The Pacific Decadal Oscillation and Australia - plus more

Here is the article about the Pacific Decadal Oscillation, with particular reference to Australia. I'll warn you in advance - it's long, a bit meandering, probably could do with more illustrations, and not a complete guide to everything. That's my excuses out of the way. I figure I've spent enough time on it so here it is. Feel free to add what you know and correct what I don't :)

In the previous article, I referred to some comments by a contributor, trying to argue that Australia's record heat of the 2012-13 summer was caused by a (non-existent) spike in the PDO index, or was an advance reaction to a spike that had not yet appeared. His comments were not easy to follow. Thing is, the PDO index wasn't positive during the Australian summer of 2012-13. In fact it did not register as positive until January 2014. There's more to it than that, in any case.

This is some of what Rum Runner wrote:

October 8, 2014 at 5:48 PM
How much of the temperature spike in 2013 was due to the spike in the PDO at that time? In fact Australian temperatures in general broadly follow the PDO (not surprisingly).
Plot:
http://oi59.tinypic.com/29p25u0.jpg
From data:
http://www.bom.gov.au/web01/ncc/www/cli_chg/timeseries/tmax/0112/aus/latest.txt
http://www.woodfortrees.org/data/jisao-pdo/from:1910

Here is Rum Runner's chart (as always, click the chart to enlarge it):

Chart 1: from Rum Runner

Here is another chart of the same thing, with an average annual PDO index plotted against the average annual temperature anomaly from 1961-1990 for Australia. The PDO index can be positive or negative. When positive it's referred to as being in a warm phase. When negative it's referred to as being in a cool phase.

Chart 2: Data sources: BoM and Nate Mantua

In Chart 2 above, the PDO index is the shaded bit coloured tan and the temperature anomaly is the blue line. I've also marked the different PDO phases as recognised in the literature as far as I can tell. (The period since the late 1990s is a bit of an odd one. It has a dip in the index, followed by a rise, followed by another dip. It's also behaved a bit differently to prior periods, which I'll touch on later.) Chart 2 also shows more clearly the fact that the temperature (the green line in Chart 1 and the blue line in Chart 2) is quite a bit lower in the first half of last century. Temperatures are going up as greenhouse gases warm the earth more. As you can see, the annual mean temperature for Australia as a whole has been going up and is considerably higher in the second half of the twentieth century, particularly so the past few years, with 2013 being the hottest year on record by a long shot.

In Chart 3 below, I've selected a shorter timeline so you can see what's happened recently more clearly.

Chart 3: Data sources: BoM and Nate Mantua

As you can see in Chart 3, there was no "spike" in the PDO index over the summer 2012-13 or the period leading up to it. The values from September 2012 through to February 2013 were all negative as follows:  -2.21;  -0.79;  -0.59;  -0.48;  -0.13;  -0.43. During the record hot year 2013, the PDO index stayed negative all year, even dropping as low as -1.25 in July 2013.

If you look closely at the period from 1980 onwards (Chart 3 above), you can see that the annual mean temperature of Australia dropped in the early 1980s while the PDO was in its warm phase. It jumped up in 1990 when the PDO dipped briefly to negative. in the late 1990s it was a tad out of phase with the PDO index but then in the early 2000s it appeared to be more in phase briefly. Then when the PDO went negative again around 2007, the temperatures remained high. It wasn't until the massive La Ninas of 2010-11 that the temperature dropped again. And despite the PDO being in a cool phase in 2012-13 (and in the absence of an El Nino), 2013 was Australia's hottest year on record.

In other words, on face value, there doesn't appear to be a strong relationship between the PDO and temperatures averaged across all Australia. If you were to rely on eyeballing the chart, you'd probably conclude that any period where the PDO aligns more or less with a change in surface temperature in Australia looks to be mere coincidence. But it would be wrong to conclude that there is no known association between the PDO and Australian weather.

The phases of the PDO and the related IPO do coincide with patterns of weather in Australia. Thing is, different parts of Australia behave differently during the two phases. But before discussing that, I'll just say a little bit about the PDO itself.

About the Pacific Decadal Oscillation

The name "Pacific Decadal Oscillation" was coined by Stephen Hare in his PhD dissertation of 1996, "Low Frequency Climate Variability and Salmon Production" (no link). Dr Hare is a fisheries biologist, now working at the Secretariat of the Pacific Community advising on scientific issues related to management of tuna stocks. Back in 1997, as part of his PhD research, he co-authored, with Nate Mantua and others, what has become the second most cited paper in fisheries history. (According to Google Scholar it has been cited 4,342 times, which is very impressive. I believe this paper, "Stages of embryonic development of the zebrafish", is the most cited fisheries paper, cited 5077 times according to Google Scholar.)

The 1997 Mantua/Hare et al paper is one many of you may be familiar with or heard of: "A Pacific interdecadal climate oscillation with impacts on salmon production." Nate Mantua and Stephen Hare later wrote another paper in 2001, simply called "The Pacific Decadal Oscillation". This oceanic phenomenon has been studied quite a lot, and continues to be a subject of active research.

Nate Mantua describes the PDO as follows:

The "Pacific Decadal Oscillation" (PDO) is a long-lived El Niño-like pattern of Pacific climate variability. While the two climate oscillations have similar spatial climate fingerprints, they have very different behavior in time. 

He has developed a PDO index, described as:

"the leading PC [Principal Component] of monthly SST anomalies in the North Pacific Ocean, poleward of 20N. The monthly mean global average SST anomalies are removed to separate this pattern of variability from any "global warming" signal that may be present in the data." 

There is a bit more information on this provided in Mantua and Hare (2001), where they write:

A PDO index developed by Hare (1996) and Zhang (1996), also used by Mantua et al. (1997), is the leading PC from an un-rotated EOF analysis of monthly, “residual” North Pacific sea surface temperature (SST) anomalies, poleward of 20°N for the 1900–1993 period of record (see lower panel of Fig. 1). “Residuals” are here defined as the difference between observed anomalies and the monthly mean global average SST anomaly (see Zhang et al., 1997).

They continue by describing the persistence of the sign of the index on the decadal scale:

A remarkable characteristic of this index is its tendency for multiyear and multidecadal persistence, with a few instances of abrupt sign changes. Based on a variety of studies, sign changes beginning in 1925, 1947, and 1977 have been labeled regime shifts (Hare and Francis, 1995; Zhang et al., 1997; Mantua et al., 1997; Minobe, 1997).

Warm and cool phases of the PDO

As described on the NOAA website, the PDO shifts between a positive phase (warm) phase and a negative (cool) phase. The PDO has a positive value (warm phase) when sea surface temperatures (SSTs) are anomalously cool in the interior North Pacific, while the Pacific Coast (of North America) is anomalously warm. And when the opposite happens, a cool anomaly along the Pacific Coast and anomalously warm interior North Pacific, then it's known as a cool phase and the index is negative. In the warm phase, the sea level pressures are below average over the North Pacific. In the cool phase the sea level pressures over the North Pacific are above average.

Nate Mantua provides a diagram illustrating the difference between a typical warm phase and a typical cool phase in the northern hemisphere winter:

Typical wintertime Sea Surface Temperature (colors),
 Sea Level Pressure (contours) and surface windstress (arrows) anomaly patterns during warm and cool phases of PDO.
Warm phase (positive PDO index) is on the left and the cool phase (negative PDO index) is to the right.
Source: Nate Mantua

The PDO vs ENSO

Mantua and Hare (2001) describes three characteristics that distinguish the PDO from the El Nino Southern Oscillation (ENSO) (my dot points):

• first, 20th century PDO events persisted for 20–30 years, while typical ENSO events persisted for 6 to 18 months; 
• second, the climatic fingerprints of the PDO are most visible in the North Pacific/North American sector, while secondary signatures exist in the tropics, while the opposite is true for ENSO; and 
• third, the mechanisms that cause PDO are not currently known, while causes for ENSO are relatively well understood (Mantua et al., 1997; Zhang et al., 1997).

Since 2001 there has been a lot of research exploring the PDO, and others have developed different ways of measuring it as well. Kevin Trenberth and John Fasullo, in their 2013 paper, "An apparent Hiatus in Global Warming", wrote: "While the ENSO interannual variations are reasonably well known, the decadal variations are not." They were referring to the Atlantic Multidecadal Oscillation (AMO) and the Pacific Decadal Oscillation (PDO) and associated Interdecadal Pacific Oscillation (IPO). This isn't surprising because the phases last for a long time. Reliable observations only go back so far, and sea surface temperatures are probably more iffy than land surface ones would be over the past century or so.

Is the PDO changing?

I came across an abstract from AGU13 Fall Meeting, by Stephanie McAfee, which suggested that the PDO these days isn't the same as it was earlier last century (extract, my paras):

...There are a number of critical differences in the climatic expression of the negative PDO of similar magnitude before and after 1976. In comparison to negative PDO winters between 1949 and 1976, recent negative PDO winters were associated with warmer SSTs in the north Pacific, a stronger Aleutian Low, more pronounced ridging at 500 hPa over western Canada, substantially higher temperatures over Alaska and western Canada, and wetter conditions over the central United States.

The difference in climate conditions associated with negative PDO indices between the mid-20th century and early 21st century suggests that either equivalent PDO index values may be derived from distinct climate forcings or that the PDO's teleconnections may not be temporally stable. ...

She's since published a paper that looks to be an expansion of this same topic, though I've only seen the abstract.

Seasonality of the PDO and leading ENSO

There are a few other things worth noting about the PDO before I move onto talking about Australia. Several papers discuss the seasonality of the PDO. For example, Newman et al (2003) write about comparing two years while the PDO is in a particular phase. It is mostly highly correlated with itself (two years later) in the northern hemisphere late winter/early spring and there's near zero correlation in November.

Further, there is reported to be a strong relationship between ENSO and the PDO (which isn't surprising). Matthew Newman et al (2003) described this as (my emphasis):

It has been suggested that decadal variability in the North Pacific may be more prominent during summer than during winter (e.g., Zhang et al. 1998). Our results suggest just the opposite: the PDO has little multiyear persistence during summer, so decadal variability of North Pacific SST is largely a winter/spring phenomenon. The confusion occurs because there is some correlation of the PDO between consecutive summers, but this is likely a consequence of the annual cycle of ENSO and the strong ENSO–PDO relationship that exists in both summer and winter (Fig. 1d); that is, a growing ENSO forces the PDO in summer, the subsequent mature phase of ENSO forces the PDO the following winter/spring, and this North Pacific SST anomaly then persists into the early part of the following summer.

The atmospheric bridge

That study also suggested that more of the decadal variability is reflected in the sea surface temperature rather than in the atmosphere. "Consistent with earlier studies, the NPI [North Pacific index] results suggest that the atmospheric component of North Pacific decadal variability is weaker than the SST component."

Kevin Trenberth and James Hurrell developed the North Pacific index as defined in their paper from 1994. As they wrote:

It has often been stated that the atmosphere has a very short memory while the ocean has enormous thermal inertia so that it is the ocean that provides the important memory for climate variations. In fact, however, it is the collaborative interaction between the atmosphere and the ocean and other parts of the climate system that gives rise to the important climate variations.

The NPI is described on the NCAR UCAR website as:

The North Pacific Index (NP index or NPI) 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 NPI is a measure of how strong the Aleutian low is during the northern winter, and "encompasses the area of maximum sea level pressure variance in the North Pacific for all months". I won't be getting into a discussion of the Aleutian low or any part of the atmospheric/ocean interaction in this article. Not because it's not important, it is central to any in-depth understanding of PDO/ENSO science. It's because I want to move on to the PDO and Australia. And, yeah, it would probably take another few days to come to grips with the subject matter. It's discussed a bit in the Wikipedia article on the PDO, with references to give you a head start.)

If you want to read more about teleconnections, the atmospheric bridge, or how ENSO affects the North Pacific, north tropical Atlantic, and Indian Oceans, in addition to the 1994 paper by Trenberth and Hurrell, there is a highly-cited 2002 paper by Alexander et al. Plus there's always Google Scholar for more recent work.

The PDO connection with temperature and precipitation in Australia

Now one of the reasons I'm writing this is to see how the PDO affects Australia. The paper by Mantua and Hare (2001) mentions Australia a few times, starting with the abstract:

A growing body of evidence highlights a strong tendency for PDO impacts in the Southern Hemisphere, with important surface climate anomalies over the mid-latitude South Pacific Ocean, Australia and South America.

Australia is big. Different parts of the country have different climates. Mantua and Hare suggest that the warm phase of the PDO coincides with anomalously warm-dry conditions in eastern Australia and anomalously wet periods in western Australia.

The cool phase of the PDO is associated with anomalously cool-wet conditions in eastern Australia. They also state that for the November to April period (the wet season in northern Australia), warm phases of the PDO coincide with anomalously warm temperatures in northwestern Australia.

Precipitation in eastern Australia

A lot of the papers relating to the PDO (or the related IPO) in regard to its impact on Australia, were about precipitation more than temperature. I came across a 2009 paper in GRL by Hamish McGowan and colleagues, which found that for more than 500 years, the PDO index (of Shen et al 2006) is highly correlated to inflows to the headwaters of the Murray-Darling system. The Darling is Australia's longest river and it flows into the Murray, which is Australia's largest river. Both are in eastern Australia. The headwaters of the Murray itself are not far north of where I live and the mouth of the Murray in South Australia. At the time of writing their paper, south eastern Australia was in arguably in it's worst drought on record (the Big Dry). Going by this paper, it was the worst drought in 529 years at least. McGowan et al wrote in the abstract:

Here we show that the >500 year PDO index of Shen et al. (2006) is highly correlated with inflows to the headwaters of Australia's longest river system, the Murray-Darling. We then use the PDO to reconstruct annual inflows to the Murray River back to A.D. 1474. These show penta-decadal and quasi-centennial cycles of low inflows and a possible 500 year cycle of much greater inflow variability. Superimposed on this is the likely influence of recent anthropogenic global warming. We believe this may explain the exceptionally low inflows of the past decade, the lowest of the previous 529 years.

They were talking about penta-decadal and quasi-centennial cycles. The second chart at the top of this article shows the period since the late 1990s as being a cool phase of the PDO, which would suggest that eastern Australia should have been wetter. (Cool phase = cooler wetter in the east; warm = warmer, dry in the east.) Well, there were arguably Australia's biggest floods on record - across the nation as a whole, in 2010-11, but you'll notice that the PDO index is positive for a short period from August 2002 to August 2007, dropping back into negative territory a few times in that period. Maybe it will take some time before it is decided exactly when the "cool" phase started and stopped. (Or maybe the PDO itself is changing with global warming.) In any case, this research would have been done six years ago, in 2008 or beforehand (it was submitted in December 2008). Therefore there would have been no clear signal of exactly when/if the PDO changed from the warm phase to the cool. The authors write (my emphasis):

Many historical hydrometeorological records document regime changes in the PDO in the mid-1940s (warm to cool) and mid-1970s (cool to warm). Interpretation of recent satellite images suggests that another change is in progress (warm to cool). Instrument records therefore only record three regime changes. This is inadequate to capture longer term variability in the PDO such as change in amplitude and/or periodicity. This information is required to improve understanding of the causes and magnitude of natural variability in regional hydroclimatology, particularly at inter-decadal time scales. In Australia, the most developed arid continent, this information is critical to the design of informed medium to long term water resource management policy.

I won't go into everything discussed in that paper. Figure 3 shows what they found in regard to the PDO and inflows to the Murray-Darling. I've added my own annotations to their Figure 3, to colour the modeled inflows blue so it stands out more, and shade the periods where the PDO index was negative. You can see that the inflows (blue line) are much higher when the PDO index is sharply negative and drop off in the warm phase of the PDO. It's not a perfect mirror image but it's close. Click the image to enlarge it.:

Figure 3: Inflows to the headwater catchments of the Murray River from A.D. 1474–1994 modelled using the Shen et al. [2006] PDO. 

Adapted from McGowan 2009.

Their research suggests that the in the warm phase of the PDO, the inflows dry up and in the cool phase the inflows increase, although it doesn't look quite as simple as that. However the scientists themselves didn't just eyeball the chart. they did statistical analysis. The conclusion of that paper is that the PDO is a "robust indicator of the hydroclimatology of the headwaters of the Murray River". They add that:

Interestingly, inflows during the past decade are far less than the lowest of the modeled inflow record for the previous 529 years. Whether this reflects a global warming signal on the hydroclimatology of southeastern Australia is unknown. However, it highlights the urgent need for future studies to reconstruct inflows to rivers of critical national importance such as the Murray River, preferably using a locally derived PDO index which is likely to better reflect local conditions.

Anyway, that's probably enough. Most of the science suggests that PDO phases do coincide with distinct patterns of temperature and precipitation in parts of Australia. Not the same patterns all over. Eastern Australia has the opposite pattern to what is observed in the west. And the north of the country is different again. Whether any of this is "cause and effect" or whether there is another underlying mechanism, the literature is not yet clear.

ENSO, the IPO and the PDO

There is a huge amount of literature on the PDO, much more than I can hope to read let alone cover in this blog article. However there was one other interesting thing that I came across, which might be of interest. It was about how ENSO events can have a different impact on Australia, depending on the phase of the Interdecadal Pacific Oscillation (IPO). The Interdecadal Pacific Oscillation shares some characteristics with the PDO, but refers to a larger region of the oceans. The IPO, like the PDO, has ENSO-like variability on decadal and longer time scales, but the area of ocean considered extends into the South Pacific and a bit of the Indian and Atlantic. In the North Pacific, the IPO and the PDO spatial patterns are similar.

A couple of papers that kept popping up were authored by Dr Scott Power in 1998 and 1999. Power et al (1999) has been cited a lot and is still being cited fifteen years down the track. It describes different ways people have tackled a quantitative representation of the IPO. The various IPO indices are, like the PDO, based on EOF analysis of SST data. For example, Power et al (1999) states that the index derived for a larger region is similar in its evolution over time to the PDO:

Additionally, Zhang et al. (1997) showed that the temporal evolution of EOFs of SST from larger regions had an inter-decadal component which is very similar to the inter-decadal component of the EOF of extra-tropical North Pacific SST.

What Power et al (1999) found was that there is a marked difference in how ENSO events affect Australia, depending on the phase of the IPO itself. Here is the abstract (my emphasis):

The success of an ENSO-based statistical rainfall prediction scheme and the influence of ENSO on Australia are shown to vary in association with a coherent, inter-decadal oscillation in surface temperature over the Pacific Ocean. When this Inter-decadal Pacific Oscillation (IPO) raises temperatures in the tropical Pacific Ocean, there is no robust relationship between year-to-year Australian climate variations and ENSO. When the IPO lowers temperature in the same region, on the other hand, year-to-year ENSO variability is closely associated with year-to-year variability in rainfall, surface temperature, river flow and the domestic wheat crop yield. The contrast in ENSO's influence between the two phases of the IPO is quite remarkable. This highlights exciting new avenues for obtaining improved climate predictions. 

I came across more than one paper that provide evidence that in the cool phase of the IPO/PDO, La Niña events bring more and heavier rainfall to eastern Australia. This is from a more recent paper by Andrew King, Lisa Alexander and Markus Donat (2013). They write:

Our results also indicate strong interdecadal modulation of the ENSO-extreme rainfall teleconnection related to the IPO. IPO-negative phases accompany significant relationships between the strength of La Niña events and extreme rainfall in eastern Australia. In El Niño seasons, there is some indication of a similar relationship during IPO-negative years; however, further investigation is required to establish the validity of this relationship. During IPO-positive phases, these relationships break down during both El Niño and La Niña seasons. Again, this interdecadal modulation is captured in the reanalysis to some degree.

[24] The IPO plays a large role in the frequency of major flood events in eastern Australia with those during the 1950s, 1970s, and 2010–2011 all occurring while the IPO has been in its negative (more La Niña–like) phase. Very intense rainfalls in eastern Australia that lead to severe flood events are often associated with strong La Niña seasons. Given that the IPO is likely to be in its negative phase, it is possible that there will be further strong La Niña events in the coming years. Therefore, there is an increased likelihood of major flood events occurring in this region while the IPO remains in its negative phase. It is worth noting, however, that the IPO has limited predictability, and its state at any present time is unknown due to the way it is calculated. This work has shown that the states of ENSO and, to a lesser degree, the IPO provide predictability of extreme rainfall events that lead to the most devastating floods.

That second last sentence brings me back to the difficulty of knowing the current phase of the IPO at any time, but probably more so when it might be changing phases. I'd say this applies to the related PDO as well. It's not enough to have the PDO index change sign. It has to be a sustained sign change. That means waiting and deciding some time after the phase change that it was indeed a phase change. If anyone has better information about this or if I'm behind the times, let me know.

Back to the original question

If you read the previous article, about the hottest summer on record, you'll have seen that it was hot all over the continent. The PDO/IPO have different effects on different parts of the country. If the PDO is in a cool phase (remember the index was negative all during that summer), then it should have coincided with cooler, wetter conditions in eastern Australia and warmer conditions in the west. When it's in the negative, La Nina's tend to be stronger. There was no ENSO event that summer.  The wet season was delayed and the whole country heated up and stayed extremely hot for around two weeks.

I won't categorically say there was no connection with anything because weather is connected. (Not that you'd take my word for it in any case. You'd rightly ask for the scientific evidence.) The other thing is that scientists don't talk about the PDO causing this or that, or the IPO causing this or that. They discuss them in terms of coinciding with particular weather events. The PDO and IPO aren't really mechanisms. They are an observed pattern of sea surface temperature and associated atmospheric conditions that have some consistency and tend to repeat and can be measured.

There's just one more thing. The PDO index has been positive since January this year. How long it stays that way - well your guess is at least as good as mine. But think back to Rum Runner's question. He asked if a (non-existent) "spike" caused the hottest summer on record. Given a warm PDO/IPO coincides with a warmer/dryer eastern Australia, why wouldn't the summer of 2013-14 have been hotter still? It wasn't.

Now some of you probably know more than you needed to know about the PDO, IPO and Australian weather. People more knowledgeable than me will probably read this and see holes in what I've managed to glean from the literature. Please let me know in the comments if that's the case, and I'll try to fix it.

To answer Rum Runner's question - finally. I found nothing to support Rum Runner's speculation and everything to suggest he was way off beam.

You can read more about what contributed to the hottest summer on record here. It was other factors that dominated, including greenhouse warming.

---

Alexander, Michael A., Ileana Bladé, Matthew Newman, John R. Lanzante, Ngar-Cheung Lau, and James D. Scott. "The atmospheric bridge: The influence of ENSO teleconnections on air-sea interaction over the global oceans." Journal of Climate 15, no. 16 (2002): 2205-2231. (link to paper)

King, Andrew D., Lisa V. Alexander, and Markus G. Donat. "Asymmetry in the response of eastern Australia extreme rainfall to low‐frequency Pacific variability." Geophysical Research Letters 40, no. 10 (2013): 2271-2277. doi:10.1002/grl.50427

Mantua, Nathan J., Steven R. Hare, Yuan Zhang, John M. Wallace, and Robert C. Francis. "A Pacific interdecadal climate oscillation with impacts on salmon production." Bulletin of the American Meteorological Society 78, no. 6 (1997): 1069-1079.

Mantua, Nathan J., and Steven R. Hare. "The Pacific decadal oscillation." Journal of Oceanography 58, no. 1 (2002): 35-44. (link to paper)

Newman, Matthew, Gilbert P. Compo, and Michael A. Alexander. "ENSO-forced variability of the Pacific decadal oscillation." Journal of Climate 16, no. 23 (2003): 3853-3857. (link to paper)

Power, Scott, Faina Tseitkin, Vikram Mehta, Beth Lavery, Simon Torok, and Neil Holbrook. "Decadal climate variability in Australia during the twentieth century." International Journal of Climatology 19, no. 2 (1999): 169-184. DOI: 10.1002/joc.1627  (link to paper)

Power, Scott, Faina Tseitkin, Simon Torok, Beth Lavery, Robert Dahni, and Bryant McAvaney. "Australian temperature, Australian rainfall and the Southern Oscillation, 1910-1992: coherent variability and recent changes." Australian Meteorological Magazine 47, no. 2 (1998): 85-101. (link to paper)

Power, S., T. Casey, C. Folland, A_ Colman, and V. Mehta. "Inter-decadal modulation of the impact of ENSO on Australia." Climate Dynamics 15, no. 5 (1999): 319-324. (link to paper)

Trenberth, Kevin E., and John T. Fasullo. "An apparent hiatus in global warming?." Earth's Future 1, no. 1 (2013): 19-32. doi:10.1002/2013EF000165

Trenberth, Kevin E., and James W. Hurrell. "Decadal atmosphere-ocean variations in the Pacific." Climate Dynamics 9, no. 6 (1994): 303-319. (link to paper)