and the PDO. So invert the standardized trade wind data. Your result should look like this [Figure 29]:

Figure 29

[Figure 29]

Note how poorly the PDO data agree with the trade wind data before the 1997/98 El Niño, while the ENSO index and the trade wind data mimic one another over the term of the data. That pretty well sums up the differences between the PDO and ENSO.

If you’d like, you’re more than welcome to continue to believe that the PDO influences the trade winds of the equatorial Pacific, but your beliefs will not be supported by data.

THE CURRENT VALUE OF THE PDO INDEX

The JISAO PDO Index values have been positive this year, and for March 2014, it is at a moderately positive value of 0.97. Curiously, around the blogosphere, we’re being told the PDO is “cool”.

DO THE CLIMATE MODELS USED BY THE IPCC PROPERLY SIMULATE THE MULTIDECADAL VARIATIONS IN THE SEA SURFACE TEMPERATURES OF THE EXTRATROPICAL NORTH PACIFIC?

I’m glad you asked. The answer is no. We’ve already established that the climate models do not properly simulate the variations in sea surface temperatures on any timescale. See the posts:

This post included discussions of the PDO and the multidecadal variations in the sea surface temperatures of the extratropical North Pacific. So let’s compare models with data, and we’ll start with a proxy for the PDO, and I’ll borrow a few graphs from an earlier post.

We established in this post that the sea surface temperature anomalies of the Kuroshio-Oyashio Extension correlated well (but inversely) with the PDO index. (See Figure 23 above.) As you’ll discover, that relationship extends back in time as well. The following is a portion of the post Questions the Media Should Be Asking the IPCC – The Hiatus in Warming, which was also cross posted at Jo Nova’s blog. (I’ve updated the Figure numbers for this post.):

2. North Pacific Multidecadal Variability

The North Pacific also exhibits multidecadal variations in sea surface temperatures. Unfortunately, the climate science community presents those variations in a very abstract (statistically created) form called the Pacific Decadal Oscillation. To further complicate matters, the variations in the Pacific Decadal Oscillation Index are actually inversely related to the sea surface temperatures of the region of the North Pacific from which the Pacific Decadal Oscillation index is derived. That, of course, is confusing to most persons.

In the North Pacific, the variations are dominated by the sea surface temperatures of an area east of Japan known as the Kuroshio-Oyashio Extension (or KOE). The inverse relationship between the Pacific Decadal Oscillation index and the sea surface temperatures of the Kuroshio-Oyashio Extension are shown in Figure 30. They’re basically mirror images of one another.

Figure 30

Figure 30

(Note: In Figure 30, the Kuroshio-Oyashio Extension data have been detrended, and they have been standardized (divided by their standard deviation) because the Pacific Decadal Oscillation Index is standardized.)

Because we’re interested in sea surface temperatures and not an abstract form of them, we’ll use the sea surface temperatures of the Kuroshio-Oyashio Extension for much of this discussion.

Let’s see whether the models can simulate the multidecadal variations in the sea surface temperatures of the Kuroshio-Oyashio Extension. For this example, we’ll subtract the global sea surface temperatures from the sea surface temperatures of the Kuroshio-Oyashio Extension. (It’s similar to the method used by Trenberth and Shea for the Atlantic Multidecadal Oscillation.) The coordinates used for the Kuroshio-Oyashio Extension data are 30N-45N, 150E-150W and the global sea surface temperature data are for all of the global oceans, 90S-90N. (The multidecadal variations in the Kuroshio-Oyashio data in Figure 31 are different those in Figure 30 because they were determined using different methods. Also note they have not been standardized in Figure 31.) As shown with the blue curve in Figure 31, there are very large multidecadal variations in the sea surface temperatures of the Kuroshio-Oyashio Extension after the global data has been subtracted.

Figure 31

Figure 31

On the other hand, using the models prepared for the IPCC’s 5th Assessment Report, we get totally different results when we subtract the modeled global sea surface temperatures from the modeled sea surface temperatures of the Kuroshio-Oyashio Extension. (See the red curve in Figure 31.) This indicates the models do not simulate the multidecadal variations in the sea surface temperatures of the Kuroshio-Oyashio Extension. Because the variations in the surface temperatures of the Kuroshio-Oyashio Extension region dominate the North Pacific, the failure of the models to show the same multidecadal variations suggests that greenhouse gases (and aerosols) used in the models are not responsible for the warming (and cooling) in North Pacific sea surface temperatures.

And let’s confirm that the models also do a horrendous job of simulating the multidecadal variations in sea surface temperatures for the entire extratropical North Pacific. In this case, we’ll simply detrend the sea surface temperature anomaly data and the climate model simulations for the extratropical North Pacific. See Figure 32.

Figure 32

Figure 32

That recent divergence between the models and the data stands out like a sore thumb.

We’ve already discussed why we use the multi-model ensemble-member mean in our model-data comparisons. (See the post here.) The model mean represents the forced component of the models. That is, the model mean presents how surface temperatures should vary if (big if) their variations were in response to the forcings used as inputs to the climate models. Obviously, the multidecadal variations in sea surface temperatures of the extratropical North Pacific are not in response to the forcings used to drive surface temperatures in models.

DID LAST YEAR’S UNUSUAL WARMING EVENT IN THE EXTRATROPICAL NORTH PACIFIC REGISTER ON THE PDO INDEX?

Last year, there was an unusual warming of the surface temperatures of the extratropical North Pacific. See the post here. Sea surface temperatures there have lowered somewhat, as shown in Figure 33, but they do not appear to have returned yet to previous levels. I’ve also included the PDO index (scaled) in Figure 33 to confirm that the warming event did not register on the PDO index as it occurred. Again, the PDO index does not represent sea surface temperatures of the North Pacific.

Figure 33

Figure 33

And while the sea surface temperatures of the Eastern Extratropical North Pacific have cooled over the past 32 years (see the graph here), the March 2014 sea surface temperatures continue to be unusually warm. See Figure 34.

Figure 34

Figure 34

It will be interesting to see what happens in upcoming months as the current El Niño conditions continue to develop.

UPDATE: CONFIRMATION – NOAA’s DESCRIPTION OF THE PDO INDEX

To confirm that the PDO index is a product of the spatial pattern of the sea surface temperature anomalies of the extratropical North Pacific and sea level pressure, refer to NOAA’s Pacific Decadal Oscillation (PDO) teleconnections webpage. There NOAA writes (my boldface):

The Pacific Decadal Oscillation (PDO) is often described as a long-lived El Niño-like pattern of Pacific climate variability (Zhang et al. 1997). As seen with the better-known El Niño/Southern Oscillation (ENSO), extremes in the PDO pattern are marked by widespread variations in the Pacific Basin and the North American climate. In parallel with the ENSO phenomenon, the extreme phases of the PDO have been classified as being either warm or cool, as defined by ocean temperature anomalies in the northeast and tropical Pacific Ocean. When SSTs are anomalously cool in the interior North Pacific and warm along the Pacific Coast, and when sea level pressures are below average over the North Pacific, the PDO has a positive value. When the climate anomaly patterns are reversed, with warm SST anomalies in the interior and cool SST anomalies along the North American coast, or above average sea level pressures over the North Pacific, the PDO has a negative value (Courtesy of Mantua, 1999).

CLOSING

Hopefully, this post provided you with a better understanding of the Pacific Decadal Oscillation…what it is, and more importantly, what it isn’t. For some reason, many people seem to have emotional ties to the Pacific Decadal Oscillation, so I suspect this post will not be well received by some. But, as always, data trumps assumptions and speculation, and I present data, not assumptions or speculation.

EARLIER POSTS IN THIS SERIES

FURTHER READING

We didn’t get a chance to discuss ENSO processes in this post. But my ebook Who Turned on the Heat? goes into a tremendous amount of detail to explain El Niño and La Niña processes and the long-term aftereffects of strong El Niño events. Who Turned on the Heat? weighs in at a whopping 550+ pages, about 110,000+ words. It contains somewhere in the neighborhood of 380 color illustrations. In pdf form, it’s about 23MB. It includes links to more than a dozen animations, which allow the reader to view ENSO processes and the interactions between variables.

I’ve lowered the price of Who Turned on the Heat? from U.S.$8.00 to U.S.$5.00. A free preview in pdf format is here. The preview includes the Table of Contents, the Introduction, the first half of section 1 (which was provided complete in the post here), a discussion of the cover, and the Closing. Take a run through the Table of Contents. It is a very-detailed and well-illustrated book—using data from the real world, not models of a virtual world. Who Turned on the Heat? is only available in pdf format…and will only be available in that format. Click here to purchase a copy. Thanks.

 

About these ads

78 thoughts on “The 2014/15 El Niño – Part 5 – The Relationship Between the PDO and ENSO

  1. Waiting for El Ninot, Episode 2:

    Estragon:
    Its nice to think that all our problems – such as the Mann-o-pause in global warming, will be behind us once el Ninot comes to the rescue. Any day now I guess…

    Vladimir:
    Actually – not wanting to damp your hopes – but what just happened to the Kelvin wave subsurface warm water in the Pacific? A week or so ago it was nice and strong but suddenly its gone AWOL.

    Estragon:
    Oh – you’re right – and that East Pacific south tropical cold water seems oddly persistent, especially along the South American west coast. The ENSO page Hovmoller diagram (anomaly plotted against time and Pacific latitude band) also shows the warm pulse bottoming out. That’s what comes of raising one’s hopes I guess.

    Vladimir:
    You jinxed it you old fool! Keep your mouth shut next time!

    Estragon:
    Nothing to be done.

  2. What do you make of the fact that a consistent trend of warming has been measured in the oceans?

  3. Paul Woland says: “What do you make of the fact that a consistent trend of warming has been measured in the oceans?”

    It’s not consistent in any way, nor is it uniform. There is also no evidence that the warming is from manmade greenhouse gases. In other words, the data indicate the oceans warmed naturally. You simply have to divide the data into logical subsets in order to see it. For an introduction to this, see my illustrated essay “The Manmade Global Warming Challenge” (42mb):

    http://bobtisdale.files.wordpress.com/2013/01/the-manmade-global-warming-challenge.pdf

    Regards

  4. ‘We cannot look at the PDO index and always determine if an El Niño or La Niña is occurring in the tropical Pacific. Sometimes, yes…but not always. The same holds true in the other direction: we cannot look at an ENSO index and assume the North Pacific will always have an El Niño-like spatial pattern during an El Niño, or a La Niña-like spatial pattern during a La Niña.’

    The thing I like about reading Bob Tisdale is that he is: exhaustive in his enquiries, astute in his observations and honest in his conclusions.

  5. Wow, yet another great and illuminating post by Bob Tisdale. It is not easy reading and the plethora of graphs induces a feeling of vertigo, but it is really worth reading all of it. Bob is totally thorough, expansive, astute and honest, which is more than can be said for the pro IPCC Gang.

  6. Bob , your account of what the PDO is , is not correct.

    “The PDO index represents the spatial pattern of the sea surface temperature anomalies in the extratropical North Pacific (20N-65N)”

    The PDO is a single figure for each date. It in no way carries “spacial pattern” information about the N. Pacific.

    “A strong positive PDO index value indicates the sea surface temperature anomalies of the eastern extratropical North Pacific are warmer than the western and central portions”

    NO. This is not what PDO measures. PDO is the deviation of N. Pacific temps from the GLOBAL AVERAGE, not from the north-western Pacific. The pattern you speak of is generally true but it is not what PDO measures nor how it is defined or calculated.

    If you are going to write stuff explaining what these indices are, at least present an accurate definition to start with.

  7. Bob Tisdale, so you do accept that the oceans are warming, as is demonstrated in the paper published in Nature? Could you explain what you think is the “natural” process behind this, if not the change in the atmospheric composition of greenhouse gases?

  8. Greg says: “Bob , your account of what the PDO is , is not correct.”

    Actually, Greg, if you had read and understood what was presented in this post you would discover that my description of the PDO is correct. I suggest you read Zhang et al (1997) so that you can confirm how the PDO is calculated:

    http://www.atmos.washington.edu/~david/zwb1997.pdf

    Additionally, in an earlier post about the PDO…

    http://bobtisdale.wordpress.com/2009/04/27/misunderstandings-about-the-pdo-%e2%80%93-revised/

    …I quoted a description of how the PDO was calculated from an email from Nate Mantua from JISAO:

    “The full method for computing the PDO index came from Zhang, Y., J.M. Wallace, D.S. Battisti, 1997: ENSO-like interdecadal variability: 1900-93. J. Climate, 10, 1004-1020.
    “They labeled this same time series “the NP index” (see their figs 5 and 6). The steps are listed below, and files described below can be found at: ftp://ftp.atmos.washington.edu/mantua/pdofiles/
    “Data used:
    * monthly 5×5 Hadley Center SST 1900-93
    “Method:
    “1. create monthly anomaly fields for all grid points
    “2. create a monthly mean global SST anomaly time series for all months, 1900-93, using gridpoints specified in file grid.temp.glob_ocean.977
    “3. create a “residual SST anomaly” field for the North Pacific by subtracting out the global mean anomaly from each North Pacific grid point in file grid.N_Pac_SST.resi.172 (20N-65N, only in Pacific Basin) for all months and locations
    np_resi(mo,loc)= np_ssta(mo,loc) – global_mean(mo)
    “4. compute the EOFs of the North Pacific residual SST anomaly fields, and ignore all missing data point (set them to zeros)
    “5. the PDO index is the leading PC from the above analysis
    “6. for PDO index values post 1993, project observed ‘North Pacific residual SST anomalies’ onto the leading eigenvector (what we call the ‘PDO pattern’ of ssts) from the EOF analysis done in step 4. We now do this with the Reynold’s and Smith Optimally Interpolated SST (version 2) data.”

    Greg says: “The PDO is a single figure for each date. It in no way carries “spacial pattern” information about the N. Pacific.”

    The PDO is a spatial pattern. The PDO index is data that represents the spatial pattern. See the link in the Wikipedia page that describes Empirical Orthogonal Function analysis:

    http://en.wikipedia.org/wiki/Empirical_orthogonal_functions

    And the Wikipedia page that describes Principal Component Analysis:

    http://en.wikipedia.org/wiki/Principal_components_analysis

    Greg says: “NO. This is not what PDO measures. PDO is the deviation of N. Pacific temps from the GLOBAL AVERAGE, not from the north-western Pacific. The pattern you speak of is generally true but it is not what PDO measures nor how it is defined or calculated.”

    You overlooked a step or two. The PDO is based on the Principal Component Analysis of the North Pacific sea surface temperatures (north of 20N) after global sea surface temperatures have been subtracted from each 5X5 grid in the North Pacific. Then the results are standardized.

    Additionally, the PDO is not the deviation of the “deviation of N. Pacific temps from the GLOBAL AVERAGE”. See Figure 19, which is a comparison of the PDO to the “deviation of N. Pacific temps from the GLOBAL AVERAGE”:

    Greg writes: “If you are going to write stuff explaining what these indices are, at least present an accurate definition to start with.”

    I, Greg, have provided correct definitions of the PDO and the PDO index. You, Greg, have expressed your misunderstandings.

    Gotta get ready for work. Please read and understand the post in its entirety. Be back later to answer any additional questions you might have.

    Regards

  9. Bob, it would be very interesting to invert one of the line in your fig 23. The NW Pacific SST seems an almost perfect mirror image of PDO.

  10. Paul Woland says:
    April 21, 2014 at 2:42 am

    What do you make of the fact that a consistent trend of warming has been measured in the oceans?

    Meant to send this link too:
    http://www.nature.com/nature/journal/v465/n7296/abs/nature09043.html

    This paper shows robust warming only because of the robust adjustment to the Argo data by the last author Josh Willis. This feat of scientific fraud has made Josh a celebrity.

    The big kink at the introduction of the Argo data compared to earlier (non-Argo) data indicates nothing much can be trusted before then. The most robust finding of the Argo data (since it is self-referenced) is the observation that the deeper layers down to 2000m (about half the average ocean depth) are warming relative to the surface layers. This indicates vertical, downward movement of heat in the upper ocean, which is the primary natural mechanism of global climate change. However the absolute conclusions of Argo – i.e. overall ocean warming, cant really be trusted due to sampling statistics issues discussed here by Willis Essenbach. Thus the most important contribution of Argo is the relative depth data, which shows downward movement of heat, which is what one expects during natural climate cooling.

  11. Paul Woland says: “Bob Tisdale, so you do accept that the oceans are warming, as is demonstrated in the paper published in Nature? Could you explain what you think is the “natural” process behind this, if not the change in the atmospheric composition of greenhouse gases?”

    Obviously, you failed to read the link I provided for you in my last reply, or maybe you read it and failed to comprehend it, because I answered your question in that essay. Based on your failure to read and comprehend the link I already provided, I have to assume you’re a new troll here. And that means I get to say, I’m not going to waste my time with someone who failed to read the answer that I offered earlier.

    Adios.

  12. Bob wrote above:
    IF THE PDO LAGS ENSO, THEN THE PDO CANNOT DRIVE ENSO

    Allan wrote in January 2008
    SINCE CO2 LAGS TEMPERATURE, THEN CO2 CANNOT DRIVE TEMPERATURE

    SAME COPNCEPT: THE FUTURE CANNOT CAUSE THE PAST.

  13. DO THE CLIMATE MODELS USED BY THE IPCC PROPERLY SIMULATE THE MULTIDECADAL VARIATIONS IN THE SEA SURFACE TEMPERATURES OF THE EXTRATROPICAL NORTH PACIFIC?

    That should not have to be a germane question. What I read here from Bob is research into patterns in the Pacific ocean temperatures. Something that seems to be ignored in the IPCC, but which goes to the heart of science.

    I appreciate the explanation, and will book mark it as there is a lot of data here that requires a longer study and more careful reading. The take away however is that we are still trying to figure out the factors that go into the ethereal global temperature.

  14. Thanks for a composed and thoughtful reply. However, I think you are misreading what is being done and the short explanation given in WP that you link to.

    “6. for PDO index values post 1993, project observed ‘North Pacific residual SST anomalies’ onto the leading eigenvector (what we call the ‘PDO pattern’ of ssts) from the EOF analysis done in step 4.

    The full EOF retains all the eigen-vectors and thus does contain spacial information in some form. However, as soon as you select only one ( the leading ) eigen-vector, which is in essence a time series of a single location, you no longer have any spacial information.

    It is a VECTOR. a single line of data. How can it contain any spacial information ??

    You can think of it as one grid cell that that is the most typical of the dataset, that captures the maximum amount of the variability that is possible to represent in one unique line of numbers.

    This one line is not the same as the average of all the cells in the region which is what you show in your figure 19 , but if you invert one of the those lines you will see they are similar.

  15. Paul Woland says:
    Could you explain what you think is the “natural” process behind this, if not the change in the atmospheric composition of greenhouse gases?

    ===

    You explain why the world cooled from MWP to LIA without someone sucking all the CO2 out of the atmosphere and you will have answered your own question.

    This is what is at the heart of all the AGW stupidity. Temperatures change. If you want to ignore that and get all anal about a 25 year cherry picked period that conveniently coincides with the increase in CO2 you can pretend a vague, spurious correlation proves causation. It does not.

    Salmon farming in the western highlands of Scotland has also increased in that period. It does not prove that salmon farming is the cause of catastrophic global warming.

  16. While the PDO cannot drive ENSO it doesn’t mean something we don’t understand drives or influences both.
    There is a 60 year climate cycle caused by something. There is a patten of moraine deposits from the last deglaciation that indicated the continental glaciers reversed their decline and briefly grew at this 60 year period.
    It doesn’t matter if the PDO is the cause, its existence allows us to make a prediction on future climate that has yet to be disproven.

  17. Paul Woland,
    The cause of the observed global warming are large changes in cloud cover.
    There was a 5% decrease in cloud cover between 1990 and 2003 which corresponds to the changes observed in global temps and the subsequent pause. This is equivalent of half the total CO2 forcing over 150 years. It is supported by Observation of outgoing longwave radiation that show the earth is loosing more heat as it warms up, the opposite of GHG theory.
    It should be noted the IPCC uses a GHG theory that predates both quantum mechanics and relativity. If you understand quantum mechanics it becomes clear the GHGs both insulate the earths surface while also cooling the atmosphere.
    No one knows the net affect of adding more as a consequences.

  18. Thank you Bob, very informative as always. Your many posts over the years have been an excellent expose on the transfer of energy from the ocean to the atmosphere. The ENSO events remind me of a lava lamp (although we only get to see a 2D cross section at the ocean surface). With the energy rich waters warmed by the sun, bubbling up and releasing that energy into the atmosphere and nothern latitudes on its complex journey out to space.

  19. Allan M.R. MacRae says:
    SAME CONCEPT: THE FUTURE CANNOT CAUSE THE PAST.

    Robert JM says:
    While the PDO cannot drive ENSO it doesn’t mean something we don’t understand drives or influences both.

    ====

    While Ray has a good point in the case of CO2 and temp where the two are related directly by several known and established natural laws, I’d be more inclined to Robert’s point on SST.

    There is no inherent reason why either ENSO or PDO should be causing the other. It is simply observed similarities. Assuming causation from nothing more than the sign of lag correlation is not justified.

    PDO is not studied because one small part of the N.Pacific has some magical properties, it is because this index is representative of Pacific-wide patterns. It tends to anti-correlate with western and central N.Pac and correlate with Nino regions and another band in the S. Pacific.

    Much of this is covered in what Bob has written, though he did not look at S.Pac here.

    Since the combined North and South Pacific oceans cover almost half the globe this can be regarded as an index of a planetary scale oscillation.

    Now once a system is in oscillation, trivial ideas of how phase lead/lag relates to causation are very different from what happens after a single event.

    I’m more inclined to see common causation than one causing the other.

  20. Another thing about fig 19, PDO is voluntarily inverted from the derived eigen-vector calculation to make it comparable with ENSO.

  21. Couldn’t comment on matters extending beyond my area of expertise, but perhaps an approach from a different perspective is welcome. Approximate masses in our solar system in metric tons:
    Sun 2×10^27, Earth 6×10^21, Oceans 1.4×10^18, Atmosphere 5.15×10^15, Atmospheric carbon dioxide 3.16×10^12.

    In accordance with Einstein theory of mass-energy equivalence energy increases exponentially with mass E=mc^2. The man-made global warming proponents seem to be pointing towards something different.

  22. Jaakko Kateenkorva on April 21, 2014 at 6:11 am

    Couldn’t comment on matters extending beyond my area of expertise, but perhaps an approach from a different perspective is welcome. Approximate masses in our solar system in metric tons:Sun 2×10^27, Earth 6×10^21, Oceans 1.4×10^18, Atmosphere 5.15×10^15, Atmospheric carbon dioxide 3.16×10^12.

    Useful stats. This means, I guess, that if the atmosphere was 100% pure CO2, and in an instant this entire atmosphere dissolved into the sea, it would add one single part per millio n of CO2 to the oceans. And the resulting pH decrease wouldake all the corals dissolve like an alke seltzer.

    As Stan Laurel would have said, “it could happen”.

  23. The PDO’s potential impact on the ENSO is that warm water or warm air could flow into / influence, the equatorial sea surface temperatures. That would be through the California Current, but this current never really makes it to the equator. There is very little influence on the ENSO from the PDO.

    An old, but useful map of world surface ocean currents.

    It turns out to be the other way around. It is the ENSO that influences the PDO pattern. The warm waters of the Pacific equatorial current eventually run up against the continental shelf at Indonesia. The water piles up and has nowhere to go. It gets pushed in three different directions.

    One is down, where it enters the Pacific Equatorial Under-Current at 150 metres deep and flows back to the East (to become the next El Nino, next La Nina). One is through the Indonesian flow-through into the Indian ocean (but there is limited flow through here).

    And one is to the north (and part of this is down under the surface to the north-west for a period of time), where it eventually flows into the Kuroshio current and the north Pacific Gyre. So the past El Nino/La Nina leaves its imprint in the middle part of the PDO pattern that sticks out into the north Pacific. Even earlier El Ninos/La Ninas form the C pattern in the eastern north Pacific.

    This is the now replaced US Navy NLOM ocean model so this is an old animation but it shows the global ocean currents really well.

    The newer Hycom model zoom-ins for the last 30 days of the relevant regions.

  24. Greg Goodman says on April 21, 2014 at 5:47 am

    Since the combined North and South Pacific oceans cover almost half the globe this can be regarded as an index of a planetary scale oscillation.

    I’m more inclined to see common causation than one causing the other.
    ________________

    A fair hypo for the possibility of an ENSO/PDO “common cause” Greg.

    But what causes the ~60 year periodicity of the PDO? It is both in-and-out-of-phase with the ~90 year Gleissberg Cycle.

    I have played with this question from time to time but have no answers.

    Is there some natural ~60-year driving frequency, or is it some sort of natural harmonic, or what?

    Any valid opinions out there, backed by credible evidence?

  25. DocWat says:
    April 21, 2014 at 5:08 am

    Too much information for simple folk. Too Much!!

    Reading WUWT is like trying to drink from a fire hose.

  26. chris moffatt says: April 21, 2014 at 7:29 am
    My bad Chris. Think I had claimed something utterly ridiculous, for example, that some human caused gas determines the properties of something hundreds thousands times heavier like the globe. Imagine that. Oh, hang on.

  27. Thanks for your efforts, Bob. I found this article difficult to follow. In one sense, I came away with the impression that the PDO may be over-hyped, and not a meaningful factor in long-term climate change (periods tend to offset). I was trained too many years ago as a synoptic climatologist, but have kept up for the most part. I came away with the feeling that the dominant ocean currents are still the conveyors of warm and cold water across latitudes. Your mention of wind patterns as important also twigged the long-understood roles of semi-permanent pressure cells in moving air masses around. What I don’t see much of anymore is the development and movement of air masses. Using 20-65º for the PDO would encompass both MT and MP air masses, which still are identifiable in weather patterns across the continents. Is anyone looking at what changes in the PDO (and AMO), both in location and intensity, have on air mass characteristics, pressure fields and concomitant weather patterns. I do know the difference between weather and climate, but today’s weather patterns can teach us a lot about how long-term climatic patterns are created. I think that the importance of SST is in how air mass and pressure patterns are affected, but I don’t see much work on these relationships.

  28. Fish tell no lies. You make a compelling case that the enso dataset drives the pdo dataset. Fish do not study datasets. If these datasets are indeed representative, what we are still really looking for is the cause of the quasi 60 year cycle driving both datasets. Follow the fish.

  29. phlogiston says: April 21, 2014 at 6:36 am
    You’re right. On the other hand, homeopathy resembling solutions exist in anthroposophy. Maybe that’s the A in AGW.

  30. In response to my statement: “The problem: there is no mechanism through which the PDO can raise or lower global surface temperatures.”

    lgl says: “But the NPI can [raise or lower global surface temperatures].
    See (3) in the abstract here: http://link.springer.com/article/10.1007%2Fs00376-012-1196-7#page-1

    The abstract doesn’t say what you suggested. Having an impact on regional surface temperatures is not the same as having the ability to raise or lower global surface temperatures.

    Cheers.

  31. Nice work again Bob, much appreciated.

    “The top map shows that surface temperatures of most of the Northern Hemisphere are negatively correlated with the PDO index. That negative correlation means that when the PDO data increases, the surface temperatures for most of the Northern Hemisphere show cooling.”

    On this point, unless its being misinterpreted by me, I will have to strongly disagree. The correlation between the value of the PDO and global temperatures is very strong. -PDO correlates with global cooling(or a halt in global warming) +PDO correlates strongly with global warming.

    Are you saying the opposite is the case? Or am I missing something?

    Here’s a graph that shows it, along with including the AMO(just compare the top graph-PDO and bottom one-Temp)

    I am not saying that the PDO causes global temps to go up or down, just that the sign of the PDO index CORRELATES with global warming and/or cooling(lack of warming) very strongly.

    In the last decade, the evidence is strong that the PDO index shift from positive to negative is correlating nicely with the stall in warming(we’ve had La Nina’s dominating). In the 80’s/90’s, with global temps in an uptrend, there was a mainly +PDO(yes, with El Nino’s dominating) .

  32. Thanks Bob.
    —————-
    Fig. 27 could be clearer if the color of NINO3.4 etc. in the “title” matched the color in the chart – such as done in Fig. 28 and those following.

  33. If I have understood the main points of this article, it is ENSO which in large part drives the PDO, not reverse. But then there is the question: why are there series of more El Niño’s and series of more La Nina’s which make that there is a 60-80 years oscillation visible in the PDO? Which correlates with the warming and standstill periods of global temperature? Similar for the NAO?

  34. Interesting.

    There is interest here in the Pacific Northwest in the PDO index as it seems to be related to land temperatures in the region.

    Annual temperatures at Corvallis Oregon, the home of Oregon State University are plotted below along with the PDO index.

  35. Bob
    “The AL is stronger and is located toward the east during strong El-Nino winters …” and strong El Ninos give global warming, or have you changed your mind?

  36. Allan M.R. MacRae says:

    Greg Goodman says on April 21, 2014 at 5:47 am

    Since the combined North and South Pacific oceans cover almost half the globe this can be regarded as an index of a planetary scale oscillation.

    I’m more inclined to see common causation than one causing the other.
    ________________

    A fair hypo for the possibility of an ENSO/PDO “common cause” Greg.

    But what causes the ~60 year periodicity of the PDO? It is both in-and-out-of-phase with the ~90 year Gleissberg Cycle.

    I have played with this question from time to time but have no answers.

    Is there some natural ~60-year driving frequency, or is it some sort of natural harmonic, or what?

    ==================

    If you look at Bob’s fig 17 , if there is a “cycle” in the SST of the PDO base region, it’s more like 45y.

    I suspect that the famous circa 60y pseudo cycle in AMO and global SST is an interference pattern of circa 9.1 y cycles of lunar origin ( Scafetta and Best land SAT ) and 10.4y solar .

    cos ( 2.pi/9.1 ) + cos ( 2.pi/10.4 ) = cos (2.pi/145.6) * cos ( 2.pi/9.7 )

    This is a 9.7y cycle with an amplitude modulation with an envelop ‘beat’ frequency of 73 years.

    This is the reason for the “phase crisis” when trying to link global temps to solar ( SSN ) cycles. It is not the only player, so it drifts in and out of phase. The fact that the drift is fairly consistently in the same direction , not arbitrary, lends weight to this interpretation.

    I see the variation in N.Atl SST to be more like this kind of fully rectified cosine , abs(cos) envelop, than a pure symmetric harmonic:

    http://climategrog.wordpress.com/?attachment_id=215

    I think we are starting to see this kind of abs(cos(2.pi/130y)) pattern forming in the Arctic too:

    In fact many of these “cycles” that pop out of spectral analysis seem to be rectified bumps, not full cosines.

    The “9.1” of Scafetta and Best papers , is IMO 9.05 years, as I detected in cross-correlation of N.Pac and N.Atl SST.

    http://climategrog.wordpress.com/?attachment_id=755

    Whether that is still there once you’ve subtracted the global mean from the top right corner of the N.Pac SST , I have no idea, though IIRC the BEST paper did C-C AMO and PDO rather than the real SST as I did.

  37. The above works into Bob’s ENSO pump hypothesis. As I’ve said before I think it’s amplitude of the oscillation that warms or cools the planet. Larger swings cause warming , periods of lesser swings allow radiative cooling.

    The moderate ENSO events of the last 10-15 years are just below what is needed to maintain a steady global SST, cooling is starting to set in.

  38. Some time ago Bob correctly criticised me for linking PDO with the 60 year climate cycle which involves a shift between El Nino dominance and La Nina dominance.

    He suggested that instead one should refer to the Pacific Multi-Decadal Oscillation (PMO) rather than the PDO and I have done so ever since. I thank Bob for that insight.

    The thing is that the PMO (not PDO) reveals upward temperature stepping during the 20th century which rather defeats the idea of the PDO as being thermally neutral over time.

    My proposal is that solar influences such as from LIA to date produce upward stepping in the PMO from one 60 year cycle to the next whereas solar influences such as from MWP to LIA produce downward stepping in the PMO from one 60 year cycle to the next.

    I appreciate that Bob prefers not to speculate beyond his data and that data is limited to recent decades but to get some idea as to how the sun and oceans interact over millennia I do think one has to go beyond Bob’s data.

    One then needs to ascertain how one could arrive at such long term changes in the balance between El Nino or La Nina dominance over multiple centuries which leads us to this proposition:

    http://www.newclimatemodel.com/new-climate-model/

    Quite simply, solar variability affects the proportion of Total Solar Irradiance (TSI) that enters the oceans which then skews the long term balance between El Nino and La Nina which then affects global air circulation.

    The present situation would appear to be that low solar activity is causing more meridional jet stream tracks and greater global cloudiness with increased global albedo for reduced solar energy into the oceans and net global cooling.,

    During the late 20th century we had the opposite scenario.

  39. lgl says: “‘The AL is stronger and is located toward the east during strong El-Nino winters …’ and strong El Ninos give global warming, or have you changed your mind?”

    I haven’t changed my mind. But the fact that the Aleutian Low is stronger and located east during strong El Nino winters still does not indicate that the NPI can raise or lower global surface temperatures as you suggested earlier.

    Regards

  40. Mike Maguire says: “On this point, unless its being misinterpreted by me, I will have to strongly disagree. The correlation between the value of the PDO and global temperatures is very strong. -PDO correlates with global cooling(or a halt in global warming) +PDO correlates strongly with global warming.
    “Are you saying the opposite is the case? Or am I missing something?”

    And you provided a link to the following graph:

    The fact that the PDO was positive during warming periods and negative during hiatus/cooling periods is simply a coincidence. Both are responding to the multidecadal variations in the strengths of ENSO events, and global surface temperatures are also varying in response to the multidecadal variations in the sea surface temperatures of the North Atlantic and of the North Pacific (the latter of which are not represented by the PDO). And of course, there is no mechanism through which the PDO can cause the multidecadal variations in global surface temperatures, since the PDO does not represent sea surface temperatures.

    The reality of the situation is, the annual PDO index correlates negatively with global surface temperature data (-0.43 correlation coefficient) from 1982 to 2013. And from 1900 to 2013, the correlation coefficient of the annual data is about as low as you can go: a correlation coefficient of 0.06.

  41. Greg says:
    April 21, 2014 at 4:33 am

    “It is a VECTOR. a single line of data. How can it contain any spacial information ??”

    By itself, it could not. But, as a vector of relative amplitudes of a functional basis, it could.

    I’m writing as one who has not delved into the particular application so I could be wrong, but I assume it is something like a set of normal modes, and the PDO index is basically a measure of the time varying amplitude of the fundamental mode.

    Would that be a fair description, Bob?

  42. Thanks, Bob. I appreciate the effort you have put into this.

    So what, in your opinion, is the main reason for calculating a PDO index at all if it is just a logical complement (mirror image) of the Kuroshio-Oyashio Extension? What value does it really add to the raw data?

    (I can see that it may be of value to the person(s) who perform and publish the calculations.)

  43. Greg quoted step 6 the Mantua email: “6. for PDO index values post 1993, project observed ‘North Pacific residual SST anomalies’ onto the leading eigenvector (what we call the ‘PDO pattern’ of ssts) from the EOF analysis done in step 4.”

    Then Greg says: “The full EOF retains all the eigen-vectors and thus does contain spacial information in some form. However, as soon as you select only one (the leading) eigen-vector, which is in essence a time series of a single location, you no longer have any spacial information”…

    Greg, you’re focusing on step 6, which is how JISAO prepared the post-1993 data, not the original data. I’m discussing how the original data is determined, and you’re discussing how JISAO adapts the current SSTa data to the old index. In other words, to support your misunderstanding/misinterpretation, you appear to be overlooking the earlier steps. And you highlighted “the leading eigenvector” in your quote, but failed to acknowledge the fact that Mantua had further defined “the leading eigenvector” as “what we call the ‘PDO pattern’ of ssts”.

    Greg closed that comment: “This one line is not the same as the average of all the cells in the region which is what you show in your figure 19 , but if you invert one of the those lines you will see they are similar.”

    I presented Figure 19 in my reply to you to contradict your earlier statement that the “PDO is the deviation of N. Pacific temps from the GLOBAL AVERAGE…” It clearly is not.

    And of course they are similar if you invert them. That was one of the points I was trying to make. If both datasets are smoothed with multiyear filters, they show similar multidecadal variations, but the PDO data are inversely related to the difference between global and North Pacific sea surface temperatures.

    However, the monthly PDO and “North Pacific Residual” data show little similarity. The correlation coefficient of those two datasets is -0.51, further confirming that the “PDO is [NOT] the deviation of N. Pacific temps from the GLOBAL AVERAGE…”.

    Greg Goodman says: “There is no inherent reason why either ENSO or PDO should be causing the other. It is simply observed similarities.”

    The processes through which ENSO creates the PDO pattern are well known and easy to describe. In fact, I described them, in part, years ago. See the post here:

    http://bobtisdale.wordpress.com/2010/09/14/an-inverse-relationship-between-the-pdo-and-north-pacific-sst-anomaly-residuals/

    There I wrote with respect to the sea surface temperature of the Kuroshio-Oyashio Extension (the west-central portion of the extratropical North Pacific):
    During La Niña events, Pacific trade winds strengthen, which reduces cloud cover over the tropical Pacific. This increases the amount of Downward Shortwave Radiation (visible light) reaching the ocean surface and, in turn, warms the tropical Pacific. The warmer water is pushed to the west by the trade winds and is carried northward by the western boundary current, the Kuroshio Current. Then the warm water is carried eastward by the western boundary current extension, the Kuroshio Extension. This is why there is the area of warm SST anomalies east of Japan during La Niña events. During El Niño events, the trade winds decrease or reverse and less warm water than normal is carried from the tropics up to the Kuroshio Extension.

    [End reprint from linked post]

    And now for the description of the warming of the eastern North Pacific during an El Niño. The sea surface temperatures of the eastern North Pacific warm in response to an El Niño due to changes in atmospheric circulation caused by the El Niño and as a result of coastally trapped Kelvin waves. After the downwelling (warm) equatorial Kelvin wave reaches the coast of the Americas, it transforms into coastally trapped Kelvin waves and they propagate poleward. Small eddies then separate from the coastally trapped Kelvin waves and they move very slowly westward as Rossby waves.

    Greg Goodman says: “PDO is not studied because one small part of the N.Pacific has some magical properties, it is because this index is representative of Pacific-wide patterns.”

    The PDO only represents the spatial patterns of the Extratropical North Pacific. Nothing more, nothing less. You’re reading much more into the PDO data than what it really represents. ENSO dominates the Pacific as a whole, and its aftereffects appear in the South Pacific as well. See Shakun and Shaman (2009) “Tropical origins of North and South Pacific decadal variability”:

    http://onlinelibrary.wiley.com/doi/10.1029/2009GL040313/abstract

    The Shakun and Shaman (2009) Conclusions read:
    “Deriving a Southern Hemisphere equivalent of the PDO index shows that the spatial signature of the PDO can be well explained by the leading mode of SST variability for the South Pacific. Thus, PDV appears to be a basin-wide phenomenon most likely driven from the tropics. Moreover, while it was already known PDV north of the equator could be adequately modeled as a reddened response to ENSO, our results indicate this is true to an even greater extent in the South Pacific.”

    And thank you for acknowledging that the PDO represents spatial patterns.

    Greg Goodman says: “Much of this is covered in what Bob has written, though he did not look at S.Pac here.”

    There is no reason to look at the South Pacific in a discussion of the PDO. The PDO data only represent the spatial patterns of the extratropical North Pacific, not the tropical Pacific and not the South Pacific.

    Greg Goodman says: “Since the combined North and South Pacific oceans cover almost half the globe this can be regarded as an index of a planetary scale oscillation.”

    The Pacific is big but not that big. The Pacific only covers 33% of the surface of the Earth…a little more than all of the continental land masses combined.

    And yes, ENSO has global impacts. It has been well understood for decades that ENSO is the dominant mode of variability on the planet. See Rasmusson and Wallace (1982) “Meteorological aspects of the El Niño/Southern Oscillation”.

    http://marineecology.wcp.muohio.edu/climate_projects_04/el_nino/ecology222.pdf

  44. Greg: It is a VECTOR. a single line of data. How can it contain any spacial information ??

    It’s all in the indexing. A matrix is merely a long vector with a formula for turning a two dimensional index i,j into a one-dimensional index. k

  45. “It’s all in the indexing. ”

    very amusing. I suppose you have enough points in your version of PDO to cover the spacial distribution of the whole gridded array.

  46. Bart says: “I’m writing as one who has not delved into the particular application so I could be wrong, but I assume it is something like a set of normal modes, and the PDO index is basically a measure of the time varying amplitude of the fundamental mode.”

    No. It’s ONE basis function , the leading component only. It has no “spacial” information. Period.

  47. Bart says: “I’m writing as one who has not delved into the particular application so I could be wrong, but I assume it is something like a set of normal modes, and the PDO index is basically a measure of the time varying amplitude of the fundamental mode.”

    Or to paraphrase you and use terms that have appeared throughout this post, the PDO index is basically a measure of the time-varying amplitude of the dominant spatial pattern.

  48. Greg Goodman says: “No. It’s ONE basis function , the leading component only. It has no “spacial” information. Period.”

    Even NOAA disagrees with you, Greg. See their Pacific Decadal Oscillation (PDO) webpage:

    http://www.ncdc.noaa.gov/teleconnections/pdo.php

    NOAA writes:
    When SSTs are anomalously cool in the interior North Pacific and warm along the Pacific Coast, and when sea level pressures are below average over the North Pacific, the PDO has a positive value. When the climate anomaly patterns are reversed, with warm SST anomalies in the interior and cool SST anomalies along the North American coast, or above average sea level pressures over the North Pacific, the PDO has a negative value (Courtesy of Mantua, 1999).

    In fact, I think I’ll update the post to include that quote.

    Cheers

  49. Thank you Greg – I’m offline for a while but will study your response later.

    Same question for Bob and all:

    What causes the ~60 year periodicity of the PDO? It is both in-and-out-of-phase with the ~90 year Gleissberg Cycle.

    Is there some natural ~60-year driving frequency, or is it some sort of natural harmonic, or what?

    Best, Allan

  50. there is no mechanism through which the PDO can raise or lower global surface temperatures, because the PDO does not represent the surface temperatures of the extratropical North Pacific (where the PDO is derived).

    I have to disagree. The locations of the warm and cold areas in the Pacific influence the jet stream. Changes in the jet stream can lead to changes in overall global clouds and their latitudinal positioning. Both of these can then influence global temperatures. This may not be the entire reason for global changes but it can have an influence.

    As I’ve mentioned before, I use the PDO index because it does correlate to global temperature trends. The planet warms during +PDO phases and cools during -PDO phases. If my above paragraph is true then, in fact, the PDO phases do have an influence of global temperatures. Bob, you need to look beyond the average water temperatures.

    Now, my own view is that both the PDO and ENSO are driven by yet another mechanism. Others have indicated this view as well. So, we end up with a chain of cause and effect. Although there is no data that I have seen to support Dr. William Gray’s thoughts, they do make sense. The driver would be the MOC (deep ocean currents). The speed of the MOC varies over time. When it speeds up you have a -PDO, when it slows down you have a +PDO.

    The faster MOC leads to more upwelling cold water. This tends to reduce the impact of El Niño which then leads to a -PDO pattern (and vice versa).

    Finally, the generally warmer oceans could be a direct result of increased solar warming. No need to invoke GHGs at all. The atmospheric warming does not correlate closely because it is tied to the PDO cycle which is driven by the MOC. The energy is released more strongly during EL Niño events. The stronger the El Niño events the more warming that can occur.

  51. Lorenz classic paper “Deterministic Nonperiodic Flow” 1962 showed how a climate simulation with no forcing can alternate between different regimes. This was using an incredibly primitive Freddy Flintstone computer. The different phases were described in phase-space terms as the two wings of the Lorenz butterfly attractor. In a chaotic system they represent periods with different probabilities of certain states, e.g. periods where el Nino and La Nina respectively predominate in ENSO:

    http://www.astro.puc.cl/~rparra/tools/PAPERS/lorenz1962.pdf

    Figure 1 (third panel) and figure 2 show the basis of the PDO.

    Since Bob himself now also describes ENSO as a nonlinear oscillator, why not make the logical step to accepting PDO as an epiphenomenon of this oscillator, i.e. a biphasic Lorenz (butterfly) attractor?

  52. Greg Goodman says:
    April 22, 2014 at 1:04 am

    It is essentially the solution of a separable partial differential equation which is the sum of products of the type

    f(t,x,y,z) = g(t)*h(x,y,z)

    The PDO index is essentially the function g(t) for the dominant such term. The spatial distribution is in the function h(x,y,z), separate from the amplitude function g(t). What I gather is that, transformed to spherical coordinates, h(r,theta,phi), within the boundaries of the North Pacific, is an odd function of the longitudinal variable theta with central node somewhere in the middle of the ocean. Something very like the animation I referred to earlier with East-to-West looking into the page, and the nodal line looking crossways North-to-South.

  53. Allan M.R. MacRae says:
    April 22, 2014 at 5:20 am

    I have been considering lunar tidal forcing. It is, I suspect, portentous that the 9.3 year cycle of the radius of nutation beating against the 11 year solar cycle gives us a period of 60 years. Some are wont to quibble, as these are all approximate values for average bulk processes, but I’m not propounding a new theory (despite the link’s headline – it wasn’t mine, I just got a comment elevated to a full post) and calling it TRUTH, yet. It’s a lead for further investigation.

  54. “The fact that the PDO was positive during warming periods and negative during hiatus/cooling periods is simply a coincidence.”

    “of course, there is no mechanism through which the PDO can cause the multidecadal variations in global surface temperatures, since the PDO does not represent sea surface temperatures.

    The reality of the situation is, the annual PDO index correlates negatively with global surface temperature data (-0.43 correlation coefficient) from 1982 to 2013. And from 1900 to 2013, the correlation coefficient of the annual data is about as low as you can go: a correlation coefficient of 0.06″

    Again Bob, I am not saying that the PDO index is causing global temperature changes. I am not sure what the mechanism is that causes global warming, cooling and in the case of the last 15 years, offsetting of greenhouse gas warming(cooling influence).

    I am very confident that the sign of the PDO correlates(not causes) strongly with global temperatures.

    A case can be made for COINCIDENCE when global temperatures go up in the 1980’s and 1990’s with CO2. It can be shown to be a coincidence if you look at the last 15 years, when CO2 direction and global temperatures were not in lock step. In the 1940’s-70’s with modest global cooling and increasing CO2, one has to again question the correlation, at least the strength of it vs natural/other factors that were at times, even more powerful than CO2 warming.

    With the PDO and global temps, the approx. 30 year periodicity has repeated enough times, with both displaying the same trend, in the same direction at the same time(global temps have a lag of a few years), to provide powerful evidence of a correlation relationship.

    Maybe its because of the lack of a clear cut physical explanation of what is causing them both to move as described that makes it hard for you to embrace this idea.

    It’s possible this is a coincidence, just like you can flip a coin and get heads or tails 6 times in a row because of random variation. Meteorological events in a chaotic short term atmospheric environment especially follow random variation at times.

    However, the chance that the PDO and global temperatures would have changes in direction and time that match up so well based on a climatic time scale, going back a century, to me would have a miniscule probability of occurring from random variation coincidence.

    Possible though.

  55. Greg Goodman: very amusing. I suppose you have enough points in your version of PDO to cover the spacial distribution of the whole gridded array.

    The word is “spatial”. Your supposition is irrelevant to the point. Have you never done any analysis or programming with arrays: vec, vech operations and such?

  56. “A case can be made for COINCIDENCE when global temperatures go up in the 1980′s and 1990′s with CO2. It can be shown to be a coincidence if you look at the last 15 years, when CO2 direction and global temperatures were not in lock step. In the 1940′s-70′s with modest global cooling and increasing CO2, one has to again question the correlation, at least the strength of it vs natural/other factors that were at times, even more powerful than CO2 warming.”

    Let start over on the confusing paragraph above, stated earlier.:
    CO2 and global temperatures both went up in the 1980’s/90’s. Coincidence? or Causation?
    CO2 has continued up since then but global temperatures have stalled. This weakens the causation argument with regards to anomalous strength of greenhouse gas warming. So does modest cooling in the 1940’s to 1970’s with increasing CO2 at the time.

    Though our planet has warmed around 1 degree the past 150 years, the warming comes in surges that match up with the +PDO. When we are not warming, it matches up with the -PDO.
    The warming has exceeded cooling, El Nino’s more than La Nina’s.

    It appears to me, that the PDO index, being mostly positive for around 30 years, followed by the PDO being mostly negative for around 30 years, then repeating again, matches up extremely well with the same periodicity of global temperature regimes.
    You can superimpose these approx 30 year global temperature regimes on top of a slowly increasing global temperature line with a slope that has not changed much for over the past 100 years.

  57. Mike Maguire says:
    April 22, 2014 at 2:28 pm (replying to)

    “A case can be made for COINCIDENCE when global temperatures go up in the 1980′s and 1990′s with CO2. It can be shown to be a coincidence if you look at the last 15 years, when CO2 direction and global temperatures were not in lock step. In the 1940′s-70′s with modest global cooling and increasing CO2, one has to again question the correlation, at least the strength of it vs natural/other factors that were at times, even more powerful than CO2 warming.”

    Let start over on the confusing paragraph above, stated earlier.:
    CO2 and global temperatures both went up in the 1980′s/90′s. Coincidence? or Causation?
    CO2 has continued up since then but global temperatures have stalled. This weakens the causation argument with regards to anomalous strength of greenhouse gas warming. So does modest cooling in the 1940′s to 1970′s with increasing CO2 at the time.

    Mike!

    You’re working too hard, but yes, the referenced paragraph needs to be simplified.

    When CO2 was steady over 450 years, apparent air temperatures fell.
    When CO2 was steady over 400 years, apparent air temperatures rose.
    When CO2 was steady for 20 years, measured air temperatures rose.
    When Co2 was steady for 15 years, measured air temperatures were steady.
    When CO2 was steady for 20 years, measured air temperatures declined.
    When CO2 rose for 20 years, measured air temperatures decreased 1/5 of one degree.
    When Co2 rose for 10 years, measured air temperatures were steady.
    When CO2 rose for 23 years, measured air temperatures rose 1/4 of one degree.
    When CO2 rose for 17 years, measured air temperatures were steady.
    Are steady now.
    And might remain steady in the future.

  58. The north Pacific trade winds correlate with a warm Kuroshio extension. That makes sense in Bob’s model because the winds are piling the hot water up, but it would be a positive feedback if it were a cause rather than a correlation.

    It makes no sense as a cause because the warm surface water should warm the atmosphere above and create surface low pressure. One might expect this would foster the Aleutian low, but it does not seem to. In fact, the Aleutian low appears to await the arrival of a cold Kuroshio tongue.

    The PDO is essentially a salmo index but it applies well (though antiphase) to anchovy and herring in California and China. The only fish index that follows enso is the Peruvian anchovy.

    For a tour de force see Klyashtorin:

    http://geosciencebigpicture.com/?attachment_id=1138

    His time frequency spectral analysis indicates that the power of the 50-70 year signal has been steadily increasing for the last millennium.

  59. Richard M says: “I have to disagree. The locations of the warm and cold areas in the Pacific influence the jet stream. Changes in the jet stream can lead to changes in overall global clouds and their latitudinal positioning. Both of these can then influence global temperatures. This may not be the entire reason for global changes but it can have an influence.”

    Sorry for the delay in replying. The correlation maps show that changes in atmospheric circulation caused by the PDO have a inverse impact on global surface temperatures.

    Regarding the rest of your comment, you discussed MOC and upwelling, but once again, the PDO does not represent those processes. And without data, your comment is, sorry to say, speculation.

  60. Tisdale is entrely correct in pointing out that the (elaborately manufactured) PDO index tends to persistently lag the NINO3.4 temperature index; thus the former cannot drive latter. And he correctly identifies the NP barometric variations as a factor influencing PDO values. But the spectral structure of the two indices is strikingly different, with more than half of PDO variance found at transdecadal frequencies, as opposed to only a fifth for NINO3.4. Furthermore, the cross-spectral coherence between the two is quite marginal at those low frequencies. This indicates that the influence of NPI is dominant there, much as SOI is dominant vis-a-vis the trade winds in the NINO3.4 region. Although the North Pacific is downstream in the hemispheric circulation, the claim that ENSO drives the PDO is tenous, at best.

  61. Bob Tisdale says on April 24, 2014 at 8:08 am

    Allan M.R. MacRae says: “What causes the ~60 year periodicity of the PDO?”

    Bob says: You’re assuming there’s a 60-year period. Not all paleoclimatological studies agree:

    http://bobtisdale.wordpress.com/2010/03/15/is-there-a-60-year-pacific-decadal-oscillation-cycle/

    Thank you Bob, I became aware of this range of opinion in 2002, when I asked Paleoclimatologist Tim Patterson when it would start getting colder again. Tim replied immediately “2020 to 2030″, based on his research and the ~80-90 year Gleissberg Cycle. I questioned Tim at the time if he was sure about this, and asked if it was it not possibly a shorter ~60 year cycle based on the PDO. Tim was convinced at that time that the cycle was longer, about 80-90years, and was related to the Gleissberg.

    Regrettably , I no longer have the time necessary to devote to this matter, and can only ask the occasional question.

Comments are closed.