The PDO is NOT a Simple Residual Like the AMO.
People understand the Atlantic Multidecadal Oscillation (AMO). It’s calculated very simply; subtract Global SST (sea surface temperature) anomalies from the North Atlantic SST anomalies. This simple process has been said to remove the global warming signal from the AMO. Many people believe the Pacific Decadal Oscillation (PDO) is calculated using the same basic equation, but it’s not. According to Nathan Mantua of JISAO, the details of how the PDO is calculated are found in this paper:
ENSO-like Interdecadal Variability: 1900–93
Calculating the PDO is a multistep process. It includes creating an SST anomaly time series for each 5 degree grid of the North Pacific (North of 20N), calculating the residual for each grid, and computing the EOFs ( empirical
orthogonal function) of these North Pacific residual SST anomaly fields. The PDO index is the leading PC (principal component) of that analysis. It’s far from a simple process.
The PDO has been found to be a function of ENSO. In “ENSO-Forced Variability of the Pacific Decadal Oscillation“, Newman et al state in the conclusions, “The PDO is dependent upon ENSO on all timescales.”
A few months ago, I discovered the instructions for retrieving Smith and Reynolds ERSST.v2 data (extended reconstructed SST) from the NOAA NOMADS system based on user selected dates and global coordinates:
http://www.ncdc.noaa.gov/oa/climate/research/sst/ERSST-ts.txt
One of the first data sets I downloaded was the time series of SST anomalies for the North Pacific, 20 to 65N, what I called the Mid-Latitude North Pacific SST Anomaly in the following graph. Note the 0.9 deg C drop then rebound in temperature from the late 19th to the mid-20th centuries. It’s tough to miss. It certainly appears to be related to Meridional Overturning Circulation (MOC), not ENSO.
http://i25.tinypic.com/2cyg07k.jpg
Using the same simple process employed to calculate the AMO, that is, subtracting the Global SST Anomaly from the Mid-Latitude North Pacific Anomaly, provides a data set that I’ve dubbed the North Pacific Residual.
http://i28.tinypic.com/jrwjk6.jpg
The North Pacific Residual bears no resemblance to the PDO. In fact, note that I had to scale the PDO to bring it back into line with the data set from which it is extracted. (The PDO data illustrated is from the ERSST.v2 data set, not the JISAO version. The curves of the two PDO data sets are similar, but the ERSST.v2 data extends further back in time.)
http://i27.tinypic.com/2n1sv49.jpg
When compared to the AMO, the two Northern Hemisphere SST oscillations complement one another from the 1920s to present. Prior to that, they were out of synch, offsetting their individual impacts on global temperature. It is no coincidence that Northern Hemisphere and global temperatures follow the rises and falls of these two residual anomalies.
http://i30.tinypic.com/11kv7r5.jpg
CLOSING
Past studies have estimated the contribution of the AMO to the rises and falls of Northern Hemisphere and Global temperatures over the 20th century. I would think that the North Pacific Residual would contribute similarly. Shouldn’t climatologists and climate change bloggers have another index of North Pacific temperature anomalies, one that could be used to determine the effect of the North Pacific SST oscillation on Northern Hemisphere and Global temperatures?
SOURCES
The links and graphs are from my series on Smith and Reynolds SST data:
http://bobtisdale.blogspot.com/2008/06/smith-and-reynolds-sst-posts.html
Sea Surface Temperature Data is Smith and Reynolds Extended Reconstructed SST (ERSST.v2) available through the NOAA National Operational Model Archive & Distribution System (NOMADS):
Jennifer Marohasy BSc PhD is a critical thinker with expertise in the scientific method.
First I didn’t know what to make of Bob’s NPR, but imo it makes better sense after reading the background via the blogspot link.
Gavin: Sorry, but I didn’t want to go into a lot of detail about the NPR. It would have made the post too long. Thanks for researching it and finding the following link. Hopefully it will help others.
http://bobtisdale.blogspot.com/2008/06/amo-versus-mid-latitude-north-pacific.html
Regards,
Bob
Very interesting; my first comment is based on these graphs;
http://i25.tinypic.com/2cyg07k.jpg
http://i27.tinypic.com/212s789.jpg
Both the Nth Atlantic SST anomaly and the Mid-Latitude Nth Pacific SST anomaly dovetail reasonably well with global SST anomalies. All show a marked increase at around about 1910; this is a good match with the takeoff of CO2 levels in the 20thC; arguably, therefore, that CO2 increase can be matched with a degassing from the oceans; SST took a downturn in the ’40’s; I have yet to look at the CO2 trend during the 40’s; but if CO2 continued to increase when there was a downturn in SST and presumably a decrease in degassing and even a net absorption then the increase in CO2 would have an anthropogenic component; but it still runs contrary to temperature! SST’s resumed their upwards trend in the 70’s as did global temp and presumably CO2 degassing; the CO2 movement when alligned with SST is a mask for natural effects on temp as indicated by the continued increase in CO2 during the 40s when temp declined.
The 1st maximum (maximum residuals) of the AMO peaked around 1850 – 1860. This coincides with a peak in NH atmospheric CO2 measured by the chemical method as Beck has identified. The standard error in the method at that time was probably about ±5-10%. Whether this coincided with the known expansion of coal mines in Europe (esp. Britain, Poland and Germany) and flowering of the steam age at that time is debatable and possibly tenuous.
However, the 2nd maximum of both the AMO and NPR peaks widely between 1930 and 1960. This coincides with a peak in NH (including Greenland) and more widespread (even a few Antarctic measurements) atmospheric CO2 level data measured by the chemical method as Beck has identified. The standard error in the method at that time was probably about ±1-2%.
The 1930 – 1960 period was characterized by rapid industrial expansion in Europe and US and the massive consumption of fossil fuels in Europe, North America and Japan associated with WWI.
The question arises as to why AMO and NPR temperatures were falling, minimal or just rising in the period 1970 – 1990?
The implication is that maxima of AMO and NPR residuals coincide very broadly with either peaks in oceanic degassing and/or peak periods of anthropogenic CO2 production. The lags that cohenite identifies are the details that require ‘teasing out’.
Clearly atmospheric and by inference surface oceanic CO2 levels are involved. In this context, it is required to attempt to understand the quantitative role of CO2 prior to the start of the post-1957 Keeling curve and, now that we have a good network of global CO2 measuring stations, regionally as well as temporally.
The fact that CO2 levels over the entire great Southern Ocean (SO) below 30 S have been continually lagging below the global mean (and NH levels) since at least 1982 and now increasingly so, such that they now lag almost as much below the global mean as the Ester Island Station in the middle of the highly oligotrophic and primary productive Southeast Pacific Gyre would seem to point towards at least one of the drivers of the NPR being derivative of cyanobacterial primary production.
As I have noted before, cyanobacterial primary production will profoundly affect not only regional CO2 masses but more importantly, a host of other weather-forcing, physicochemical effects.
We should also clearly not be wedded to the notion that the low frequency cycles of these residuals should be of constant duration.
They may well be driven by significant inputs of CO2, N, Fe and Si fertilization of these ‘lungs of the planet’ (i.e. the gross oceanic cyanobacterial biomass) by significant eruptive events.
Typo – WWII not WWI!
Looking at Mauna Loa;
1958 beginning of recordings: 314.61
1976 end of -ve PDO, general upturn in SST: 332.72
An increase over the 18 years of 18.11
End of 1994, 18 years into the +ve PDO and the general upturn in SST: 359.63; an increase of 26.91 over the equivalent 18 years into the different climate regime; a 48% increase over the rate of CO2 increase during the cooler phase. If we extend the comparison further upto the super El Nino of 1998;
1998 end concentration: 367.69; which is a staggering increase of 8.06 since 1994 and 34.97 since 1976; if we look at an 18 year margin going backward from 1998 to the end of 1980;
1980: 339.29
1998: 367.69
Change: 28.4
Compared with the change between 1958 and 1976 of 18.11, or 48%, the rate of change between 1980 and 1998 is 28.4 or 57%.
It is beyond doubt that CO2 is not a cause of climate change but a product of it; if climate is a game of table tennis, CO2 is the ball.
I’m still interested in how the various oceanic processes Bob is delineating intereact and whether they do in concert or whether, as is more likely there are regionalised differences; and I think, given the strongly contrasting regional weather differences produced by El Nino that we are gong to be talking about the myth of the global average temperature; see map on p. 4, FIG 7z-5 for representation of global El Nino weather;
http://www.physicalgeography.net/fundamentals/7z.html
Deserts are now sinks; Steve, the part about the desert blooming cyanobacteria should interest you;
http://www.ecostudies.org/press/Schlesinger_Science_13_June_2008.pdf
In 1980 CO2 at Mauna Loa (MLO) was about 0.20% above the global average, indicating net upwelling. Between 1989 and 1993 CO2 at MLO was only 0.09 – 0.10% above the global averages, indicating minimal upwelling. Between 1995 and 1998 CO2 at MLO was 0.21 – 0.28% above the global averages, indicating maximal upwelling/outgassing.
Between 1999 and 2007 CO2 at MLO ranged from 0.15 – 0.28% above the global averages indicating only a minor diminution in upwelling/outgassing following the 1998 ENSO SST maximum and a rapid return to maximal upwelling over the last decade.
However, in 1982 CO2 over the entire Southern Ocean (SO) below 30S lagged about 0.30% below the global averages and increased to a value 0.60% below the global averages by 2006, falling only slightly to 0.52% below the global average in 2008.
There is some evidence to indicate such lag values were much higher again (than 0.50%) prior to 1982 but unfortunately this involved only 1 – 2 monitoring stations.
From 1982 onwards as the number of SO stations rose form 3 to 10 this (rising) lag value trend line only fell slightly from 0.56 – 0.43% BELOW global averages from 1991 – 1993 indicating a small hiatus in increasingly efficient net CO2 removal/downwelling (and attendant physicochemical effects on climate) immediately following Pinatubo in 1991.
Regardless of Pinatubo, this indicates strong AND improving net downwelling/removal (relative to each annual global average CO2 level), especially from 1993 onwards, by the SO since at least 1982.
Excepting the period 1993 – 1998, this has coincided with a period of low or negligible SST rise over the Pacific as a whole and certainly over the SO.
The NPR also ranged from -0.23 – +0.10 C.
Over the entire period from 1994 – 2007 CO2 at Easter Island (EIC) just north of 30 S averaged a tight 0.64±0.06% below the global averages indicating near constant and maximal net removal or downwelling of CO2 i.e. the optimal balance between SST (= CO2 solubility) and primary productivity FOR the temperature range of the Southeast Pacific Gyre. The temperature range of the Southeast Pacific Gyre is significantly higher than that of the SO.
However, the entire (colder) SO below 30S has been closely approaching the EIC level of net CO2 removal and hence maximizing its primary productivity over the 8 year period 1999 – 2007.
Over the same period global temperatures have been flat to falling.
One must be cautious about assuming CO2 matched industrial economic growth prior to the 1950s. Because prior to this point much of rural North America, Europe and other developed nations still had no electricity and vehicle traffic was still relatively infrequent. Rural populations still comprised more than 50% of total population prior to this date as well. So the key emitters, power and auto, were somewhat retarded.
In fact, if Mauna Loa was 314ppm in 1958 and human CO2 emissions started impacting from 1750, from an equilibrium of 280ppm then the average CO2 increase over the 208 year interval was only 34/208 or 1.63ppm (0.58 of 1%) per decade.
According to Beck’s chemical methods data, most of that early increase was entirely within the normal range of variation. And in any event could not even be regarded as statistically significant until about 1940.
The key issue then hinges on the fact that if the pre-1910 oscillation was entirely natural why, then, would one conclude that a subsequent one of equal amplitude was not, also, entirely natural?
Hi Ian
In a nutshell, what I am NOT saying is that CO2 is driving climate strongly via the so-called Greenhouse Effect – as we now know there are a significant number of negative feedbacks which show the overall CO2 sensitivity is low – probably significantly less than 1.5 C (and maybe less than 0.5 C – but I’m not betting on it).
However, what I think we are really missing here is the response of the global oceanic CO2 system of: surface dissolution, CO2 fertilization of primary productivity stimulation and subsurface conveying’ to (variously) increased atmospheric CO2 either anthropogenic or for other reasons at various times (in terms of effect on weather).
Significant nutrient supply variation in atmospheric CO2 AND N, Fe and Si (the latter 3 via continental dust and volcanic fallout on the oceans), in my view, induces significant variations in regional, oceanic and global climate principally via the ‘spin-off’ effects of cyanobacterial blooming, i.e. the albedo, SST, relative humidity and lapse rate effects described previously.
Further to this. I forgot to point that it can be equally misleading (and fairly unproductive) to just look at absolute CO2 levels.
We know that it is (now) the anthropogenic input which is the the principal component of (positive) change in absolute CO2 levels. Hence I have been looking for possible indirect climatic effects of that principal comment. The oceans are clearly the places to look.
Just as Bob has described the derivation of the AMP and the (different) PDO above, I have for some time been looking at, in effect, the residuals between NOAA-computed monthly global means for atmospheric CO2 and monthly means over whole regions – in this case the SO especially, OR residuals between other well defined locations (showing near constant rates of change) and regions of the planet.
As Bob and Cohenite would agree, this is one of the best ways to tease out cause and effect.
Maybe I’ll just have to start providing some definitions and acronyms for my C-based residuals so they start looking as sexy as T-based residuals.
This:
http://i30.tinypic.com/11kv7r5.jpg
shows that the amplitudes of the NPR residual are decreasing with time while the amplitudes of the AMO are not.
How about a SPR?
Interesting, I had intended to write this up earlier but never got to it. The Mauna loa data set also shows a shortening of the interval between maximum and minimum annual values over the past four decades of 2 to 3 weeks. Ralph Keeling hadn’t noticed it when I spoke to him about it so he detrended the data and it was still there.
This has contributed to some of the build up in CO2 and would be consistent with a shorter growing season that has absorbed less carbon. The only problem is that the temperature data has not reconciled.
One plausible explanation could be that improved annual crop varieties have shortened the growing season and have reduced the CO2 absorption required to produce a crop.
Steve Short: Hot off the press, which means I haven’t spent a lot of time looking at the results.
The first graph is a comparison of Mid-Latitude South Pacific and Global SST Anomalies.
http://i38.tinypic.com/6rrhna.jpg
The second graph is the requested South Pacific Residual.
http://i36.tinypic.com/166azph.jpg
Bob’s comparison of SO and Global anomalies reinforces my conclusion: partial pressure of CO2 is the 800 lb. gorilla.
Biogenic pulses simply don’t fit the seasonal signal; they fall off while the SS temperature is still peaking.
Not talking about biogenic pulses of CO2 at all. No way. No whichaway (trivial effect).
A complete and utter misunderstanding by Gary!
FACTS SHEET:
Marine cyanobacteria absorb CO2 and bicarbonate from seawater (subject to the constraints of SST) and emit O2. In so doing they:
(1) In some cases secrete biogenic calcite i.e. the coccolithophores. Blooms of this give the sea surface a much enhanced albedo.
(2) All cyanobacterial blooms cause the sea to be covered with monoloayers or thicker of lipids, sterols etc due to predation by zooplakton and cell lysis by cyanobacteriophages (viruses if you will). This reduces the rate of evaporation relative to a sea surface without monolayers etc. For example, for years long chain alcohols have been used to lower evaporation of water surface in reservoirs etc. There are textbooks on this.
(2) Cyanobacteria emit a compound which decomposes into dimethylsulfide (DMS) which passes into the air. This nucleates low level cloud, not only releasing latent heat but further increasing low level albedo and lowering SST (therby increasing CO2 solubility and so on and so forth).
Marine cyanobacterial blooms have profund effects on SST, surface and low altitude albedo and cloud mass, relative humidity and lapse rate. They can be thousands of km2 in size. They are easily photographed and studied from satellites.
Go figure……
Bob – not sure where this fits but there appears to be quasi-decadal components as well.
Journal of Climate
Volume 18, Issue 1 (January 2005)
Rainfall Variability at Decadal and Longer Time Scales: Signal or Noise?
http://www.bom.gov.au/bmrc/clfor/cfstaff/sbp/journal_articles/holger_jclim_2005.pdf
I’m now confused where PDO, IPO and quasi-decadal fits together. Or echoes of the same process ?
Luke: Whenever I see quasi-decadal in the title or abstract of a paper, I presume the author is avoiding the use of TSI or solar influence, since the solar cycles are quasi-decadal. Many times I’m right. Why they shy away from solar in the title is beyond me. Unfortunately, the term quasi-decadal oscillation is applied to any number of variables, like precipitation in your linked paper, so I can’t isolate what you’re referring to.
As for the IPO: According to Chris Folland of Hadley Centre, “The Interdecadal Pacific Oscillation (IPO) is (almost) the Pacific-wide manifestation of the Pacific Decadal Oscillation of Mantua et al (1997), with as much variance in the Southern Hemisphere Pacific down to at least 55oS as in the Northern Hemisphere. The IPO is a multidecadal sea surface temperature (SST) pattern quite like that of ENSO, but differing in several ways.” He concludes that opening paragraph in the following with, “The physical nature of the IPO is under investigation; it is still not clear, despite the above studies, to what extent the IPO is really independent of ENSO red noise and especially of SST variations near a decadal time scale.”
ftp://www.iges.org/pub/kinter/c20c/IPO.doc
I’ve covered the PDO-ENSO relationship in a number of posts at my blogspot. At the end of the second linked post is a quote from a paper that states, “The PDO is dependent on ENSO on all timescales.” It may also appear in the first link.
http://bobtisdale.blogspot.com/2008/06/common-misunderstanding-about-pdo.html
http://bobtisdale.blogspot.com/2008/06/chicken-or-egg-pdo-or-enso.html
The three PDO data sets (JISAO, ERSST.v2, ERSST.v3) are discussed here:
http://bobtisdale.blogspot.com/2008/06/three-pdo-data-sets.html
If we assume the PDO is dependent on ENSO, is there enough of a correlation between the PDO and the IPO to assume that the IPO is also dependent on ENSO?
The following three graphs illustrate the correlation between the PDO and IPO. The first is raw data. Note: There aren’t any 5 and 6 deg C basin-wide oscillations in the Pacific. The amplification of the signal is a result of the statistical process they use to extract the IPO from Pacific SST data. (The same note applies to the PDO, as well.)
http://i33.tinypic.com/vpg4l2.jpg
In the second, I’ve smoothed the data with a 37-month filter.
http://i33.tinypic.com/2e6c8eq.jpg
In the third, I’ve scaled the IPO to highlight the correlation with the PDO. There are some minor variances, just as there are differences between the three PDO data sets.
http://i38.tinypic.com/nz69y.jpg
I hope that helps. I haven’t yet gotten around to graphically comparing the PDO and NINO3.4. Someday.