San José State University
Thayer Watkins
Silicon Valley
& Tornado Alley

A Major Source of the
Near-Sixty Year Cycle in
Average Global Temperatures
is the Pacific (Multi)Decadal Oscillation

The record of average global temperatures from 1880 to 2008, as given below by the National Oceanic and Atmospheric Administration (NOAA) in terms of temperature anomalies (deviations from the long term average), shows a pattern of a cycle combined with a long term upward trend.

The profile of the cycle and the slope of the long term trend can be estimated by means of regression analysis. The details of the estimation procedures are given elsewhere.

The coefficient of determination (R²) for this regression is 88 percent. This means that 88 percent of the variation in average global temperature is explained by a long term trend of 0.5°C per century and a cycle which consists of an approximately thirty-year upswings and downswings.

The regression analysis provides the basis for forecasts and backcasts which given in Cycles and Trends in Average Global Temperature and Their Projection and confidence limits for those forecasts and backcasts are given in Confidence Limits. The forecasts and backcasts are based upon the duration of the upswings and downswings being thirty two years and thus the cycle period being sixty four years. The cycle has persisted for the 155 years for which the global temperature data is available.

The long term trend of 0.5°C per century quite likely is due to human activities, but from a variety of them rather than solely the production of carbon dioxide. The clearing of land for agriculture and city building goes back into the 19th century and is probably the major source. The increase in water vapor in atmosphere in arid and semi-arid areas due to irrigation, landscape watering, hydro projects and the burning of hydrocarbon fuels fuels is another. The increase in water in the atmosphere leads to a greater greenhouse effect and also the effect of more clouds on the climate. The other result of burning hydrocarbon fuels and of burning carbon (coal) is the increase in the carbon dioxide content of the air and its effect on the greenhouse effect. The long term trend is discernable but not catastrophic. The catastrophic predictions derive from an unjustified extrapolation of the short term cycle.

The immediate question is what natural phenomenon can account for the approximately thirty-year cycle of upswings and downswings. One plausible candidate for this cycle is the Pacific (Multi)Decadal Oscillation (PDO). The Pacific Decadal Oscillation is a climate index based upon patterns of variation in sea surface temperature of the North Pacific. It is available from the Northwest Fisheries Science Center, a division of the NOAA Fisheries Service. It has been tabulated for the period 1900 to 2009 and is maintained by Dr. Nathan Mantua. It is the principal component of sea surface temperatures in the northern Pacific Ocean. According to ther Northwest Fisheries Science Center,

The PDO index is correlated with many records of North Pacific and Pacific Northwest climate and ecology, including sea level pressure, winter land–surface temperature and precipitation, and stream flow. The index is also correlated with salmon landings from Alaska, Washington, Oregon, and California.

The PDO is highly correlated with sea surface temperature in the northern California Current (CC) area; thus we often speak of the PDO as being in one of two phases, a "warm phase" and a "cool phase," according to the sign of sea–surface temperature anomalies along the Pacific Coast of North America. These phases result from winter winds in the North Pacific: winter winds blowing chiefly from the southwest result in warmer conditions in the northern CC. Conversely, when winds blow chiefly from the north, upwelling occurs, leading to cooler conditions in the northern CC.

The term Pacific Decadal Oscillation was coined by Fisheries scientist Steven Hare in 1996 as a result of his work on the connections between Alaska salmon production cycles and Pacific climate. The Pacific (Multi)Decadal Oscillation is sort of the big brother of El Niño, the Southern (Pacific) Oscillation (ENSO). The ENSO occurs sporadically with a time interval between episodes of something on the order of ten years. When an El Niño occurs there is often a spike in the average global temperature and weather is affected around the world. The perturbations due to ENSO events typically last only six to eighteen months. The oscillation is between warming phases (El Niño) and cooling phases (La Niña). The perturbations due to ENSO events typically last only six to eighteen months.

The PDO involves a much bigger area of the ocean than the ENSO and the switch between warming and cooling phases takes a correspondingly longer time. Also because of the larger oceanic area involved the effect of the cycle of the PDO is also correspondingly larger spatially.

Here is the graph of the annual average for the PDO. The quantity plotted is a simple annual average of the monthly data on the PDO as tabulated by the Northwest Fisheries Science Center.

There is a good deal more noise for this data than for the average global temperature (AGT) data but the PDO index has a declining phase from 1900 to about 1919, as does the AGT. Then there is an upswing that lasts until the late 1930's, as does the AGT. From there there is a downswing lasting until about 1970, as there is for the AGT. From cerca 1970, for both the PDO index and the AGT there is an upswing. For the AGT the upswing lasts until about 2005 whereas for the PDO the downswing begins before 1990.

When a bent-line regression is fitted to the PDO index the result is as shown below. In the analysis the turning points were varied and selected to maximize the coefficient of determination.

The coefficient of determination (R²) for this equation is 0.325. This indicates that there is a good deal of the variation in the PDO index that is not explainable by a pattern over time.

As can be seen the slopes of the upswings are nearly equal and the slopes of the downswings are nearly equal. A regression line was fitted to the data in which the slopes of all the upswings are the same and all the slopes of the downswings are the same. As can be seen below the result looks nearly the same as the unconstrained regression shown above.

The coefficient of determination (R²) for this equation is 0.316, nearly the same as that for the unconstrained regression.

The big question is how closely can the cyclic profile of average global temperature (AGT) be related to that of the PDO index. The two profiles are shown together below.

Or, to more easily view the correspondences and non-correspondences:

There appears to a lag between the turning points of the PDO index and those of the AGT. This is as would be expected. However the magnitude of the lag varies which considerably weakens the use of the PDO as a predictor of the AGT. Although the correspondence is intriguing it is impossible to make it precise. The periods of the upswings and downswings for the PDO index is variable and is roughly 26 or 27 years. The corresponding period for the AGT is about 32 years. It is to be noted that a spectral analysis of average global temperatures comes up with a cycle period of about 52 years, corresponding to trends of 26 years in either direction.


Thus while the Pacific (Multi)Decadal Oscillation appears to be involved in the cycles of the average global temperature there has to be other factors also involved. The strong possibility is that the other oceanic oscillations such as the Atlantic Multidecadal Oscillation are involved as well as the Pacific one. It is already accepted that the ENSO accounts for significant perturbations in the AGT record.

The results of the above indicates that there is anthropogenic global warming but of a non-catastrophic 0.5° per century and probably only partially due to due to increased carbon dioxide in the atmosphere. The more perceptible changes in average global temperature are undoubtedly due to oceanic oscillations, one major one being the Pacific (Multi)Decadal Oscillation. More work needs to be done on the PDO index to remove the apparent noise in the data. More work also needs to be done on the matter of cycle lengths and time lags. And finally the other oceanic oscillations have to be examined as supplements to the effect of the PDO.

Over all the result is so reasonable and should have been obvious from the beginning. Average global temperature is driven by oceanic cycles and a secular trend of non-catastrophic proportions probably due to human activities. Once the tunnel vision focus on carbon dioxide is given up the truth emerges easily.

HOME PAGE OF applet-magic
HOME PAGE OF Thayer Watkins