Natural Variability

Natural Variability

The first edition of the textbook Physics of the Air by W. J. Humphreys was published in 1920 with a second edition appearing in 1929 and the third [and final] edition appearing in 1940.

Physics of the Air was considered “a standard work” between the 1920s and 1940s and provides a wonderful insight into Atmospheric Science before the era of Settled Science.

This posting references the second edition [1929] of Physics of the Air [that can be read or download via the website] but it should be noted that the preface to the third edition [1940] notes that it “contains no radical departures from either the plan or the scope of the second” although the third edition does correct a few errors and includes some additional information.

The chapter called Vulcanism: Observational in this “standard work” is especially interesting because its observational approach provides a refreshing counterbalance to the mind numbing orthodoxy [and scientific illiteracy] of modern Settled Climate Science.

Solar irradiance was measured [in this era] using pyrheliometers which were [very sensibly] located “at the surface of the Earth” [rather than in a Low Earth Orbit of some flavour].

This ensured that the [net] solar irradiance measured at the surface of the Earth included the gains and losses attributable to atmospheric gases, atmospheric dust and [even] solar variability.

The [very striking] results [from this era] indicate that combining sunspot numbers with surface solar irradiance produces a curve that “actually parallels the curve of temperatures” with “remarkable fidelity”.

Vulcanism: Obervational

It will be interesting and profitable, now, to consider the supplementary portion of the theory of the relation of vulcanism to climate.

That is, to consider the observational evidence, pyrheliometric or other kind, bearing on the effect of volcanic dust on solar radiation, and, thus, obtain some idea of those absolute values essential to even a rough determination of the climatic consequence of volcanic dust in the high atmosphere.

Pyrheliometric Records.
Direct measurement of solar radiation by means of the pyrheliometer, an instrument that measures the total heat of sunshine, shows marked fluctuations from year to year in the intensity of this radiation as received at the surface of the earth.


This subject has been carefully studied by Dr. H. H. Kimball, of the United States Weather Bureau, who prepared the accompanying table, graphically represented by Fig. 223.

Annual Average Pyrheliometric Values

Since the yearly values are given in terms of the average value for the entire period, it is obvious that percentages of this average do not represent the full effect of the disturbing causes, of which volcanic dust certainly is the chief.

The following table of intensities was computed from observational data obtained at the following stations:

Montpellier, France, monthly means (noon values) 1883-1900
Pavlovsk, Russia, monthly maxima 1892-1913
Lausanne, Switzerland, monthly means (noon values) 1896-1904
Warsaw, Poland, monthly maxima 1901-1913
Washington, D. C., and Mount Weather, Virginia, monthly means for air mass 2.0 1905-1913
Simla, India, monthly means (noon values) 1906-1913
Paris, France, monthly maxima 1907-1913

Annual Average Pyrheliometric Table

A few other stations have been established since 1908.

The marked decrease in the pyrheliometric readings for 1884, 1885, and 1886, doubtless were largely, if not almost wholly, due to the eruption of Krakatoa in the summer of 1883; the decreased values of 1888 to 1892, inclusive, occurred during a period of exceptional volcanic activity, but were probably owing essentially to the violent eruptions
of Bandaisan (1888), Bogoslof (1890), and Awoe, on Great Sangir (1892); the low values of 1903 to the eruptions of Santa Maria (1902), Pele (1902) and Colima (1903); and the low values of 1912-1913, to the explosion, June 6, 1912, of Katmai.

The slight depression in the curve corresponding to the year 1907, during which no violent eruptions were reported (this docs not exclude the possibility of such occurrence in remote and unfrequented regions), according to Dr. Kimball, probably was caused by local haze at Washington, D. C., where his observations were made, and elsewhere, and this supposition is partially supported by the fact that his values for the year were not uniformly low, and by the further fact, inferred from a publication by Gorczynski, that during that year the solar radiation was but little below normal at Warsaw, Poland.

By July, 1914, the depression due to the dust from Katmai had fully passed, and from that date to the present (July, 1928), there have been no violent volcanic eruptions and the pyrheliometric values have been remarkably constant except for a slight unexplained irregularity in 1920-1921.

There is, then, abundant pyrheliometric evidence that volcanic dust in the upper atmosphere actually does produce that decrease in direct solar radiation that theory indicates it should, and, as the theory is well founded and the observations were carefully taken, this mutual confirmation may be regarded as conclusive both of the existence of volcanic dust in the upper atmosphere (isothermal region) and of its efficiency in intercepting direct radiation from the sun.

It should be remembered, however, in this connection, that the intensity of the solar radiation at the surface of the earth depends not only upon the dustiness of the earth’s atmosphere, but also upon the dustiness, and, of course, the temperature, of the solar atmosphere.

Obviously, dust in the sun’s envelope must, more or less, shut in solar radiation just as, and in the same manner that, dust in the earth’s envelope shuts it out.

Hence, it follows that when this dust is greatest, other things being equal, the output of solar energy will be least, and when the dust is least, other things being equal, the output of energy will be greatest.

Not only may the intensity of the emitted radiation vary because of changes in the transparency of the solar atmosphere, but also because of any variations in the temperature of the effective solar surface, which, it would seem, might well be hottest when most agitated, or at the times of spot maxima, and coolest when most quiescent, or at the times of spot minima.

Now, the dustiness of the solar atmosphere, manifesting itself as a corona, certainly does vary through a considerable range, from a maximum, when the sun-spots are most numerous, to a minimum, when they are fewest; and, therefore, partly because of changes in the transparency of the solar envelope, and partly because of changes in the solar surface temperatures, if, as in all probability they do, such temperature changes take place, we should expect the solar constant also to vary from one value at the time of spot maximum to another at the time of spot minimum, and to vary as determined by the controlling factor, dust or temperature.

If the above reasoning is correct, it follows that pyrheliometric readings are functions of, among other things, both the solar atmosphere and our own terrestrial atmosphere; and as the former is altered chiefly by sun spots or at least varies with their production and existence, and the latter by volcanic explosions, a means is at hand for comparing the relative importance of the two radiation screens.

Figure 224 shows one such comparison.

The upper curve gives smoothed annual average pyrheliometric readings (not solar constants, though closely proportional to them) and the lower curve sun-spot numbers.

It will be noticed that, in their most pronounced features, the two curves have little in common, and that the great drops in the pyrheliometric values occur simultaneously with violent volcanic explosions, as already explained, and not at the times of sun-spot changes.

Hence, it appears that the dust in our own atmosphere, and not the condition of the sun, is a very important, if not the controlling, factor in determining the magnitudes and times of occurrence of great and abrupt changes of insolation intensity at the surface of the earth.

Relation of Pyrheliometric values, sunspots and volcanic eruptions

Temperatures at the Surface of the Earth.

If a veil of dust actually should intercept as much as one-fifth of the direct solar radiation, as Fig. 223 indicates that, at times, it does, it would seem that in those years the temperature of the atmosphere at the surface of the earth should be somewhat below the normal.

Of course, the great supply of heat in the ocean would produce a lag in this effect, particularly over the oceans themselves, and, besides, there must be both an increase of sky light by scattering and some interception of earth radiation by the dust which, since it is at great altitudes, receives the full, or nearly the full, planetary radiation of the earth.

This increase of sky radiation, together with the return terrestrial radiation, obviously compensates in some measure for the loss of direct insolation.

Measurements, however, made by Abbot, at Bassour, Algeria, during the summer of 1912, show that at this time and place the direct radiation and the sky radiation, which obviously included both the scattered solar radiation and some return terrestrial radiation, were together less by about 10 per cent than their normal combined values; and there is no reason to think that in this respect Bassour was at all different from other places, certainly a large portion of the northern hemisphere, at least, covered by the veil of dust.

Clearly, then, if this decrease in the radiation received were universal and should continue indefinitely, the ultimate radiation of the earth would also decrease to the same extent, or 10 per cent.

Now, since the earth, or rather the water vapor of the atmosphere, mainly, radiates substantially as a black body, and, therefore, proportionally to the fourth power of its absolute temperature, it follows that a 10 per cent change in its radiation would indicate about a 2.5 per cent change in its temperature.

But the effective temperature of the earth as a full radiator, which it closely approaches, is about 252 Abs.

Hence a change of 10 per cent in the radiation emitted would imply 6.3 C. change in temperature, an amount which, if long enough continued, would be more than sufficient to produce glaciation equal, probably, to the most extensive of any known ice age.

As above implied, not much lowering of the temperature could be expected to take place immediately; however, some early cooling over land areas might well be anticipated.

To test this point, the temperature records of a number of high altitude (together with two or three very dry) inland stations have been examined.

High altitudes were chosen because it was thought that the temperature effects of dust in the upper atmosphere probably are most clearly marked above the very and irregularly dusty layers of the lower atmosphere; and the condition that the stations should also be inland was imposed because these are freer, presumably, than many coast stations, from fortuitous season changes.

Thus, stations in the eastern portion of the United States were rejected because of the great differences in the winters, for example, of this section depending upon the prevailing direction of the wind, 2 a condition wholly independent, so far as known, of variations in the intensity of direct radiation.

The number of stations was still further limited by the available recent data.

Hence the records finally selected, and kindly put in shape by the Climatological Division of the United States Weather Bureau, P. C. Day in charge, were obtained at the following places:

American Stations

In Table III, the first column gives the year in question.

The second column gives the average departure in degrees Fahrenheit, for the seventeen American stations, of the annual average maximum, as determined from the monthly average maxima, from the normal annual maximum, or average of a great many annual average maxima.

The third column gives smoothed values, determined from the actual values in the second column as follows:

America Average Temperature Departures

in which S is the smoothed value, b the actual value pertaining to the particular year for which S is being computed, a and c the actual values for the next previous and the next succeeding years, respectively.

The fourth and fifth columns give, respectively, the actual and the smoothed average departures of the annual average minima, while the sixth and seventh columns give the corresponding average departures of the annual average means.

Figure 225 shows the graphical equivalents of the smoothed portions of Table III, to, and including, 1912.

America Average Temperature Graph

The values after that date show nothing interesting except, roughly, the usual inverse relation of temperature to sun-spot numbers.

It will be noticed that the three curves of Fig. 225, marked maximum, minimum, and mean, respectively, are, in general, quite similar to each other.

Hence, because of this mutual check and general agreement, it seems reasonably certain that any one set of temperature data, the means, for instance, furnishes a fairly safe guide to the actual temperature and climatic fluctuations from year to year, or period to period.

Table IV gives the weighted actual average departures and the smoothed departures in degrees Fahrenheit of the annual mean temperatures of the selected seventeen American, seven European, and one Indian stations listed in Table I.

World Temperature Departures

in which D is the weighted departure, A the smoothed average American, E the smoothed average European, and I the smoothed Indian, departure of the mean annual temperature from the normal annual temperature.

Table IV, extended, as well as the scanty early data, mainly from the given stations, will permit, back to 1872, is graphically represented by the continuous, light curve at the bottom of Fig. 226.

In 1880, and again in 1901, the curve probably does not very closely represent worldwide temperature departures, being, presumably, at both places quite too low, owing, in each case, to an abnormally cold single month in America.

The dotted curve from 1907 to 1911 gives the average temperature departures for the American stations only, and, presumably, represents world temperature departures much more closely than does the continuous light line for the same time.

This is because of two or three exceptionally cold summer months in Europe.

The dotted curve from 1872 to 1900 gives the smoothed averages of the annual temperature departures from the normal temperatures of the following stations as computed from the actual departures given by Nordmann; Sierra Leone, Recife (or Pernambuco), Port au Prince, Trinite, Jamaica, Habana, Manila, Hong Kong, Zikawei, Batavia, Bombay, Island of Rodriguez, Island of Mauritius.

Smoothed pyrheliometric, sunspots and temperature curves

All these, or practically all, are low-level stations, and most of them either tropical or semitropical, and, therefore, should show in general, from altitude influence alone, a smaller, and from latitude influence alone, a greater, abnormality than do the stations whose temperature departures are given by the continuous fine-line curve.

Hence, all things considered, the average temperature departures as calculated from the two sets of stations agree remarkably well, so that one can say with, presumably, a fair degree of confidence, that the heavy curve T approximately represents the average of the departures of the mean annual temperatures from the normal annual temperatures of equatorial and high altitude regions of the earth, or that T, with the above restrictions, is the curve of world temperatures.

Much additional statistical evidence bearing on this point and supporting the conclusion just given has been published by Mielke.

This consists of the average annual temperatures from 1870 to 1910 of 487 widely distributed stations, with, however, numerous and extensive breaks in fact, the records of only a few stations cover the entire period.

By grouping these stations according to zones, tropical, subtropical, warm temperate, cold temperate and frigid, and then averaging and smoothing the zonal annual temperature departures, giving all equal weight, values were found which run substantially parallel to those already found but of less (about one-half) amplitude, quite as anticipated from the fact that stations above the dust, fogs, and many clouds of the lower atmosphere, must be more sensitive to variations in the transparency of the outer atmosphere and to solar changes than are those (the great majority) located at, or not more than a few hundred meters above, sea level.

Either curve might, therefore, be used in a discussion of the causes and periods of temperature changes, but in what follows the curve of larger amplitudes or the curve of high altitude stations will be used because:
(a) data for it, but not for the other, are available through the period of the Katmai veil of dust,
(b) it is freer from surface disturbances and, therefore, more representative of solar and high atmospheric conditions,
(c) high altitude temperatures are more effective than those of sea level in modifying glacial conditions.

Relation of World Temperatures to Pyrheliometric Values.

Curve P, also of Fig. 226, gives the smoothed course of the annual average pyrheliometric readings, as computed from the actual values given in Fig. 223.

The insolation intensity data, covering the whole of the depression that had its minimum in 1885, were obtained at a single place, Montpellier, France, by a single observer, L. J. Eon, who confined himself to noon observations with a Crova actinormeter.


It may be, therefore, that merely local and temporary disturbances produced a local insolation curve that was not quite parallel to the curve for the entire world.

At any rate, the drop in the solar radiation values obviously was due to dust put into the atmosphere by the explosion of Krakatoa in August, 1883, and it would seem that the effects of this dust both on the surface temperatures and on pyrheliometric values must have been greater during the latter part of 1883 and in 1884 than they were in 1885, when much of the dust, certainly, had already settled out of the atmosphere, and this supposition is well supported by the pyrheliometric and temperature drops that immediately followed the volcanic explosions of 1903 and 1912, and their partial recovery within a single year.

Nevertheless, the pyrheliometric values must be accepted as obtained.

Indeed, this exceptional lag is not quite unprecedented, since the coldest year following the similar, though more violent, explosion of Asamayama, just 100 years earlier, was not the year of the explosion, 1783, nor the following year, but 1785.

It is probable that in the earlier, as certainly in the later, of these unusual cases the dust was thrown to such great altitudes that the finer portions were nearly, or quite, 2 years in reaching the lower level of the isothermal region.

Clearly, too, much of this dust, while perfectly dry, probably was so fine as merely to scatter even solar radiation, and yet, on reaching the more humid portions of the atmosphere, the particles may have gathered sufficient moisture to assume reflecting size, and, therefore, seriously to interfere with insolation.

This is merely suggested, but in no wise insisted upon, as a possible explanation of the unusual pyrheliometric lag after the explosion of Krakatoa.

It is obvious, from a mere glance, that the pyrheliometric and the temperature curves, or curves P and T, have much in common.

This is especially marked by the large and practically simultaneous drops in the two curves in 1912, following the eruption of Katmai.

But while a relation between these curves thus appears certain, the agreement is so far from perfect as to force the conclusion that pyrheliometric values constitute only one factor in the determination of average world temperatures.

Sun Spots and Temperature.

It has been known for a long time that the curve of sun-spot numbers, curve S (Fig. 226), and the curve of earth temperatures, curve T, follow or parallel each other in a general way, in the sense that the fewer the spots the higher the temperature, with, however, puzzling discrepancies here and there.

Both these facts, the general agreement between the phenomena in question and also their specific discrepancies, are well shown by the curves S and T of Fig. 226, and, while the discrepancies are marked, it is obvious that, on the whole, the agreement is quite too close to leave any doubt of the reality of some sort of connection between sun spots and atmospheric temperatures.

Just how, or by what process, this relation, conceivably, may exist will be discussed below.

Combined Effect of Insolation Intensity and Sun-spot Influence on Atmospheric Temperatures.

Since it is obvious that the insolation intensity and the number of sun spots each exerts an influence on the temperature of the earth, it is clear that some sort of a combination of the two curves P and S should more closely parallel the temperature curve T than does either, alone.

It is probable that the sun-spot effect is not directly proportional to the actual number of spots, but, however this may be, the direct combination of the curves P and S gives the resultant P + S, which, as a glance at the figures shows, actually parallels the curve of temperatures T with remarkable fidelity.

Exactly this same combination, from 1880 to 1909, has been made by Abbot and Fowle, whose lead in this important particular is here being followed, and the resultant curve found to run closely parallel to the curve of “smoothed annual mean departures” of the maximum temperatures of 15 stations in the United States.

Probably the most striking point of agreement, one that must strongly be insisted upon, as shown by Fig. 226, between the combination curve and the temperature, occurs in 1912, when, in spite of the fact that the sun spots were at a minimum, indicating that, according to rule, the temperature should be high, the temperature curve dropped greatly and abruptly; obviously, because of the simultaneous and corresponding decrease in the intensity of solar radiation produced by the extensive veil of Katmai’s dust, precisely as happened at spot minima after the explosion of Asama, in 1783.

Both cases, since they occurred during spot minima, show distinctly the great influence volcanic dust has on terrestrial temperatures.

Volcanic Disturbances of Atmospheric Temperature Since 1750.

It must be distinctly remembered that the earlier temperature records, because of their limited number, if for no other reason, give only the general trend of world temperatures.

Again, the record, back to 1750, of even violent volcanic eruptions is necessarily incomplete; and, besides, not all great eruptions decrease the surface temperature only those that drive a lot of dust into the stratosphere, and, even then, decrease it perceptibly in only those regions, usually extensive and at times world wide, over which the dust spreads.

Pronounced and long-continued sky phenomena, therefore, of the type that followed the eruption of Krakatoa, furnish the best evidence of volcanic violence in the sense here used.

Finally, there can be no particular test save where the temperature is low in comparison with that which the number of sun spots would indicate.

Obviously, then, no matter how close the actual relation between the phenomena may be, the errors and the incompleteness of the recorded data would prevent the discovery of more than a general relation.

Of course, it will naturally occur to one to ask about special cases, such as the cold years of 1783-1784-1785, and, in particular, 1816, the famous “year without a summer,” “poverty year,’ or “eighteen hundred and froze-to-death.”

The first of these, 1783-1785, followed, as already explained, the great explosion of Asama, in 1783, while the second, the “year without a summer, ” that was cold the world over, followed the eruption of Tomboro, which killed 56,000 people, and blew up so much dust that “for three days there was darkness at a distance of 300 miles.

There is a detail in the temperature curve, for the years 1886-1887, that needs special attention.

The temporary depression where, seemingly, the temperature should be steadily rising, obviously was due to the great eruption of Tarawera (June 10, 1886), in New Zealand.

This volcano is a little more than 38 south of the equator, and, therefore, furnishes a good example of an eruption on one side of the equator affecting the temperature far to the other side.

Doubtless, however, when the dust gets but a little way into the stratosphere, the effect is greatest on the volcano’s side of the equator.

But, if the temperature was decreased by Tarawera, why, one might ask, was not the pyrheliometric curve similarly affected?

It was, for several months after the eruption, as the individual monthly values show, but the annual means, plotted in the figure, have the effect of making the pyrheliometric disturbance from Tarawera appear only as a retardation in the recovery from the effects of Krakatoa.

Neglecting the smaller irregularities which may or may not have been of world-wide occurrence, and remembering that, other things being equal, temperature maxima are to be expected at the times of spot minima and temperature minima at the times of spot maxima, the marked discrepancies and their probable explanations may be tabulated as follows:

Temperature and Sunspot Discrepancies

It may be concluded, therefore, that the variations in the average temperature of the atmosphere of the kind and magnitude shown by actual records depend jointly upon volcanic eruptions, through the action of dust on radiation, as already explained, and upon sun-spot numbers, through, presumably, some intermediate action they have upon the atmosphere possibly of the nature explained in the next chapter.

Physics of the Air – W. J. Humphreys – 1929 – McGraw-Hill Book Company

Which just goes to show what happens when you let the Space Cadets [in Houston] launch Climate Science into Low Earth Orbit.

Seems like Phil, Mikey and The Team need to get their feet back on the ground so they can become acquainted with sunspot numbers, atmospheric dust and the sharp end of a pyrheliometer.

Gallery | This entry was posted in Atmospheric Science, Books, Earth. Bookmark the permalink.

One Response to Natural Variability

  1. kuhnkat says:

    Don’t tell Willis over at WUWT that real scientists running real experiments proved that volcanoes caused temperature drop. His statistical game didn’t come out that way…

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s