SOLAR WIND HAS TWICE
THE GLOBAL WARMING EFFECT
OF EL NIÑO
THE CONSENSUS ON CLIMATE
MISTAKENLY ATTRIBUTES SOLAR WIND WARMING
TO MANMADE CARBON DIOXIDE
------------------------------------------------------------
SOLAR WIND, EL NIÑO/SOUTHERN OSCILLATION,
& GLOBAL TEMPERATURE:
EVENTS & CORRELATIONS
by Jeffrey A. Glassman, PhD
Revised 7/10/07
-
ABSTRACT
Classical and advanced signal analysis techniques applied to the climate data of global temperature, solar wind, and El Niño/Southern Oscillation (ENSO) reveal new events and correlations in graphical form. The results include:
1. Major state changes appear in the global temperature record around 1934.4 and 1979.5.
2. A major state change occurred in the solar wind index around 1937 to 1939, and a secondary state change occurred in the 1970s.
3. Major state changes occurred in the Southern Oscillation Index beginning about 1919.3 and 1979.4. A large state change occurred during the brief period of 1940.2 to 1942.0.
4. The state changes are real in the records, but may be due either to data acquisition artifacts or to real physical phenomena.
5. The Southern Oscillation Index has a weak cyclic behavior with a period of 3.38 years.
6. Global temperature lags the Southern Oscillation Index by about 5 months.
7. The global temperature record appears to suffer from excessive processing.
8. High correlations found by other investigators may be the result of prior data smoothing.
9. The low level of correlation between temperature and other parameters may be due to excessive noise, equivalently due to low signal to noise ratio. More importantly, it may be due to the closed loop gain of a mechanism in the climate, unknown to the Consensus on Climate, that regulates global surface temperature.
10. Global temperature is weakly correlated with ENSO. The SOI could account for 4.6% of the measured variation in global temperature.
11. Global temperature and the solar wind index are correlated. The solar wind index may contribute as much as 8.9% of the processed global temperature variations.
12. Global temperature lags the solar wind index by about two to five years.
13. ENSO and the Southern Oscillation affect the global surface temperature. The reverse, that temperature might affect either, is not true.
ENSO may, as the Consensus says, devastate, but it has only half the capacity of the solar wind to warm the planet. By omitting the solar wind, the Consensus underestimates the natural causes of global warming, simultaneously overestimating the anthropogenic sources by the equivalent of two ENSOs, assigning the error to carbon dioxide emissions.
INTRODUCTION
Global surface temperature is the essence of global climate, and by definition the only sensible measure of Global Warming. Global temperature is influenced in part by the solar wind, according to a chain of connections admitted by the IPCC in its Third Assessment Report (TAR). The Consensus on Climate, which finds its voice in the IPCC reports, recognizes (1) a dependence of global temperature on clouds, (2) a positive correlation between clouds and galactic cosmic rays, and (3) the physics of cosmic ray flux modulation by the solar wind. Nonetheless, the Consensus refuses to adopt this three part model into its climate models for lack of evidence. Climate Change 2001, ¶6.11.2.2, pp. 384-385. More importantly, the TAR never uses the record of solar wind measurements, a record longer than the record of temperature from thermometers.
According to the Consensus, Earth’s surface temperature is strongly affected by the El Niño/Southern Oscillation (ENSO), so named because of the strong indication of El Niño events in the Southern Oscillation Index (SOI), a continuous variable. The Consensus is convinced that ENSO causes human suffering, and that on some time scales it is the cause of the strongest natural fluctuations in the climate. Climate Change 2001, Box 4, p. 52. However, the Global Circulation Models, formerly known as Global Climate Models but more aptly named Global Catastrophe Models, (GCMs), have yet to demonstrate sufficient accuracy in replicating ENSO to even answer the pointed question posed by the Consensus: whether anthropogenic greenhouse gases cause a positive feedback by precipitating warming El Niño events. Climate Change 2001, p. 151; ¶7.6.5, pp. 453-455. The SOI sample record is substantial, but the TAR uses it merely to remark on its variability and to label the El Niño events for qualitative discussion.
These issues are well-suited to ordinary signal analysis of the three time series because the method can recognize signals at the threshold of detectability. It can, among other things, locate events in the record, and measure the correlation and lead or lag between temperature and the other parameters, all key to model building. That analysis, not reported in the IPCC and apparently never addressed by the Consensus, is initiated here for what it reveals, and as background for upcoming climate studies.
DATA SOURCES
Records are available on-line of monthly measurements of the Temperature anomaly, the Southern Oscillation Index, and Solar Wind Index.
Temperature anomaly
The Temperature anomaly is the subject of several reductions by the IPCC (e.g., Fourth Assessment Report, Summary for Policy Makers, p. 14, Fig. 5; Climate Change 2001, Fig. 2.7c, p. 114. See also http://svs.gsfc.nasa.gov/vis/a000000/a001000/a001008/a001008_pre.jpg), including the infamous reduction for the Northern Hemisphere (Climate Change 2001, Fig. 2.20, p. 134) known derisively outside the IPCC reports as the Hockey Stick reconstruction. The phrase solar wind appears just once in the TAR, and that is in the title of a reference about cloud creation. However, the TAR uses not a single datum from the solar wind database. The Fourth Assessment Report also mentions the solar wind, but again just once, and that is to say that the effects of solar wind fluctuations are ambiguous. Id., ¶2.7.1.3, p. 192.
Global temperature data are available from January, 1880. http://data.giss.nasa.gov/gistemp/tabledata/GLB.Ts.txt. (see also ftp://ftp.ncdc.noaa.gov/pub/data/anomalies/monthly.land_and_ocean.90S.90N.df_1901-2000mean.dat.) The global temperature as reported by the IPCC extracted from Figure 2.7(c) is shown in Figure 1. The report states without elaboration that the data are an “optimum average”.
Figure 2.7: …[C]ombined land-surface air and sea surface temperatures (ºC), 1861 to 2000, relative to 1961 to 1990, for … (c) Globe. …[S]hown are the unsmoothed optimum averages – red bars… . Climate Change 2001, p. 114. [Click figures to enlarge.]
Figure 1
IPCC Data Distribution Center
Although not stated in the Third or Fourth Assessment Report, the IPCC maintains a separate Data Distribution Center. It includes information on the Temperature anomaly but not on the solar wind.
The Temperature anomaly is an IPCC-calculated global figure based on a grid of temperature differences spanning the globe and recorded by the Climate Research Unit (CRU), School of Environmental Sciences, University of East Anglia (UEA). The CRU provides additional information on-line about the Temperature anomaly and its treatment by the IPCC, however its key records exceed the number of rows permitted in Microsoft Excel.
The CRU data are the differences between a calculated mean temperature for each grid section and its average from 1961 to 1990. The IPCC interpolates for missing data points, and its final temperature anomaly is a weighted average over the globe and over an unspecified time interval. In the processing, adjustments are included for station altitude, and to convert sea surface temperature to local atmospheric temperature.
In spite of these complications, the National Oceanic and Atmospheric Administration (NOAA) says,
By adding the long-term monthly mean temperature for the Earth to each anomaly value, one can create a time series that approximates the temperature of the Earth and how it has been changing through time. National Climatic Data Center, Global Surface Temperature Anomalies, 2/6/06. http://www.ncdc.noaa.gov/oa/climate/research/anomalies/anomalies.html
According to the Consensus, the long-term mean surface temperature is 14ºC. Climate Change 2001, p. 89.
Southern Oscillation Index
The available Southern Oscillation Index (SOI) data, designated “aa”, begin January, 1876. ftp://ftp.bom.gov.au/anon/home/ncc/www/sco/soi/soiplaintext.html. {Begin rev. 7/10/07} The Third Assessment Report charts them in Figure 7.9 after subtracting the average for the first 100 years. A copy is Figure 2, below. The offset by the average has little effect; it is trivial because by design, the Index is nominally zero. {End rev. 7/10/07}
Figure 7.9: Darwin Southern Oscillation Index (SOI) represented as monthly surface pressure anomalies in hPa. Data cover the period from January 1882 to December 1998. Base period climatology computed from the period January 1882 to December 1981. The step function fit is illustrative only, to highlight a possible shift around 1976 to 1977. Climate Change 2001, p. 455.
Figure 2
Solar Wind Index
{Begin rev. 7/10/07} Solar Wind Index (“aa”) data start January, 1868. ftp.ngdc.noaa.gov/STP/SOLAR_DATA/RELATED_INDICES/AA_INDEX/ [rev. 7/10/07].{End rev. 7/10/07} The Third Assessment Report refers to Tinsley, B.A., 1996, Correlations of atmospheric dynamics with solar wind-induced changes of air-earth current density into cloud top, J. Geophys. Res., 101, 29701-29714 ($9) in its discussion of a correlation between galactic cosmic rays and clouds. Climate Change 2001, ¶6.11.2.2, pp. 384-385. The Consensus summaries a mechanism proposed by Tinsley to link cosmic rays and clouds, but without mentioning the solar wind at that point or anywhere else in the Report.
SIGNAL ANALYSIS
Referring to these three records as raw data, the objective now is to mine them for correlations and abrupt changes in statistics. These changes delineate changes in state of the data. They might signify data acquisition artifacts, or actual climate events or patterns. The techniques rely on numerical analysis, witnessed by graphs, and employ ordinary techniques used in engineering signal analysis.
Data Reduction.
Data reduction may employ advanced techniques, but if any part of those techniques is kept hidden, the method is secret and beyond mere esotericism. An example is the application of Principle Component analysis where the principle components are not fully disclosed. Principle Component analysis is esoteric, not wrong. However discarding components or keeping the selection method secret violates scientific principles, and is sufficient to invalidate the results.
The objective of data reduction is to reveal features in the underlying physical processes, uncovering any artifacts caused by faulty data acquisition along the way. It is to unveil real events and patterns with which to perfect models, and from those models to make predictions.
Data reduction used to support a subjective conclusion is for the movies. Data reduction to make the data better looking, or trendy, or to create correlations in support of conjectures is for the advertising business.
In science, where a method fails to produce a useful result for a model, the application fails. The method is not discarded merely for being instantaneously inapplicable.
On the other hand, where a method does produce a useful result, it is successful even if the method may be deemed heuristic for lack of theoretical exposition. Here applied is a technique that reveals profound changes in the statistics of the data, changes which could be artifacts of the record preparation, but which are reproducible and which could not have been produced by the data reduction technique.
Temperature and the Solar Wind.
The signal analysis begins with a co-plot of the Solar Wind Index, aa, and the so-called global “Temperature anomaly”. Plotting both records along the same time scale creates a parametric view of the data, with time the parameter. This is Figure 3.
In this chart, the solar wind rises slowly over its history, marked by decadal clumping. Temperature has a lazy-w shape. The solar wind starts relatively well organized, and later diffuses. The temperature history starts disorganized and later coalesces. Stated another way, the variability of the Solar Wind Index increases over its record, and the variability of the Temperature anomaly decreases. No rationale is known for this organizational behavior, nor for its complementary appearance.
Solar Wind Trend
Filtering helps measure the rise in the solar wind record. Figure 4 shows four overlaid, low pass filter reductions, with time constants of one, two, three and five decades. At the shortest time constant, the solar wind reveals a step-ramp-step shape, with breaks around 1920 and 1960.
The long term, steady rise in both the Solar Wind Index and the Temperature anomaly suggest a correlation between the two traces.
Temperature & Solar Wind Events
Independent, best fit, three-step fits to Temperature and the Solar Wind, beginning in common from 1880, reveal no particular relationship. Figure 5.
In the legend, the numbers in parenthesis are one standard deviation error between the best fit and the data over the full record, expressed in the units of the ordinate. The digit preceding the standard deviation is the number of segments in the mathematical model. Best fit modeling characterizes the data objectively. The dates and step magnitudes are determined mathematically, once the investigator selects the number of segments. The fits are arithmetically best in that they minimize the sum squared error for the model type, and as is evident, not necessarily the best shape subjectively. For example, Temperature appears as though it would be better modeled with a step-step-ramp shape.
Cumulative Signal Analysis.
Numerical analysis of cumulative data reveals characteristics of the physics hidden by noise in the raw data. An illustration applied to the Temperature anomaly and the Solar Wind Index is in Figure 6.
The segmented, best fit curves comprise straight lines between the end points and the labeled graph markers on or near the cumulative curve. The best fit endpoints in this analysis lie on the cumulative data, and were not subject to optimization. The slope of each segment is the height of step in the corresponding best fit to the raw data. The step height approximates the mean value of the raw data in the segment interval span. The curves have the smallest Root Mean Square (RMS) error between the raw data, not the cumulative data, and the stairstep fits, as finally determined by the Excel 2004 solver routine, for the operator-selected number of segments.
The Temperature record contains a state change beginning at about 1934.4 and 1979.5, with second order state changes at about 1918.2, 1946.5, and 1997.4. Any of these may be data acquisition artifacts, as when instrumentation technology, standards, or the set of measuring stations changed. For example, new standards for thermometer measurements became effective circa 1920, and satellite measurements were added in the ’70s. No discrete, climate event appears in these data from 1880 to the present which suggests an anthropogenic source or event.
Masking Effects of Scaling
Cumulative analysis removes noise from the measurements, much as do its cousin, the running-average, pass band filters, and other classes of filters. Noise is the variation in the data from any source other than the parameter under examination. The cumulative technique has the advantage of lack of subjectivity, the bane of science. The analysis is immediately applicable to the Temperature anomaly because the original investigators reduced that record to approximately zero mean. If a constant 14ºC had not been subtracted from the Temperature, its cumulative graph would be the ramp in Figure 7.
The plot in Figure 7 comprises individual, unconnected dots for each data point. The state change information is still in the data, but it is no longer resolvable to the unaided eye. The state structure might be visible with a magnifying glass on a graph rendered at upwards of 600 dots per inch. Greater resolution would be required had the temperature been recorded in total degrees Kelvin. This demonstrates the masking effects of merely an unfortunate choice of units. The technique of offsetting a bias or mean is common in signal analysis, and, in some circumstances, necessary. It is the creation of an arbitrary zero point, as commonplace as using the Fahrenheit and centigrade scales. It causes no loss of information so long as the offset is stated. That is, the raw record can be restored exactly by calculating sample by sample differences in a cumulative history, and restoring the offset.
The Solar Wind Index retains a large average value. So in the cumulative, its signal changes ride masked atop a strong ramp. Figure 6. Another operation is necessary to prepare this raw record for cumulative analysis.
Raw Solar Wind vs. Cumulative Temperature anomaly
A comparison of Temperature with its seven-segment fit to the raw Solar Wind Index with its three-step fit is shown in Figure 8. The coincidental break around 1934 is suggestive of an actual climatic event. This coincidence is a form of correlation, and like correlation does not prove a cause and effect, but suggests where a causative event might lie. On the other hand, its absence is convincing evidence against a cause and effect.
The step models fitted to the Solar Wind Index and Temperature in Figures 5 and 8 could be fanciful. The fitting of a step function yields steps even if the record is a pure ramp or other kind of smooth curve. The fact that the steps are unequal may be due to noise, or to an acceleration in the record. No such choice occurs with the cumulative technique, next applied to the Solar Wind Index by offsetting the raw data by the full record mean.
Solar Wind Events
The Solar Wind Index contains changes, including one profound change, previously unknown at least to the Consensus on climate. A primary state change occurred around 1937-1939, and a secondary state change around 1980. These are revealed in Figures 9 through 12 with progressively increasing numbers of segments, 3, 4, 7, and 9, respectively. Since the RMS error to the raw data is already on these charts, the standard deviations referenced to the cumulative data are the RMS error between the segmented ramps and the cumulative curves.
Cumulative Solar Wind vs. Cumulative Temperature anomaly
The next three charts show a cumulative analysis of the offset Solar Wind Index parametrically with that of the Temperature anomaly. The best fits are independent, and for 3, 4, and 7 segments, with the RMS errors referenced to the raw data. Increasing the number of segments has a relatively minor effect on the accuracy (mathematically the accuracy cannot decrease), and quickly produces diminishing returns.
A reasonable conclusion is that the evident Solar Wind events did not precipitate the evident global temperature events.
Autocorrelation Functions of Solar Wind & Temperature anomaly
Next in Figure 16 are the autocorrelation functions for the Solar Wind Index and Global Temperature.
The well-known 11 year solar cycle appears in the Solar Wind Index, but not in the Temperature record. (Compare with “The surface temperature response to the 11-year cycle is found to be small (citations).” Climate Change 2001, p. 708.) The breadth and shape of the Temperature autocorrelation function suggests weighted filtering over a window of about 20 years. Such filtering can distort signal analysis, requiring special considerations. All the correlation function calculations here employ the tape-loop algorithm.
While global temperature is correlated with the solar wind, the cyclic behavior of the solar wind is not evident in the temperature. One reason might be heavy temporal smoothing of the Temperature anomaly record.
Cross Correlation of Solar Wind & Temperature anomaly
Next is the cross correlation function between Temperature and Solar Wind Index. Figure 17.
This curve is not sharp enough to yield a conclusive lead/lag relationship, perhaps again because of temperature processing. Physically, global temperature should not lead the Solar Wind Index. The data suggest that the Temperature lags the Solar Wind Index by about two to five years.
Temperature anomaly vs. Solar Wind Scatter Diagram
The most significant relationship between Temperature and the Solar Wind appears in the cross-plot scatter diagram with the linear fits. These are in Figure 18 with zero time offset between the traces.
The pair of lines are the result of simply changing which variable is assigned as the dependent variable, and it illustrates the centroid of the scatter, and the correlation between the variables. The legend includes the slope of each straight line fit, where the product of the slopes is the coefficient of determination, r2, the square of the correlation coefficient, r. The smaller the acute angle between the lines, the greater the correlation, and the lines cross at the means of the two variables. (If the lines are perpendicular, the traces are called orthogonal. Only if the lines are perpendicular and cross at (0,0), are they strictly called uncorrelated.) The line T(aa) is bold to emphasize the feasibility of temperature depending, in part, on the Solar Wind, while the reverse, aa(T), is not possible.
In signal analysis terms, the coefficient of determination is a measure of the mutual power between the normalized variables. In statistical terms, the coefficient represents the variability in a given observation due to an explanatory variable. The solar wind could account for 8.9% of global temperature in a linear regression.
On the Low Coefficient of Determination, r2.
Landscheidt
Another investigator came to quite different conclusions about the relationship between the solar wind and global temperature. This investigator, obviously outside the Consensus (the Consensus did not analyze the solar wind), wrote:
Abstract. Near-Earth variations in the solar wind, measured by the geomagnetic aa index since 1868, are closely correlated with global temperature (r = 0.96; P < 10-7). Geomagnetic activity leads temperature by 4 to 8 years. Allowing for this temperature lag, an outstanding aa peak around 1990 could explain the high global temperature in 1998. After 1990 the geomagnetic aa data show a steep decline comparable to the decrease between 1955 and 1967, followed by falling temperatures from 1961 through 1973 in spite of growing anthropogenic CO2 emissions. This points to decreasing global temperature during the next 10 years. Landscheidt, T., Solar wind near earth: indicator of variations in global temperature, Proceedings of 1st Solar & Space Weather Euro Conference, 9/29/00, p. 1. http://www.mitosyfraudes.org/Calen/SolarWind.html.
Smoothed yearly aa index [ordinate]. Smoothed yearly Northern Hemisphere temperature anomalies [abscissa]. Figure 1. Scatter plot of yearly means of the geomagnetic aa index and Northern Hemisphere land air and sea surface temperature anomalies 1868 -1998. The aa data are shifted to offset a 6-year lag of temperature. The slope of the regression line and the aggregation of the slightly smoothed data around the straight line fit indicate a close correlation (r = 0.75) which is highly significant ( P < 10-7). Id., p. 2.
A scatter plot of the raw yearly data shows a promising positive correlation between aa and temperature (r = 0.48). Id., p. 3. The data were subjected to three-point smoothing. The least-squares linear fit line indicates a strong correlation ( r = 0.75) which explains 56% of the variance. This correlation is highly significant. After the shift, the record of yearly means is reduced from 131 to 125 data points as the data lost by shifting cannot be replaced. [¶] Three-point smoothing, applied once, reduces the number of independent data to 42. Id., p. 3.
Regardless of statistical rationale, the yearly averaging at the outset produces additional correlation out of whole cloth, exacerbated by the additional three-point smoothing. Amplified correlation attained by smoothing is a mathematical recreation, but is less likely to lead to a physical model with predictive power than working from raw data.
Köhnlein
Still another investigator on a related subject followed a similar course, writing
A linear fit of daily solar wind parameters todaily sunspot numbers does not lead to very useful results. The residual scatter is simply too high. If, however, the plasma- and sunspot data are smoothed beforehand by an averaging procedure, then structures show up which are persistent in each of the last two sunspot cycles. This averaging procedure can either be accomplished by running means over, say, 3 years or by the lower terms of a Fourier analysis. Both versions lead to the same results. Köhnlein, W. Cross-correlation of solar wind parameters with sunspots (‘long-term variations’) at 1 AU during cycles 21 and 22, Astrophysics and Space Science, v. 245: 81-88. 11/13/96. p. 83. www.springerlink.com/index/N675G9N846K16722.pdf
Indeed, it is the so-called scatter (noise) that causes decorrelation. Smoothing removes noise, and thereby increases correlation mathematically, but it is no part of the underlying physics. Alternatively stated, smoothing is a regression process that progressively reduces data toward a curve. Two data streams similarly filtered appear correlated. Two straight lines are perfectly correlated, even though the sources were independent.
A strong r-squared, that is, one near unity, is excellent support for a cause and effect model between the dependent variable and the predictor, or explanatory, variable. Conversely, a small value may be due to noise alone, indicating that the independent variable has little predictive power, if any. A weak value might be a masking by noise. It can be an indication of a poor signal-to-noise ratio, a common challenge in communications, astronomy, and climatology. The noise can be an additive interference, contributions from multiple sources, or limitations in the instrumentation and data reduction. An example of a noisy source is the algorithm to reckon a global temperature from widely scattered measurements, over land and sea, and complicated by weather phenomena, altitude, seasonal and diurnal effects, measurement complexities, estimations and arbitrary weightings.
Open Loop Inferences from Closed Loop Measurements
Yet another cause of poor correlation, one not recognized by the Consensus, is the phenomenon of closed loop behavior. During the time that a separate phenomenon regulates the dependent variable, the closed loop response to a predictor variable can be sharply attenuated. The response is reduced by the closed loop gain. Earth’s global temperature is likely just such a variable. The mean temperature around 14ºC is not just some accidental, instantaneous value as the climate slowly wanders between the temperature of Venus and that of Neptune.
Current Global Catastrophe Models force greenhouse gas accumulation to drive the climate to end life-as-we-know-it – except for the dominant greenhouse gas, water vapor. And except for clouds, which have yet to be modeled successfully. And except for albedo, which the Consensus treats as a constant known to one significant figure.
Why Is Earth’s Climate so Stable?
The notion of a Delicate Blue Planet is romantic and juvenile. The idea of a tipping point is a manifestation of paranoia if believed, or mischief if not. Round boulders do not perch on the sides of hills, and cones are not found standing on their tips. Minute disturbances quickly produce a new, quasi-stable state. Global Climate Models are not designed with any stable state. They are as chaotic as the weather. They tip over in a random direction due to one disturbance or another, hence Global Catastrophe Models. At the start of a run, the GCM stands on its tip. The Consensus computes the average cause and effect from runs with a number of these unstable Global Conical Models. See especially Climate Change 2001, Chapter 8, Model Evaluation, pp. 471-512. Catastrophe is certain. The only questions by this paradigm is how fast, how far, which direction, and from which causes.
A rational approach to Earth’s climate begins with the observation that it is in a conditionally stable state, and the scientific challenge is first to model the variables that regulate that state. The global temperature does not move much with changes in the solar wind or ENSO. In part, that is because the temperature is regulated, and measurements can only be made in closed loop.
The next task for climatology is to determine the margin for closed loop control. The atmosphere has lots of room for more water vapor, more clouds, and a greater albedo, or the reverse, thanks to the immense reservoir of liquid water and its heat capacity.
Still, comparing cumulative temperature to the raw solar wind index as done for Figure 8 but with four segments for each trace instead of three, results in a weak and somewhat subjective support for a model for temperature dependence on the solar wind. It is shown in Figure 19.
The correlation in Figure 18 supports a dependence of global temperature on solar wind, and is an important conclusion of this signal analysis. Figure 19 suggests the dependence does not strongly arise from a state change in the solar wind. It shows the effects of constraining the breaks in the temperature model to coincide with those in the solar wind. The constraint increases the standard deviation of the error by more than 50% (4.8 to 7.5). Experimentation by lagging the temperature breaks might produce a better fit.
Regardless, the correlation need not be strong because the solar wind is not a direct source of warming. Instead the theory of cloud formation makes the solar wind a gate to admit greater solar radiation through reduction in cloud cover and hence reduction in Earth’s albedo.
Temperature and ENSO.
Third for signal analysis is the Southern Oscillation Index (SOI), a strong, climate measure over the South Pacific, well‑established as an indicator of Los Niños (the harmful El Niño and his amiable sister, La Niña) events. Sir Gilbert Walker, the discoverer, apparently was responsible for setting the Index to be neutral at zero, and negative toward an El Niño event. A sustained, strong excursion in negative or positive territory of the SOI invariably indicates such events, so climatologists give the phenomenon the name ENSO for El Niño/Southern Oscillation. See Figure 2, above.
ENSO Events
Raw SOI data are featureless compared to its colorful behavior in the cumulative. The comparison is shown in Figures 20 and 21 along with three- and five-piece linear fits, respectively. As before, the standard deviations given are for the respective fits.
The cumulative curves are surprising. Nothing in the cumulative signal analysis could have caused these statistical changes. The analysis proceeds mechanically without regard to the time coordinate.
When the Consensus analyzed the variability of ENSO, it divided the instrument record into four main epochs: the first 40 to 50 years, the period of 1920 to 1960, an intervening period, and the last 40 to 50 years, with special remarks for the period of low SOI from 1990 to 1995. Climate Change 2001, p. 151. The data are more precisely characterized by three first order epochs, separated at 1916 and 1977, with a brief, strong retrace in the middle epoch between 1940 and 1942.
ENSO vs. Temperature anomaly Events
Parametric comparisons of ENSO and the Temperature anomaly are in Figures 22 and 23, the cumulative first, yielding the dates demarking the first order events, and second in raw data form, providing the best fits.
At the opening of the record in 1880, SOI is slightly negative, during which time the climate is dominantly below average 0.3 ºC. This lasts until 1918 when ENSO enters a 60 year cooling state, interrupted for just two years beginning in 1940 by a sharp warming signal. Meanwhile, global temperature is within ± 0.1ºC of its long term average. In the middle of 1977, ENSO turned sharply toward El Niño by 4.2 units. During this time until the present, global temperature accelerated an average of over 0.4 ºC.
ENSO Effects
The Consensus claims that ENSO has global implications, and that its contribution to global temperature is probable.
Warm episodes of the El Niño-Southern Oscillation (ENSO) phenomenon (which consistently affects regional variations of precipitation and temperature over much of the tropics, sub-tropics and some mid-latitude areas) have been more frequent, persistent and intense since the mid-1970s, compared with the previous 100 years. Climate Change 2001, Summary for Policy Makers, p. 5.
Warm phase ENSO episodes have been relatively more frequent, persistent, or intense than the opposite cold phase during this period. [¶] This recent behavior of ENSO is related to variations in precipitation and temperature over much of the global tropics and subtropics and some mid-latitude areas. The overall effect is likely to have made a small contribution to the increase in global surface temperature during the last few decades. Climate Change 2001, p. 103.
Thus, cooler nutrient-rich waters upwell from below along the equator and western coasts of the Americas, favouring development of phytoplankton, zooplankton, and hence fish. Climate Change 2001, Technical Summary of the Working Group I Report, p. 52.
By cumulative analysis, the change attributed to the “mid-1970s” can be objectively set to the period of 1977.5 to 1979.4. The quantification of the ENSO signal to Los Niños events alone by the Consensus obscures indications of major shifts and trends in Pacific circulations.
In the upwelling discussion, the Technical Summary omits the coincident CO2 increases observed by Keeling and Revelle:
During “normal” years the partial pressure of carbon dioxide in surface ocean waters near the equator in the Eastern Tropical Pacific is 60 to 80 parts per million higher than in the atmosphere, and there is a flux of about 0.6 gigatons of carbon from the sea to the air. During these years the Southern Oscillation Index is positive, that is, the air pressure off the coast of South America is higher than in the far western Pacific. The excess CO2 is carried to the surface by water upwelling from depths of between 50 and 150 meters. These upwelling waters are also rich in plant nutrients, resulting in intense biological production and the settling out from the surface layers into deep waters of particulate organic matter containing nearly 1 gigaton of carbon.
During “El Nino” years, when the Southern Oscillation Index is negative, upwelling and biological productivity virtually cease, the surface waters are depleted in nutrients, and the carbon dioxide partial pressure in the sea is about the same as in the atmosphere. Consequently there is no appreciable flux of CO2 from tropical waters into the air. Bold added, Keeling, C.D. and R. Revelle, Effects of El Nino/Southern Oscillation on the Atmospheric Content of Carbon Dioxide, Meteoritics, Vol. 20, No.2, Part 2, June 30, 1985. P. 437.
These El Niño factors are discussed in the main body of the TAR, but in the context of “CO2 variability” and a “reduced upwelling of CO2-rich waters”. Climate Change 2001 in ¶3.5.2, pp. 208-209. Regardless, the Consensus concludes with a contrary observation:
In any case, the slowdown (of the early 1990s) proved to be temporary, and the El Niño of 1998 was marked by the highest rate of CO2 increase on record, 6.0 PgC/yr. Id., p. 210.
Signal analysis supports a different view of the relationship between ENSO and global temperature.
ENSO & Temperature anomaly Relationship
First is the autocorrelation function of the Southern Oscillation Index. Figure 24.
This figure shows that the well-known cyclic behavior of ENSO has a period of 3.38 years (“preferred period of about three to six years”, Climate Change 2001>, Technical Summary, p. 52), and again no evidence of the solar 11-year cycle.
Next is the cross correlation function between Temperature and the negative of the Southern Oscillation Index. Figure 25.
Because a negative going SOI is a warming trend, the cross-correlation calculation includes a sign change to preserve the usual orientation of the function. A well-defined peak in correlation of 0.46 occurs at a five month lag in the temperature record.
The TAR says,
Whether global warming is influencing El Niño… is a key question, especially as El Niño affects global temperature itself. Citations omitted, Climate Change 2001, p. 151.
The five month lag in temperature suggests the answer is no. Surface temperature is a weak, lagging indicator of ENSO, not a predictor.
Next is the scatter diagram of Temperature lagged by five months and the Southern Oscillation Index for the period of 1880 to 2007. Figure 26.
The chart shows the small but significant relationship that global temperature tends to increase with decreasing SOI. In a linear regression, the SOI could account for 4.6% of the variability (power) in the temperature anomaly.
The Measured Correlations Are Not Likely Due to Noise
Finally as a demonstration and validation of the conclusions, examine the correlation between the Solar Wind Index and the Southern Oscillation Index in a scatter diagram. The two should be orthogonal (r2=0). As shown in Figure 27, r2 = 0.0015.
An elementary simulation shows the probability of r2 being at least this large due to noise alone is a weak but not improbable 10%. Accepting the hypothesis that the traces are orthogonal is bolstered by the facts that r-squared for temperature and solar wind is 60 times larger, and for temperature and the southern oscillation it is 31 times larger.
These results support a model in which global temperature does not affect ENSO. Because the solar wind affects global temperature, if the temperature in turn influenced ENSO, the solar wind and ENSO would be correlated. They are not measurably correlated. This conclusion is supported by the fact that temperature lags ENSO.
CONCLUSIONS
According to the Consensus, “ENSO … play[s] a fundamental role in global climate” (Climate Change 2001, Technical Summary, p. 51) The Consensus on Climate said of its destructive power,
Changes associated with ENSO produce large variations in weather and climate around the world from year to year. These often have a profound impact on humanity and society because of associated droughts, floods, heat waves and other changes that can severely disrupt agriculture, fisheries, the environment, health, energy demand, air quality and also change the risks of fire. ENSO also plays a prominent role in modulating exchanges of CO2 with the atmosphere. The normal upwelling of cold nutrient-rich and CO2-rich waters in the tropical Pacific is suppressed during El Niño. Id., p. 52.
although the ENSO correlation with carbon dioxide proved fleeting:
In any case, the slowdown (of the early 1990s) proved to be temporary, and the El Niño of 1998 was marked by the highest rate of CO2 increase on record, 6.0 PgC/yr. Climate Change 2001, p. 210.
The Consensus accounts for the various natural sources for climate variability, and the remainder it must attribute to man:
Any human-induced changes in climate will be embedded in a background of natural climatic variations that occur on a whole range of time- and space-scales. Climate variability can occur as a result of natural changes in the forcing of the climate system, for example variations in the strength of the incoming solar radiation and changes in the concentrations of aerosols arising from volcanic eruptions. Natural climate variations can also occur in the absence of a change in external forcing, as a result of complex interactions between components of the climate system, such as the coupling between the atmosphere and ocean. The El Niño-Southern Oscillation (ENSO) phenomenon is an example of such natural “internal” variability on interannual time-scales. To distinguish anthropogenic climate changes from natural variations, it is necessary to identify the anthropogenic “signal” against the background “noise” of natural climate variability. Climate Change 2001, Technical Summary, p. 25.
In this accounting, the Consensus places ENSO third on its list after solar radiation and volcanoes. It excluded the solar wind, arguing
We conclude that mechanisms for the amplification of solar forcing are not well established. … At present there is insufficient evidence to confirm that cloud cover responds to solar variability.Climate Change 2001, ¶6.11.2.2 Cosmic Rays and Clouds, p. 385
The evidence has been hidden in the climate records for decades. It was not just that the solar activity was linked to cloud cover, but that it was linked to global surface temperature. The solar wind could account for 8.9% of the variation in the Temperature anomaly, 1.93 times the power of ENSO, which accounts for 4.6% of the surface temperature.
Climate signal analysis establishes the following:
1. Major state changes appear in the global temperature record around 1934.4 and 1979.5.
2. A major state change occurred in the solar wind index around 1937 to 1939, and a secondary state change occurred in the 1970s.
3. Major state changes occurred in the Southern Oscillation Index beginning about 1919.3 and 1979.4. A large state change occurred during the brief period of 1940.2 to 1942.0.
4. The state changes are real in the records, but may be due either to data acquisition artifacts or to real physical phenomena.
5. The Southern Oscillation Index has a weak cyclic behavior with a period of 3.38 years.
6. Global temperature lags the Southern Oscillation Index by about 5 months.
7. The global temperature record appears to suffer from excessive processing.
8. High correlations found by other investigators may be the result of prior data smoothing.
9. The low level of correlation between temperature and other parameters may be due to excessive noise, equivalently due to low signal to noise ratio. More importantly, it may be due to the closed loop gain of a mechanism in the climate, unknown to the Consensus, that regulates global surface temperature.
10. Global temperature is weakly correlated with ENSO. The SOI could account for 4.6% of the measured variation in global temperature.
11. Global temperature and the solar wind index are correlated. The solar wind index may contribute as much as 8.9% of the processed global temperature variations.
12. Global temperature lags the solar wind index by about two to five years.
13. ENSO and the Southern Oscillation affect the global surface temperature. The reverse, that temperature might affect either, is not true.
ENSO may devastate, but it has only half the capacity of the solar wind to warm the planet. By omitting the solar wind, the Consensus underestimates the natural causes of global warming, simultaneously overestimating the anthropogenic sources by the equivalent of two ENSOs, assigning the error to carbon dioxide emissions.
BIBLIOGRAPHY
IPCC, Fourth Assessment Report (4AR), 2007
IPCC, Third Assessment Report (TAR), Climate Change 2001
Keeling, C.D. and R. Revelle, Effects of El Nino/Southern Oscillation on the Atmospheric Content of Carbon Dioxide, Meteoritics, Vol. 20, No.2, Part 2, June 30, 1985
Köhnlein, W. Cross-correlation of solar wind parameters with sunspots ('long-term variations') at 1 AU during cycles 21 and 22, Astrophysics and Space Science, v. 245: 81-88. 11/13/96
Landscheidt, T., Solar wind near earth: indicator of variations in global temperature, Proceedings of 1st Solar & Space Weather Euro Conference, 9/29/00
National Oceanic and Atmospheric Administration (NOAA), National Climatic Data Center, Global Surface Temperature Anomalies, 2/6/06
National Oceanic and Atmospheric Administration (NOAA), National Climatic Data Center, Southern Oscillation Index (aa)
Tinsley, B.A., 1996, Correlations of atmospheric dynamics with solar wind-induced changes of air-earth current density into cloud top, J. Geophys. Res., 101, 29701-29714 ($9)
© 2007 JAGlassman. All rights reserved.



























