METEOROLOGY OF THE LOWER LAYERS OF THE STRATOSPHERE

G. M. B. Dobson, A. W. Brewer, and B. M. Cwilong*)

I. INTRODUCTION

Since the time of its discovery at the end of the last century, the stratosphere has always aroused great interest not only among meteorologists, but also among scientists of other specialties. If at first the causes of its origin seemed so mysterious that many even doubted the reality of its existence, it is now generally agreed that the transition from the troposphere to the stratosphere corresponds to the transition from convection to radiation as the chief factor regulating the temperature of the air. The aim of the present lecture is to attempt to present a picture of the conditions in these high layers of the atmosphere as they are known today.

The temperature of any part of the atmosphere, of course, depends entirely on the energy radiated by the sun, but this process is often indirect and complex. Solar radiation, being considerable in the visible part of the spectrum and in the near infrared region, undergoes only slight absorption by the principal constituents of the atmosphere on its way downward, to the earth, where it produces heating of the soil. The air, with the exception of its very highest layers, is heated only slightly directly by the sun’s rays. In contrast to sunlight, the return radiation coming from the surface of the earth, being intense in the far infrared part of the spectrum, undergoes strong absorption by the atmosphere. Three secondary constituents of the atmosphere bear the main responsibility for the absorption, namely water vapor, carbon dioxide, and ozone; each of them possesses one or several strong absorption bands in the far infrared part of the spectrum. If no other processes act in the atmosphere, then the temperature of the air is determined by the absorption and emission of this long-wave radiation. In the lower part of the atmosphere the temperature must decrease rapidly with increasing height above the earth, but at a very great height, where the absorbing gases are too rarefied to cause strong absorption of radiation, the temperature of the air will be low and, with increasing height, will

*) Bakerian Lecture: “Meteorology of the lower stratosphere.” By G. M. B. Dobson, F. R. S.; with A. W. Brewer and B. M. Cwilong. Proc. Roy. Soc. A, 185, 144—175, 1946. Translated from the English by Prof. I. A. Khvostikov.

…only slowly. However, a rapid fall of temperature near the earth corresponds to an unstable state; very strong convection arises, which brings the magnitude of the temperature gradient to a stable value. There can hardly be any serious doubt as to the correctness of this general picture of temperature regulation in the atmosphere, namely, that in the lower part, in the troposphere, the controlling process is convection (at a definite soil temperature, determined by solar radiation), while in the upper part (the stratosphere and the warm layer above it) radiation is controlling.

As is known, Lord Chervell was the first to suggest that absorption of the reverse infrared radiation, coming from the earth and the lower layers of the atmosphere, by carbon dioxide and ozone must take part in establishing radiative equilibrium in the stratosphere, in addition to the influence of absorption by water vapor (Lindemann, 1919). Until recently, however, it was usually assumed that absorption by water vapor, extending over a wide region of the infrared spectrum, is of predominant importance. In Part 3 we shall consider the question of how fully this hypothesis is confirmed by the results obtained in recent years.

Of the three principal absorbing gases in the atmosphere, carbon dioxide has a very strong absorption band with its center near \(14.7\,\mu\) and several weak bands. Water vapor has a very complex absorption spectrum with weak bands in the near infrared, a strong band near \(6.5\,\mu\), and a very extensive region in the longest-wave part of the spectrum, where absorption is strong. Ozone, besides weak absorption bands, has a strong but extremely narrow band near \(9.7\,\mu\). It should be noted that this band falls in that part of the spectrum where both water vapor and carbon dioxide are almost transparent and where the intensity of the reverse radiation is large, since it comes from the earth’s surface almost without experiencing absorption by other gases.

The reverse radiation, passing upward through the lower layers of the stratosphere, does not all originate at one and the same level. Radiation of those wavelengths for which water vapor and carbon dioxide are very transparent, i.e., the region near \(10\,\mu\), comes predominantly from the earth’s surface, which has a mean temperature of approximately \(280^\circ\) abs. But those wavelengths which correspond to strong absorption by water vapor or carbon dioxide originate in the upper part of the troposphere, at such levels where there is already insufficient water vapor and carbon dioxide for strong absorption; the mean temperature of these layers is probably about \(233^\circ\) abs. Other wavelengths, which correspond to partial absorption by water vapor or carbon dioxide, originate in the lower layers of the troposphere, where the temperature is, say, \(260^\circ\) abs. This question was examined in detail by Simpson (1928) in his work “Further Studies in Terrestrial Radiation.” There he gave a curve showing the spectral composition of the radiation reaching the upper part of the stratosphere. This diagram,

if it is corrected by changing the temperature of 218° abs., used by Simpson, to 233° abs., gives a reasonable estimate of the radiation of various wavelengths reaching the base of the stratosphere (Fig. 17). In this way one can calculate the equilibrium temperature of some gas, which it must reach if only this gas is present and only it absorbs radiation of the given wavelength. The results of such calculations are given in Table 1, it being assumed that the absorption and re-emission of radiation in the high layers are so small that they may be neglected.

Table 1.

It is difficult to estimate the true radiative equilibrium temperature of water vapor in the stratosphere because of its very complex absorption spectrum, but a value of 190° abs. does not seem implausible. Thus we may take the equilibrium temperature of carbon dioxide as approximately 200° abs., that for water vapor as about 190° abs., while for ozone it comes out to about 250° abs. It may be expected that an increase in the relative content of ozone in the stratosphere is accompanied by an increase in temperature, whereas an increase in the relative fraction of water vapor should lower the temperature.

In order to try to calculate the actual temperature expected in the stratosphere, we must know the relative amounts of these three important gases. This question will be considered in the following part of the present work.

II. WATER VAPOR, OZONE AND CARBON DIOXIDE IN THE UPPER LAYERS OF THE ATMOSPHERE

a) Water vapor

1) Description of the hygrometer

Until quite recently there was no reliable information on the amount of water vapor in the high layers of the atmosphere, although it was usually assumed that these layers of air are close to saturation. With respect to

moisture of the air in the stratosphere, our ignorance was very complete and led to widely differing views. The great difficulties of measuring the humidity of the air at these heights are due to the fact that extremely small quantities of water vapor have to be measured at low temperature. Thus, if the air is saturated at 220° abs. (the mean temperature of the stratosphere in temperate latitudes), then one cubic meter of air contains only 26 mg of water vapor, while at 190° abs. (the mean temperature of the stratosphere above the equator) it contains only 0.3 mg/m³. For this reason, most of the usual methods of measuring humidity become inapplicable. Thus, for example, the difference in the readings of thermometers with dry and wet bulbs becomes vanishingly small, even if the air is very dry; materials sensitive to moisture, such as hair, take so long to absorb from the air the necessary amount of moisture (or to give it up to the air) that the lag becomes inadmissible; direct determination by taking air samples at altitude in vessels is made very inaccurate because of adsorption of most of the water on the walls of the vessel. It would be possible to measure the absorption of solar infrared radiation in the region of the strong water-vapor band at 6.5 μ, but, besides the need for rather complicated apparatus, lengthy and difficult laboratory measurements of the coefficient of absorption of water vapor at low temperature and low pressures would be required, since at present it is known for these conditions only inaccurately. Only the dew-point or frost-point method is free of the principal difficulties, and this method was developed by us for application under the specified special conditions. Since no instrument suitable for use on small free balloons carrying radiosondes had been developed, all measurements were made on airplanes.

As was indicated above, saturated air at a temperature of 190° abs. contains only 0.3 mg/m³ of water vapor, and only a small part of this amount is deposited on the mirror of the instrument if the measurements are carried out carefully. Therefore, if we want the method to operate satisfactorily, we must be able to detect extremely small deposits of ice and to determine whether this small deposit is increasing or whether the ice is evaporating. A method was found for measuring deposits much smaller than one-thousandth of a milligram of ice. Two types of instruments were constructed: in one of them the deposit is visible to the observer, while in the other the amount of deposit is measured photoelectrically from the amount of scattered light. In the instrument with visual reading such sensitivity is achieved by using a jet of air about a millimeter in diameter, directed across the flat surface of the cold mirror of the hygrometer. In this way the water vapor is deposited in the form of a noticeable stripe across the mirror, and the contrast with the surrounding clean part of the surface greatly helps to make the thin layer of deposit visible to the eye. The illumination of the mirror is also carried out carefully. The mirror is made of

of aluminum, and its upper flat surface blackens, while uniform illumination from all sides at small angles is achieved by means of a reflector surrounding the thimble; the elliptical reflector is made of glass or of transparent synthetic

Labels in the figure: transparent elliptical body; illuminating lamp; heating; resistance thermometer winding; pump; Dewar vessel; magnifier; jet; tube for the air under investigation; thimble; pump control handle; view of the ellipse in plan.

Fig. 1. Diagram explaining the principle of the construction of a hygrometer permitting determination of the ice point by visual observations.

plastic, with the lamp placed in a hole drilled at one focus of the ellipse, and the thimble located in the hole at the other focus. In this way an effect similar to dark-field illumination is obtained, making it possible to see very faint traces of hoarfrost clearly. To assist the eye, a magnifying lens is provided, covering this small cell from above.

The lower part is hollow, and the thimble is cooled to the desired temperature by forcing a jet of cold gasoline into this cavity. To cool the gasoline, the pump is surrounded by a mixture of solid carbon dioxide and gasoline. If desired, liquid air may be forced into the thimble if it is necessary to attain very low temperatures. With the aid of such a device it is not difficult to cool the thimble to the desired temperature and to maintain it constant within fractions of a degree. The construction of the instrument is shown in the schematic drawing in Fig. 1.

The procedure for carrying out the measurements is as follows: first, the thimble is cooled until a deposit is obtained. To make sure that this is ice and not supercooled water*), the thimble is cooled still further, after which it is heated to such a degree that the deposit almost, but not quite entirely, evaporates; then it is again cooled to a temperature close to the presumed frost point. Having held the thimble at this temperature for 30–60 sec., the deposit is again examined in order to determine whether it is increasing or evaporating. Proceeding in this way, one finds two values of the temperature at which the deposit just begins to increase or to diminish. The mean of these temperatures will be very close to the true frost point. It is important to emphasize the fact that the temperature of the thimble can be maintained one or two degrees below the true frost point of the air without any formation of deposit at all, and that at this low temperature the elasticity of the ice vapor is so small that the coating evaporates very rapidly; if the temperatures are determined by the first appearance of the deposit during cooling and at the moment when the deposit just begins to be destroyed by heating, then the mean of these two temperatures may not be the true frost point of the air.

Fig. 2. Diagram explaining the principle of the construction of a hygrometer that makes it possible to determine the frost point with a photoelectric deposit indicator.

Instead of visual observations of the deposit of ice or dew on the surface of the thimble, the amount of deposit can be determined photoelectrically. The general principles of the instrument are explained by Fig. 2. In this case

*) Supercooled water can easily be deposited on the thimble and remain liquid for a long time when cooled down to a temperature of as much as 240° abs.

the top of the thimble is covered with a polished sheet of gold and does not blacken, and the instrument is operated in such a way that, in the absence of deposit, only a very small amount of light falls on the photocell; this amount increases greatly when a small deposit of ice appears, owing to the scattering of light. The current of the photocell is amplified and read on a microammeter, which is connected so that the reading is close to zero when the surface of the thimble is clean. The reading of the microammeter is thus a rough measure of the amount of deposit on the thimble. Although the instrument contains more parts and requires more attention to keep it in order than an instrument of the visual type, it is nevertheless simpler to operate, since all that is required (after obtaining a small deposit and checking that it is ice and not water) is to bring the temperature of the thimble to the value at which the microammeter reading remains unchanged, which indicates that the amount of deposit is constant. The time spent on the measurement is very short, since the actual amount of deposit is quite immaterial; the only necessary condition is the constancy of this amount. Consequently, two independent measurements can be made in one minute. As with most instruments, the complication entails both disadvantages and advantages, and in any case one cannot be certain that an error has not crept into the measurements which would have been immediately noticed in measurements carried out by a good observer on an instrument with visual reading.

In principle, only one step remains in order to make the instrument fully automatic: the photocell current can control the cooling of the thimble, so that a small constant amount of deposit will be continuously present and the thimble will remain for a prolonged time at the frost-point temperature of the air. Improvements of the instrument in this direction have been undertaken.

Very careful critical experiments were made to check that the readings from these instruments give the true frost point of the air even at very low temperatures. For this purpose air was prepared having a frost point of 194° abs.: from the pump the air first passed through dry silica gel, whereby it lost most of the water vapor, then through a cleaned copper coil immersed in a bath of solid carbon dioxide and acetone, and finally through cotton wool placed in a metal cylinder—also cooled in a mixture of solid carbon dioxide and acetone—in order to trap individual ice particles that might have passed through the coil. It was assumed that the air emerging from this system should have a frost point equal to the temperature of the cooling bath, and the temperature difference between the thimble and the bath was measured with a thermocouple. Two somewhat different instruments of the visual type were used for observations carried out by three

observers, and it was clearly shown that the instruments indicate the true ice point and that the probable error of measurement under laboratory conditions is approximately \(\pm 1^\circ\mathrm{C}\). In the case of work on an airplane at great heights, the lack of convenience and the shortage of oxygen will have an effect, as a result of which the probable error will apparently increase. But if the air has a higher ice point, then work with the instrument is made easier owing to the larger quantity of water vapor contained in the air; in this case the accuracy of the measurements should improve considerably.

A similar critical test of instruments of the photoelectric type showed that a single determination for air with an ice point of about \(194^\circ\) abs. can easily be made in about half a minute, as is shown by comparison with measurements over the course of 5 minutes by means of visual instruments at the same low temperature; the accuracy of the measurements proves to be much higher. Since instruments of the photoelectric type make smaller demands on the training of the observer, it appears that, under conditions of aircraft measurements at great heights, instruments of this type should provide higher accuracy than visual ones.

Tests of this kind showed that down to an ice point of \(194^\circ\) abs. the hygrometer described gives completely reliable results. However, at lower temperatures an effect was discovered which, as we now understand it, creates a definite lower limit of temperatures at which this hygrometer can be used. Initially the effect was noticed when the thimble of the hygrometer was cooled to approximately \(160^\circ\) abs., while the air being used had an ice point of about \(194^\circ\) abs. The thimble was cooled to this low temperature in order to obtain a deposit of ice rapidly, but, contrary to expectations, no visible deposit at all formed, although the thimble remained at this low temperature for 15 minutes. However, when the thimble was heated to a temperature only \(1^\circ\) below the ice point of the air being used, the usual deposit of frost appeared. Further investigations showed that the effect begins to manifest itself at still higher temperatures. With the aid of an instrument of the photoelectric type, the rate of growth of the deposit was measured at different thimble temperatures and at different air humidities. It turned out that the magnitude of the deposit varies greatly from one experiment to another, but the general character of the phenomenon can be seen from the curves in Fig. 3. (Since repeated measurements did not give exactly identical values, there is no certainty that the curves will always intersect one another.) From these curves it can be seen that there exists a natural limit to the dryness of air that can still be measured by our hygrometer; it may be expected that if, nevertheless, drier air is used, a curve of the form \(ABCD\) will be obtained, when there exists only a very narrow temperature interval between \(A\) and \(C\) in which a measurable ...

to a change in precipitation. In still drier air no visible precipitation will be obtained at any temperature. It should be noted that the effect is not caused by a shortage of suitable nuclei on which water vapor can sublime, since if the thimble is covered with a thin crust of hoarfrost before being cooled to low temperatures, then the formation of precipitation ceases at the same low temperature.

Fig. 3. Observed rate of ice deposition in the hygrometer at different thimble temperatures and at different moisture contents in the air.

There is reason to believe that water vapor continues to be deposited even at very low temperatures, but rather in a glassy form and not in the form of ice crystals, and therefore the precipitate remains invisible. Thus, for example, if the thimble has been cooled, say, to \(150^\circ\) abs. by a jet of very dry air, then for about a minute we admit air with a frost point of about \(220^\circ\) abs., and after this again pass a jet of very dry air through the apparatus, very slowly warming the thimble at the same time, then a considerable deposit appears when the temperature of the thimble reaches approximately \(200^\circ\) abs., despite the fact that the temperature of the thimble is much higher than the frost point of the surrounding air. The deposit again evaporates in a minute or two. The simplest interpretation of this phenomenon is that, while the thimble is very cold, an invisible, very thin layer of vitreous ice is deposited on it, which slowly changes into crystalline ice as the thimble is warmed.

Bergon and Oliver (1936), applying the method of X-ray diffraction to the study of the sublimation of water vapor at low temperatures and low pressures, showed that if water vapor condenses on a metallic surface at a temperature below \(163^\circ\) abs., vitreous ice is formed. If this surface has a somewhat higher temperature, then a crystalline or semicrystalline deposit is formed. If the temperature of vitreous ice is raised, it turns into crystalline ice.

Another phenomenon, causing considerable trouble at low frost points, consists in the fact that it often proves extremely difficult to obtain a deposit at all if the thimble is cooled directly from room temperature.

of temperature, especially if its surface has been freshly polished. After cooling of the thimble, after some time a deposit begins slowly to form; it grows the faster, the longer the thimble remains at a low temperature, but after roughly an hour the deposit forms readily even at an ice point of 190° abs. It appears that the surface of the thimble is in some way “activated.” This “activation” does not disappear when the thimble is heated until the deposit has completely evaporated, but often it disappears when the thimble is heated to room temperature for the same interval of time. It was precisely the variability of this “activation” of the thimble surface that made it difficult to obtain repeatable values for the curves in Fig. 3.

2) Measurements of humidity in the upper layers of the atmosphere

According to the most generally accepted theory of the distribution of water vapor in the stratosphere, it was considered that in the tropopause the air is saturated or close to saturation and that in the stratosphere slow mixing takes place, which mixes air from the upper layers of the stratosphere with air situated immediately above the tropopause; as a result, the weight ratio of water to air is constant for all altitudes in the stratosphere and is equal to their ratio in the tropopause.

To test this theory, one may use certain observations carried out in the upper layers of the air.

Several ascents have already been made during which, with instruments of the type described above, humidity measurements were made; their results have shown the significance of such measurements for the study of the dynamics of the atmosphere, both of the stratosphere and of the troposphere. The chief attention in this case was concentrated on the stratosphere, but consideration of some results obtained for the middle and lower troposphere may make it easier to understand the results relating to the stratosphere.

Fig. 4. Ice point and air temperature according to observations in an anticyclonic inversion on 21 March 1945.

The study of ascents occurring through stable layers of the troposphere showed that only slow mixing takes place.

In the early months of 1945 the weather was of such a type that descending inversions were a very common phenomenon, i.e., very stable layers, which often persisted for several days and, if destroyed, soon appeared again. A large number of soundings were made through these inversion layers; a typical result is shown in Fig. 4, where a very stable layer is seen between 914 and 1680 m. The curve of frost points is of special interest. Near the top of the stable layer the frost point (and therefore also the amount of water vapor) passes through a minimum, and there exists there a depleted layer only 600 m thick, containing less water vapor than the layers lying immediately above and below it. Figure 5 gives the data from this same ascent: the

Fig. 5. Weight fraction of water vapor in the air on March 21, 1945.

logarithms of the weight of water vapor per gram of air are plotted as a function of height. The scale of the figure has been enlarged sufficiently to show more clearly the details of the inversion layer. This ratio—the weight fraction of water vapor in the air—changes very rapidly with height, which is the most remarkable fact: between 914 and 1530 m, in a thickness of only 616 m, the weight fraction of vapor in the air decreases in the ratio 13:1, and then between 1530 and 1680 m it again increases in the ratio 9:1. These depleted dry layers occur in most of the ascents, although in some they are less conspicuous than in others; these layers persist for several days without weakening, so that any mixing between the different layers must have been very slow.

These data are of great interest from the general meteorological point of view, but for our purposes the chief significance is that in stable air mixing between different layers must be extremely slow and that any theory based on vertical exchange of air in the stratosphere cannot—

may be correct. Thus the need for actual observations in the stratosphere becomes very real.

Up to now we have made thirteen ascents into the stratosphere. The ascents of August 26 and September 7, 1943, took place up to the beginning of the stratosphere and revealed a large decrease in humidity at the greatest heights reached, but the first good observations in the stratosphere were made during the ascent of December 22, 1943, the results of which are presented in Fig. 6; we believe that these are, in general, the first careful measurements of humidity made in the stratosphere. The ascent took place in a current of polar air, which was rapidly descending, as was indicated by the distribution of temperature and humidity in the troposphere. The lower boundary of the subsidence was less than 3 km, and the upper boundary was 8 km. The tropopause air did not descend.

Fig. 6. Frost point and air temperature according to observations during the first ascent accompanied by measurements in the stratosphere. Boscombe Down, December 22, 1943.

Immediately upon entering the stratosphere the frost point falls very rapidly, and in the upper part of the ascent the frost point could not be measured, since it was not possible to cool the thimble sufficiently to obtain a visible deposit.

Such a very rapid fall of the frost point immediately upon entry into the stratosphere is typical of all the ascents made so far; it also occurs on the other typical ascents shown in Figs. 7 and 7a. These ascents relate to various meteorological conditions; nevertheless, for many ascents it remains necessary to determine whether a fall of the frost point above the tropopause always occurs. The sudden decrease in relative humidity is consistent with the well-known fact that the fog trail usually left by an aircraft at great heights becomes much shorter when the aircraft enters the stratosphere, which indicates very dry air in which the trail quickly evaporates.

It is not easy to explain cases in which an aircraft leaves a long-persisting trail at a height of several thousand feet above the indicated tropopause. It is possible that the height taken for the tropopause is

incorrect, but there remains the possibility that at times the air in the lower layers of the stratosphere is close to saturation.

Further confirmation is still required as to whether the frost point actually continues to fall at still greater heights in the stratosphere. Up to now, ascents reaching great heights deep into the stratosphere, in which the hygrometer provided adequate cooling, have confirmed that the frost point remains, roughly speaking, constant, ranging from 190° to 205° abs.

Fig. 7. Frost point and air temperature according to observations of May 5, 1944.

Let us now consider how air in the stratosphere can become so extremely dry. It appears that the only way in which water vapor can disappear from stratospheric air is the cooling of rising air to a low temperature, causing sublimation of the water vapor. The ice particles that form settle out and leave the air. It is natural to suppose that this process occurs near the equator, where the stratosphere is coldest and where the ascending currents in the troposphere are very intense. The mean temperature of the stratosphere here is about 190° abs., although at times it may decrease to 182° abs. Day-to-day variations here are much smaller than in temperate regions.

At first glance it might seem that at these very great heights, where the temperature is very low and the air is very clean, considerable supersaturation should be required for sublimation to occur, and that if it does occur, the clouds formed should consist of the finest crystals, capable of settling only extremely slowly. In this connection, experiments at very low temperatures performed with Wilson’s chamber are of interest: moist air is suddenly expanded and, as a result, becomes supersaturated. It was found that even in very clean air (from which all nuclei had been removed) a cloud of ice crystals is readily formed, provided only that the temperature has fallen below 232° abs., and that sublimation occurs at supersaturations much lower than the threshold necessary for the formation of water droplets on ions.

at temperatures above 232° abs.*). Further, when the temperature falls below 190° abs., predominantly larger grains of ice are formed, like small hailstones, and not a cloud of small ice crystals. These large grains immediately fall out and thereby remove water from the cold air; they probably must be completely invisible from the ground, so that no clouds whatever will be seen. If certain causes lead to the formation of dry air, then at the same time there must occur an astonishingly slow diffusion of air through the tropopause, since on those days when the exceptional dryness of the air was established, there could have been no recent currents of air from low altitudes. In the case to which Fig. 6 refers, the air of the stratosphere was probably over Hudson Bay several days before that. If air passes to another level without condensation or evaporation of water, then the frost point changes only slightly; an ascent of 1 km causes the frost point to fall by approximately 1°. Consequently, no considerable changes in the frost point can be obtained in this way.

Fig. 7a. Frost point and air temperature according to observations at Boscombe Down, 30 May 1945, 13 h 00 min. The frost point in the upper part of the curve is the lowest of all observed.

Unfortunately, up to now we have not had the opportunity to carry out a sufficient number of humidity measurements in the stratosphere in order to clarify its

) Author’s note at proof correction.* The highest temperatures in the expansion chamber, as obtained from the ratio of the volumes before and after expansion, at which the first particles of ice appear, are recorded quite distinctly; assuming the expansion to be adiabatic, the following was obtained:

1) for air freed from nuclei, 225.5° ± 0.1° abs.,
2) in open air, 235.5° ± 0.5° abs.,
3) for air polluted with tobacco smoke, 242° ± 0.5° abs.

However, experimental proof has recently been obtained that the minimum temperature values reached as a result of expansion are in fact higher than those obtained from the formulae for the adiabatic process. Experiments are now being conducted for the exact determination of the minimum temperature values. As the most probable values for the above-mentioned cases at the present time the following may be indicated: 1) 232° abs., 2) 241° abs., 3) 246° abs.

changes from polar air to equatorial air, have not been able to establish its mean value over regions of high and low pressure. It seems probable that, when such investigations become possible, they will lead to very important discoveries.

b) Ozone

Let us now turn to a consideration of the role of ozone in the upper layers of the atmosphere, and first give a brief review of the already published work on this question, after which we shall consider some new results obtained in recent observations. Most measurements make it possible to determine the total amount of ozone in the atmosphere over the place of observation, although it is possible, even by means of observations made from ground level, to obtain some idea of the distribution of ozone over the various levels of the atmosphere.

Ozone has a very strong absorption band in the ultraviolet region of the spectrum between 3300 and 2200 Å, and this is used in most work for measuring the amount of ozone. With the aid of a suitable monochromator, two wavelengths are isolated in the solar radiation, and from the ratio of their intensities the absorption by ozone is computed, and hence the amount of ozone. One wavelength is chosen almost outside the ozone absorption band, and the other within the band, but not too far inside the band, since otherwise the incoming radiation will be excessively weakened and accurate measurement of its intensity will become impossible. The measurement of the relative intensity of the two selected wavelengths may be carried out photographically or photoelectrically; the latter method has the great advantages of simplicity and speed of observation, so that a single observation and the necessary calculations can easily be completed in 10 minutes.

From measurements like those described above it is known that, over any given place, the total amount of ozone in the atmosphere may at times fall to 1.5 mm*) and rise to 4.5 mm. This amount is, of course, very small if compared with a gas such as carbon dioxide, which has an equivalent thickness of 2.4 m; but if one judges by the strong absorption of radiation by it, even this small amount proves to be very important. Measurements made in the most diverse parts of the world show that near the equator the amount of ozone is always small and usually very constant, the mean value being 2.0 mm. On the other hand, in high latitudes considerable annual variations are observed: the amount of ozone is very large in spring, but in autumn it probably exceeds the amount of ozone near the equator by no more than 25%. In temperate regions,

*) The total amount of ozone in the atmosphere over any given place is usually given as the thickness of an equivalent layer of pure gas at standard temperature and pressure.

where most of the principal measurements were carried out, the average amount likewise has its greatest value in spring and its smallest in autumn in both hemispheres. The amount of ozone also varies considerably from day to day, depending on meteorological conditions; being small in the warm sector of a depression and in an anticyclone, it is large in the cold part of a depression. It seems beyond doubt that in most cases polar air is distinguished by a high ozone content, and equatorial air by a low one; nevertheless at present one cannot judge with certainty the origin of air masses on the basis only of changes in the ozone content.

An estimate of the vertical distribution of ozone in individual atmospheric layers can be obtained from observations of the absorption, due to ozone, of direct solar radiation and, at the same time, of light scattered by the air and arriving through the atmosphere from the zenith under a clear blue sky. Other measurements were performed with the aid of a spectrograph carried aloft by a balloon-sonde and measuring the amount of ozone above the level of observation (Regener, 1934). There is no reason why an appropriate spectrograph could not be taken up in an aircraft and ultraviolet spectra of the sun photographed at different altitudes; from these the concentration of ozone at different levels could be determined.

The determination of the mean height of ozone in the atmosphere was also made by Strong (1941). He measured the total amount of ozone from the absorption of the sun’s ultraviolet rays, as described above, and in addition the absorption in the band \(9.7\,\mu\). He showed that, for an unchanged amount of ozone, the absorption of this band varies in proportion to the fourth root of the total pressure under which the ozone is found. From these two measurements he was able to calculate the pressure at the mean height of the ozone layer, and hence its height. The method, of course, gives no information about the vertical distribution of ozone, but its great advantage is that it makes it possible, in a single set of measurements made at a definite moment, to estimate the mean height; thus changes of height can be followed just as changes in the amount of ozone can. The method is applicable only under conditions when the sun is unobscured, and therefore cannot be used under a cloudy sky.

The number of measurements of the vertical distribution of ozone is very small in comparison with the number of measurements of the total amount of ozone. The measurements show that the concentration of ozone in the atmosphere increases with height, reaches a maximum at about \(25\) km and, apparently, falls to a very small value at an altitude of \(50\) km. They also show that fluctuations from day to day occur chiefly in the lower stratosphere, say from \(10\) to \(20\) km—an important point to which we shall return in § III.

Methods of measuring the total amount of ozone used in earlier observations required direct solar rays, so that

observations became impossible in cloudy weather. Since the passage of fronts separating different air masses is usually accompanied by abundant cloudiness, in these cases it was possible to obtain only limited information on the details of changes in the amount of ozone. Recent measurements now make it possible to fill this gap. The passage of a warm front, when cold air is replaced by warm air, is accompanied by a decrease in the amount of ozone; since the warm front advances obliquely, at an angle, roughly speaking, of \(1/150\), the decrease in ozone takes place before the front reaches ground level, just as the appearance of high clouds indicates the approach—

Fig. 8. Synoptic maps showing the passage of a warm front and the measured amounts of ozone, expressed as percentages of the mean seasonal value.

—of the warm front some time before the front reaches the surface. In Figs. 8, 9, and 10 a well-defined case of a warm front is presented; the amount of ozone is given as percentages of its normal value for the given time of year. Figure 8 shows the pressure distribution and the position of the frontal surface over three days. On the first day, when a region of low pressure covered the British Isles, the amount of ozone at all stations exceeded the normal seasonal value. On the second day, in western Ireland, the ozone content fell to \(88\%\) of normal by 13 hours and to \(76\%\) by 15 hours. At Lerwick the ozone also decreased to \(96\%\), but at the other stations the ozone content remained just as high as, or even higher than, on the preceding day. At this time the warm front was approaching western Ireland, but it was still 600 miles from the coast. On the third day the ozone content over all British stations had fallen to quite small values, although the front still remained west of Ireland. Unfortunately, at the British stations the decrease in ozone content could not be followed in greater detail by means of night obser-

tions, but two such observations in western Ireland show how rapid this decrease was. On the fourth day the front (now already occluded) passed through the British Isles, but another warm front, which approached Ireland, maintained the low ozone content at Valentia.

It is interesting to note that on the Scottish islands, which remained to the north of the surface of the warm sector, still smaller amounts of ozone were observed—a circumstance that was often repeated in other similar cases.

Fig. 9. Composite map showing the distribution of ozone relative to the warm front indicated in Fig. 8.

In Fig. 9 the front is plotted as it corresponds to the time of 13 hours on the second day; the ozone-content values are marked at the points corresponding to the positions of the stations that made the observations. Earlier and later ozone measurements are also included here; they are placed at points corresponding to the approximate positions of the stations relative to the front at the moment of measurement. The isopleths of the amount of ozone show the general distribution of ozone relative to the front. Fig. 10 represents an attempt to give a cross section of the atmosphere at right angles to the direction of the front. Measurements of the height of the tropopause were not made, so it had to be inferred by analogy with other similar cases. The line in the upper part of the diagram indicates the region, relative to the position of the front, in which the most rapid decrease in the amount of ozone occurs.

Fig. 10. Cross section of the atmosphere through a warm front (see Fig. 8), showing the region of decreased ozone amount.

It can be seen that the rapid decrease in ozone occurs in that region where the front passes through the troposphere at a height of about 8 km, but we think that the decrease in ozone is more likely to have

a direct continuation of the front into the stratosphere. There is no information on the inclination of the front in the stratosphere, but a backward inclination, over the warm air, is probable. There is also no information on the thickness of the transition region occupied by the front.

This case was chosen in order to show the large changes in the amount of ozone during the passage of a warm front, and also that, as a rule, the first signal of the approach of a warm front that can be obtained from ground observations is a decrease in ozone content, which is observed before the pressure begins to fall and often before the appearance of any frontal clouds. However, there are rare cases of marked warm fronts that do not show these ozone changes; it may be supposed that such fronts do not extend upward into the stratosphere.

Fig. 11. Synoptic maps showing the passage of a cold front and the increase in the amount of ozone behind the front. Ozone amounts are indicated as percentages of the mean seasonal value.

In the case of cold fronts, ozone changes, as might be expected, have, generally speaking, the opposite character compared with the changes at a warm front; usually, during the passage of a cold front, the ozone content increases. Further, since a cold front is usually inclined backward, but at a greater angle than a warm front, the change in the amount of ozone is observed soon after the passage of the cold front at the surface. As in the case of a warm front, not all cold fronts show changes in ozone content; these cases are again explained by the assumption that the cold front does not extend upward into the stratosphere.

Figures 11 and 12 show a typical case of the passage of a cold front*). In this case the increase in ozone content

*) It should be noted that this front is shown as an occlusion. This is due to the fact that historically it was connected with a warm front, but it has all the

was clearly noted both at Oxford and at Eskdalemuir, and all observations were made with good accuracy. It should be noted that the increase at Eskdalemuir was exceptionally rapid. Fig. 12 is an attempt to depict a cross-section of the atmosphere at right angles to the cold front, although there were no reliable data that would have made it possible to fix with great accuracy the position of the cold front in the higher layers of the air. The height of the tropopause also had to be determined conjecturally from observations made at some distance from the front.

The position of the front in the troposphere, as in Fig. 10, was fixed on the basis of such meteorological data as deserve confidence; but the position of the front above the tropopause was determined solely on the basis of a consideration of the changes in the amount of ozone. It is possible that the frontal region in the upper layers of the atmosphere is more diffuse than near the earth’s surface.

Fig. 12. Cross-section of the atmosphere through a cold front (see Fig. 11), showing the region of increase in the amount of ozone.

In some cases the increase in ozone content began, evidently preceding the presumed position of the cold front at the earth’s surface; but in many of these cases the temperature of the high layers of the atmosphere also began to fall before the cold front, which indicates that a change of air masses had in fact already taken place. One case is especially interesting: when, under a completely cloudless sky, an increase in ozone was observed, and only a few hours later a cold front arrived. A high cold front passed east—northeast over the British Isles, accompanied at a distance of about 100 miles by a cold front at the earth’s surface, the second cold front gradually catching up with the first. The upper cold front reached Oxford at about 19:00. There appeared there many alto-cumulus castelatus and a certain number of tower-like cumulus clouds with bases situated very high. The second front passed at about 21:00. The results of measurements of the ozone content and information on weather changes are given in Table 2 and in Fig. 13.

There can be no doubt that the increase in ozone content at 15:06 was associated with the upper cold front, which—

with characteristics of a genuinely cold front. See the temperatures in northwestern Ireland.

Table 2

4 hours later it passed over the station. In this case there were no indications of a decrease in the temperature of the high layers of the air which would have anticipated the presumed position of the front at the earth’s surface, but this could simply have escaped attention.

Fig. 13. Changes in the amount of ozone in the atmosphere under a cloudless sky, preceding an upper cold front, 6 June 1942.

Let us now turn to changes in the amount of ozone in the air when an occlusion passes over the observing station. It was established that the phenomena which are usually all called occlusions may be subdivided, according to their influence on the ozone content, into three groups:

1) True occlusions, representing the upper continuation of the warm sector of a depression, a tongue of warm air extending upward and clearly overtaking the end of the warm sector at the level of the earth. Such occlusions (as should have been expected) show a decrease in ozone content, often extending for several hundred miles ahead of the warm sector at the earth’s surface.

2) Some occlusions exert no influence on the ozone content; presumably, they do not extend to stratospheric heights.

3) Some occlusions show a noticeable increase in the amount of ozone during the passage of the occlusion over the observing station. The increase may amount to 25% or more of the initial amount of ozone and usually persists only for a few hours; but if the occlusion is nearly stationary, the high ozone content may be maintained for an entire day. Occlusions

Fig. 14. Composite diagram of ozone distribution relative to certain types of occlusion.

of this third kind may either be associated with the warm sector at some distance from the center of the depression, or they may not be associated with any warm sector at all. In all cases the air above the occlusion appears to be cold rather than warm. Fig. 14 presents a generalized diagram showing the distribution of ozone around occlusions of the third type.

A very large increase in ozone was found (in this case it should be considered to occur in the troposphere) during thunderstorms and with cumulus and cumulonimbus clouds. It was natural to determine whether the electrical disturbances associated with such locally ozone-producing occlusions were connected with them. However, many occlusions showing a typical increase in ozone are accompanied only by drizzle or light rain and have no effect on the course of the barometer, so that thunderstorm conditions are unlikely.

Very rarely, a marked increase in the amount of ozone, similar to that in an occlusion, was observed in perfectly clear weather, when there were no signs of a front or thunderstorm conditions that could have caused this increase. A typical example is given in Table 3, where all measurements, except the first two and the last, were made in direct solar radiation. The observations took place

near the center of the anticyclone. In the afternoon powerful cumuliform clouds formed, but certainly only after the maximum ozone values.

Table 3

Finally, it was established that a very large increase in ozone content can occur during thunderstorms. Fig. 15 gives an example, when two thunderstorms passed one after the other directly over the head and caused a very large increase in the ozone content in the atmosphere.

Fig. 15. Large increase in the amount of ozone during a thunderstorm on July 13, 1941.

It must be remembered that such measurements have to be made under cloudy skies and their accuracy is much lower than under direct sunlight. However, the influence of error probably more often decreases than increases the result of the measurements. Most of the ozone formed during a thunderstorm is probably within the limits of the thunderclouds, and not in the stratosphere. In this case it should not play

plays no role in the radiation balance within the stratosphere. Further, it was found that cumulus and partly stratocumulus clouds, although not accompanied by lightning, usually cause a high ozone content; when considering changes in stratospheric temperature caused by changes in the observed amounts of ozone, observations that have similarly experienced local influences of electrical discharges must be excluded.

In an earlier published work one of us gave a diagram showing the mean distribution of ozone around a typical cyclone and anticyclone. Further observations made it possible to present the distribution of ozone separately for a typical depression having a warm sector and for a depression that is completely occluded and has only cold air associated with it. Fig. 16 shows these results. In the case of an occluded depression there are no areas where the amount of ozone would be less than the seasonal norm, and the maximum of ozone is located in front of the center of the depression, and not behind it, as is the case for a depression with a warm sector.

Fig. 16. Composite diagram showing the distribution of ozone around a typical young cyclone with a well-defined warm sector and around an old occluded cyclone.

c) Carbon dioxide

The concentration of carbon dioxide in unpolluted air at all points on the globe where measurements have been made is about 0.03% by volume (Planet, 1939; Callendar, 1938). There is every reason to suppose that throughout the entire thickness of the troposphere, where turbulence is so strong, the concentration of this gas is the same. Analysis of air samples taken from the stratosphere, carried out by Glückauf (1944), showed that there are no differences in the concentration of carbon dioxide

in the stratosphere and near the ground. Measurements of the ratio of the concentrations of helium and oxygen in the atmosphere above 20 km have shown that at these altitudes there is sufficiently strong mixing of the air to prevent any noticeable separation of the gases by gravity, so that the only thing that can be assumed about the relative concentration of carbon dioxide is that it remains unchanged at these great altitudes.

The hygrometer based on the frost point and described in paragraph II, (a), can also be used for measurements of the concentration of carbon dioxide in the air. If a thimble is rapidly cooled in dry air such as is found in the stratosphere to a very low temperature, then the amount of ice deposited on it will be small and will not prevent the point of deposition of solid carbon dioxide from being clearly observed. The concentration of carbon dioxide in the stratosphere is much greater than the concentration of water vapor, and the deposition point can be determined very accurately. Up to the present time only a few, and rather rough, measurements of the concentration of carbon dioxide in the stratosphere have been made; these measurements show that the concentration is about 0.03% by volume, as indeed was to be expected.

In Part III below it is assumed throughout that the concentration of carbon dioxide remains constant in all parts of the atmosphere.

III. METEOROLOGICAL CONSIDERATION OF THE QUESTION

In the preceding sections those instruments have been described by means of which accurate measurements can be made of the amount of three important gases—water vapor, carbon dioxide, and ozone—at all altitudes of the atmosphere that can be reached by multi-seat aircraft. There exist sufficient scientific personnel for the further conduct of such work, and therefore the chosen path appears clear: it will lead to a considerable increase in our knowledge in the area under discussion. There is no doubt that it is of great scientific interest and, perhaps, will acquire practical significance for meteorology.

1. Relative significance of the various gases in the stratosphere

Since the absorption produced by a given mass of gas depends on pressure and temperature, measurements of the absorption coefficient for infrared radiation are necessary precisely under those conditions which prevail in the high layers of the atmosphere. Usually absorption changes in proportion to the square root of the total pressure, but Strong (1941) showed that absorption within the 9.7 μ band of ozone changes in proportion to the fourth root of the total pressure. Furthermore, when radiation corresponding to a certain absorption band passes through

medium that equally absorbs all wavelengths of this band, then the attenuation of radiation, of course, follows the law \(e^{-kx}\) (where \(k\) is the absorption coefficient of the medium and \(x\) is the mass traversed by the ray). However, if in reality the absorption band consists of a number of separate lines, then it is clear that this law is not fulfilled. In those cases where the absorption is so small that even at the center of a line it is not complete, the partial absorption may be considered proportional to \(x\); when the absorption is greater than this, but the lines of which the band consists are separated from one another by distances greater than their widths (usually in those cases where the absorption amounts to from 10 to 50%), the partial absorption is approximately proportional to \(x^{\frac{1}{2}}\); in those cases where the lines overlap one another and the absorption is large, the partial absorption changes more slowly than \(x^{\frac{1}{2}}\) (Cooling, 1943). All this shows that at present it is not possible to give exact quantitative results, and one must confine oneself to indicating general qualitative conclusions.

Fig. 17. Approximate distribution of energy in the spectrum of radiation coming from below through the lower layers of the stratosphere.

In considering the role played in questions of stratospheric meteorology by water vapor, carbon dioxide, and ozone, it is necessary to try to estimate the amount of energy absorbed by each of these gases. Simpson (1928) gave an estimate of the amount of terrestrial radiation reaching the upper part of the stratosphere and of its spectral composition. On his advice we modified these results, obtaining an estimate of the radiation reaching the lower surface of the stratosphere. This result is represented by the curve in Fig. 17, drawn with a thick line: it gives the energy for various wavelengths from which one must proceed in calculating absorption by gases in the stratosphere. Each gas will be considered in turn.

Ozone. Observations have reliably established the fact of absorption by ozone in the region between 9.3 and \(10.2\mu\). Adel and Lampland (1940), Adel (1942), and Strong (1939) investigated the absorption of solar rays in this band and found that at the center of the band there is absorbed

50–60% of the incident energy. Therefore, for the outgoing diffuse radiation, for which the effective path length must be approximately doubled, the absorption will amount to 70–80%. The hatched area between 9 and 10 μ in Fig. 17 represents the total amount of energy which, it may be assumed, is absorbed by atmospheric ozone from the flux of outgoing terrestrial radiation. In order to obtain the amount of energy absorbed by ozone at some level, it is necessary to take into account both the absorptive power of ozone at that level and the intensity of the radiation passing through it. To estimate the latter, one must include the absorption and re-emission of radiation by ozone both in the higher layers and in the lower ones. Table 4 gives values of the absorptive power of ozone at various levels, calculated on the assumption that it varies in proportion to the square root of the ozone concentration and to the fourth root of the total pressure. This table clearly

Table 4

Absorptive power of ozone at various levels

shows that the greater part of the energy of wavelengths from 9 μ to 10 μ is absorbed in the lower layers of the stratosphere.

Ozone located in the upper warm region—say, at about 50 km—also emits energy in the 9.7 μ band, and part of this energy is absorbed by the ozone located below; but the amount of ozone at these great heights is so small that the amount of emitted energy is of no serious significance. The same applies to carbon dioxide and water vapor in the upper warm region; their concentrations are so small that the energy emitted by them is still less substantial than in the case of ozone.

Carbon dioxide. Simpson (1928) estimated the energy absorbed in the 15 μ band by carbon dioxide in the stratosphere. He showed that, for parallel-directed radiation, the absorption at the center of the band is about 90%, so that diffuse terrestrial radiation of this wavelength is almost entirely absorbed in the stratosphere. Elsasser (1942) likewise came to the conclusion that almost all radiation coming from the Earth within the 15 μ band is absorbed by carbon dioxide in the stratosphere. In Fig. 17, the hatched area at 15 μ approximately shows the amount of energy absorbed by carbon dioxide in the stratosphere. Since the total pressure and the total amount of carbon dioxide decrease with height, the greater part of the absorbed energy belongs to the lower layers of the stratosphere.

From Fig. 17 it is evident that the total amount of energy absorbed in the stratosphere by ozone and carbon dioxide is, roughly speaking, the same.

Water vapor. It represents a much more difficult problem. Many different circumstances indicate that in reality the absorption is considerably less than was at one time supposed, namely:

1) The actual humidity, measured on aircraft and described in paragraph II, shows that the amount of water vapor is much smaller than had been expected.

2) Adel (1942) showed that in the spectra of atmospheric absorption there is a fairly transparent region beginning already at wavelengths greater than 16 μ; he obtained the solar spectrum over an interval up to 24 μ.

3) The effect of low pressure and low temperature in the stratosphere reduces the absorption by a given amount of water vapor.

Elsasser (1942) concluded that at those heights (they belong to the region of the stratosphere) where carbon dioxide radiates into space (the absorption by the overlying carbon dioxide is small), the radiative action of carbon dioxide is equal to or exceeds its magnitude for water vapor.

The assumption is not without foundation that the energy absorbed by water vapor in the stratosphere from the outgoing terrestrial radiation is approximately equal to the energy absorbed by carbon dioxide or ozone. In Fig. 17 it is indicated by broad sections shaded with thick vertical lines, which may roughly represent absorption by water vapor.

Although more accurate data are required for this, nevertheless the supposition, put forward as early as by Lord Chervell (Lindemann, 1919), is not unfounded: that water vapor, carbon dioxide, and ozone are approximately equally important in the radiation balance in the stratosphere, and that the actual equilibrium temperature depends on the relative concentration of these three gases.

There is one circumstance that should never be forgotten, namely that the temperature of the stratosphere can change as a result of absorption and emission only very slowly. The total amount of energy coming from the earth through the stratosphere may be taken, on the average, as approximately 0.3 cal/cm² min, and the total energy absorbed by both carbon dioxide and ozone is hardly more than 5% of this quantity (we suppose that the energy absorbed by water vapor is also of approximately the same magnitude). Even if we assume that all this radiation is absorbed in the layer between 10 and 15 km and that no energy losses occur by radiation, the temperature of this layer will rise only by about 0°.6 per day. The actual changes caused by an increase or decrease in absorption must be still much smaller, so that very small thermodynamic

changes in temperature produced by ascending or descending currents may give an effect of comparable magnitude.

Below we consider how the observed changes, under the conditions existing in the stratosphere, can be explained by a change in the relative amount of the three absorbing gases. The concentration of carbon dioxide probably changes very little, and without particular danger one may assume that it remains constant.

Since too little is known about changes in the amount of water vapor under various conditions, it has to be neglected, although significant errors may arise because of this. In any case, the following phenomena are consistent with the hypothesis that an increase in the concentration of ozone causes an increase in temperature in the stratosphere and vice versa. (It should be noted that percentage changes in ozone concentration in the lower layers of the stratosphere are approximately twice as large as the percentage changes in total ozone.)

2) Annual changes in the temperature of the stratosphere

Fig. 18 shows the annual changes in air temperature in the stratosphere and in the upper and lower troposphere. The data were obtained from more than 2000 ascents made at ten stations at latitudes between \(43^\circ\) N and \(68^\circ\) N. If the stations are then divided into three groups according to latitude, then in each group we find identical characteristics. This gives confidence that they are indeed representative. The earth’s surface is heated by solar radiation, and it should be expected that the maximum and minimum of the temperature of the earth’s surface must occur after the solstice. The lower troposphere is heated mainly from the earth’s surface by convection and condensation of water vapor, and the maximum and minimum of its temperature occur in accordance with the thermal regime of the earth’s surface, with some lag. Moving another step higher, we find that the upper troposphere is heated by the lower troposphere, and again there is a further lag in phase. However, if we take the next step upward, into the lower stratosphere, we find a great difference: the maximum and minimum of temperature are shifted back, toward the time of the solstices. It would seem that there is every reason to think that the lower stratosphere

Fig. 18. Annual changes of temperature in the troposphere and stratosphere.

is heated chiefly by absorption of radiation coming from the troposphere and from the earth’s surface, and this radiation has its maximum and minimum long after the solstices. If the absorbing capacity of the air in the stratosphere remains constant, then one would expect that we should find the maximum and minimum of temperature shifted toward later months relative to the maximum and minimum in the troposphere. If, however, ozone plays an essential role in the absorption of infrared radiation in the stratosphere, then one should expect precisely such annual changes of temperature as are actually observed. Since ozone has a maximum in spring and a minimum in autumn, the amount of ozone, say, in the month before the summer solstice must be greater than in the month after it, and this raises the temperature of the stratosphere before the solstice in comparison with the temperature after the solstice.

This question may be interpreted in various ways. The temperature of the troposphere at the time of the spring equinox is approximately 8°C lower than at the time of the autumn equinox, so that one might expect that the equilibrium temperature of the stratosphere in spring would also be 8°C lower than in autumn (that part of the radiation going into the upper layers of the atmosphere which comes from the earth’s surface must likewise be appreciably smaller in spring than in autumn). In reality, however, the temperature of the stratosphere in spring and autumn is almost exactly the same. In temperate latitudes the total ozone content in a vertical column of air is approximately 0.08 cm greater in spring than in autumn; thus, neglecting changes in the concentration of water vapor, precisely this circumstance appears to be the cause of the increase in the temperature of the lower layers of the stratosphere by approximately 8°C; in other words, 0.010 cm of ozone raises the temperature by 1°C (of course, the greater part of this amount of ozone is located in higher layers than the lower stratosphere and does not take part in heating it).

3) Annual variations in tropopause height

The tropopause has its lowest height in spring and its greatest in autumn. This can be explained immediately if one accepts the hypothesis that the tropopause is a transition layer between the upper region, where the temperature is determined by radiative equilibrium, and the lower region, where the temperature is determined by the temperature of the earth’s surface and by the transfer of heat upward through turbulence, etc.

If there were no ozone at all in the stratosphere and its absorbing capacity remained unchanged throughout the year, then annual variations of temperature would occur with some lag relative to the annual variations of temperature in the upper troposphere, just as the latter lag behind the annual variations of temperature in the lower troposphere, etc. This would lead to such annual variations of tropopause height that the maximum height would have

…would occur in summer, and the minimum in winter. If, however, the presence of ozone leads to a maximum and minimum of stratospheric temperature at the time of the solstices, then it should be expected that the height of the tropopause will be lowest in spring and highest in autumn, as is in fact observed.

4) Latitudinal changes in stratospheric temperature and tropopause height

The relatively large ozone content in the atmosphere that occurs at high latitudes should, from the standpoint of the hypothesis developed here, produce the observed excess of the stratospheric temperature in this region as compared with the temperature of the stratosphere near the equator. On the average for the year, the total amount of ozone at latitude \(60—70^\circ\text{N}\) exceeds the amount of ozone over the equator by \(0.130\ \text{cm}\), while the stratospheric temperature is higher by \(30^\circ\text{C}\). This indicates (if one disregards the influence of differences in moisture content, which still remain unknown) that an increase in the amount of ozone by \(0.004\ \text{cm}\) leads to an increase in temperature by \(1^\circ\text{C}\). Considering the annual changes of temperature in temperate latitudes, we arrived at the conclusion that an increase in the amount of ozone by \(0.010\ \text{cm}\) should produce a rise of \(1^\circ\text{C}\). This agreement may be regarded as good, since the influence of many other factors was neglected. An increase in the temperature of the stratosphere should also lead to a decrease in the height of the tropopause at high latitudes.

5) Change of temperature with height in the stratosphere

The intensity of terrestrial radiation passing through the stratosphere should gradually decrease as height increases, owing to the back radiation downward of part of the absorbed energy. If the composition of the stratosphere were the same at all heights, one would expect a gradual decrease of temperature with height. In reality, however, with the exception only of very high layers, the temperature remains constant at different heights or even increases slightly with height*), as was shown by Dines (1928) and Samuels (1929). It is known that the absolute concentration of carbon dioxide and water vapour should decrease with height, while the concentration of ozone increases noticeably. This change in the relative amounts of the indicated absorbing gases may cause an increase of temperature in the high layers, so that the absence of a decrease of temperature with height should not seem surprising. It is possible that already at high—

*) There is no doubt that the observed rise of temperature with height is not caused by insufficient ventilation of the thermometers or by heating by direct solar rays at high altitudes. The increase in temperature is observed not only by day but also at night, so that there is every reason to regard this effect as real.

at an altitude of about 20 km the absorption of ultraviolet solar rays by ozone can also produce some heating, although, of course, to a much greater extent heating by this means must occur in layers lying 15–20 km higher.

It is often suggested that the difference in stratospheric temperature in low and high latitudes is due to processes of general circulation between the equator and the poles, which cause a slow ascent of air in the stratosphere near the equator and a slow descent near the poles. It is difficult to imagine a general circulation of such a kind in which the air at a height of several kilometers above the tropopause in low latitudes would not undergo a much greater ascent (and as a consequence would not be cooled much more) than the air situated directly above the tropopause. The fact that the fall of temperature does not continue to great heights in such ascending currents indicates that some other effect is predominant. Such a predominant effect can only be radiational, and if it is necessary to assume that radiation does not play any significant role, then there are no grounds for rejecting the supposition that the difference in the observed amount of ozone between the equator and the poles is the cause of the temperature differences, without postulating the existence of ascending and descending currents.

b) Changes in ozone content and meteorological conditions

The observed changes in the amount of ozone in the high layers of the air are sometimes large and rapid. The frequently observed connection of these changes with the passage of “fronts” indicates that changes in ozone are often caused by a change of “air masses.” It seems beyond doubt that the ozone content in a given air mass undergoes only slow changes when the air mass is displaced from one latitude to another, provided there are no other local and rapid changes. The fact that the temperature of the stratosphere, owing to absorption and emission of radiation, can undergo only slow changes indicates that the frequently observed considerable changes in stratospheric temperature are in fact caused by a change of air masses. The general picture appears to be that changes in the relative concentration of ozone and water vapor determine slow temperature changes extending over large regions, while rapid changes both in the amount of ozone and in temperature, observed at any given place, are due chiefly to a change of air masses.

In conclusion, one may consider three fundamental questions posed by Simpson at the end of his paper “Further Studies of Terrestrial Radiation,” namely:

a) Why does the temperature of the stratosphere not decrease with height?

b) Why does the temperature of the stratosphere increase in passing from low latitudes to high ones?

c) Why is the beginning of the stratosphere over the equator situated higher than over the polar regions?

We may try to answer these questions as follows:

a) According to the assumption put forward by Simpson himself, because the ratio of the amount of water vapor to ozone decreases with height in the stratosphere, which compensates for the slight decrease in the flux of long-wave radiation in the upper layers.

b) Because, for reasons not yet entirely clear, the amount of ozone and, possibly, the ratio of the amount of ozone to the amount of water vapor is greater over the polar regions than over the equator.

c) Because the temperature distribution caused by turbulent mixing of the air extends over the equator to greater heights, until that temperature is reached at which the air is in radiative equilibrium.

Measurements of humidity in the upper layers of the air were carried out as part of the work of the Meteorological Research Flight of the Meteorological Office, and we express our gratitude to the Director of the Meteorological Office for permission to publish this part of the work. We are also very grateful to the Officer Commanding, Aeroplane and Armament Experimental Establishment, for his constant assistance, and to the Air Force pilots and observers by whom the measurements were performed.

The ozone measurements at Lerwick and Eskdalemuir were made at the observatories of the Meteorological Office, and the measurements at Valentia at the observatory of the Meteorological Service in Éire; we offer our thanks to the staff of these observatories for their assistance and cooperation.

All the work was carried out in close contact with the Meteorological Office, and we express our special gratitude to the Director for his constant assistance and interest.

References Cited

Adel, Astrophys. Journ. (2), 94, 451 (1942).
Adel and Lampland, Astrophys. Journ., 91, 1 (1940).
Burton and Oliver, Proc. Roy. Soc. A, 153, 166 (1936).
Callender, Quart. Journ. Roy. Met. Soc., 64, 223 (1938).
Cowling, Rep. Progr. Phys., 1942/43, 29 (1943).
Dines, Mem. Roy. Met. Soc. (18), 2, 137 (1928).
Elsasser, Harv. Met. Stud., No. 6 (1942).
Gluckauf, Nature, 20 May, p. 620 (1944).
Lindemann, Phil. Mag., 38, 669 (1919).
Paneth, Quart. Journ. Roy. Met. Soc., 65, 304 (1939).
Regener, Phys. Zschr., 35, 788 (1934).
Samuels, Mon. Weath. Rev., September, p. 382, Washington (1929).
Simpson, Mem. Roy. Met. Soc., 3, 21 (1928).
Strong, Journ. Opt. Soc. Amer., December, 29, 520 (1939).
Strong, Journ. Franklin Inst., No. 2, 1 (1941).