NEW DATA ON TWILIGHT LUMINESCENCE OF ATMOSPHERIC SODIUM

The discovery of the presence of considerable quantities of sodium in the high layers of the atmosphere (60–100 km above sea level) immediately raised the question of the ways in which it penetrates into these layers. Thus, studies of the distribution of sodium with height proved to be most closely connected with the general problem of the structure and dynamics of the atmosphere, which in recent times has been acquiring ever greater importance. Among the various methods for determining the concentration of sodium at various heights, a special place is occupied by studies of its twilight luminescence, discovered in 1937 by M. F. Vuk and B. I. Chernyaev¹. Measurements of the intensity of the sodium \(D\)-line, present in the spectrum of the glow of the twilight sky, as a function of the zenith distance of the Sun, have already made it possible to draw a number of substantial conclusions². The reviewed work, carried out in the autumn of 1948 by S. F. Rodionov and E. N. Pavlova³, is devoted to the further development of these studies and contains new, very significant results.

Unlike the preceding measurements, which were carried out by methods of photographic photometry with the aid of low-dispersion luminous spectrographs, S. F. Rodionov and E. N. Pavlova used a photoelectric photometer and a monochromator with dispersion in the region of 5900 Å equal to 135 Å/mm. The light receiver was a photoelectron multiplier with a cesium cathode of the L. A. Kubetskii system. The measurements were made during evening twilight at an altitude of 2200 m (Adyl-Su, Caucasus), the instrument being directed at an angle of 30° to the horizon toward the south or southwest. The dependence of the photocurrent was measured directly every 4.3 Å in the interval of wavelengths from 5825 to 5929 Å at various immersions of the Sun below the horizon. Measurement of one such spectrum took from 1 to 2 minutes. Figure 1 gives the results of measurements for one evening. All the curves cover a time interval of about 20 minutes.

On curves 1, 2, 3 the \(D\)-line of sodium emission is quite clearly visible, the doublet remaining unresolved because of the large width of the monochromator slit (0.12 mm). The course of the intensity of this line as a function of the zenith distance of the Sun reproduces in general outline the data of other authors², and the absolute value of the intensity is in satisfactory agreement with the data of Brückner and Kastler⁴.

What is substantially new is that, at zenith distances smaller than 95°, the emission line is replaced by an absorption line, which again disappears at zenith distances smaller than 92°. The authors

emphasize that here we are dealing with a regularly observed phenomenon. Fig. 2 gives a summary curve of all measurements of the intensity of the sodium \(D\)-line (the line intensity minus the intensity of the background of scattered sunlight in the adjacent regions of the spectrum) at various zenith distances of the Sun.

Fig. 1. Spectra of the twilight sky at various zenith distances of the Sun.

The authors adduce very convincing arguments in favor of the view that the absorption line observed at zenith distances from \(92^\circ\) to \(95^\circ\) is due to atmospheric sodium.

Against identifying it with a Fraunhofer line there is evidence both in the fact that it is not observed during the day and in the fact that, with the slit width used by the authors, a Fraunhofer line should not be observed because of its low intensity. In exactly the same way, the water-vapor absorption band (6000–5850 Å) under the experimental conditions should have had a width of about 150 Å, whereas the width of the observed absorption line is only 9 Å.

Fig. 2. Intensity of the atmospheric sodium \(D\)-line as a function of the zenith distance of the Sun.

The authors seek an explanation of the observed absorption line in the quenching of the “resonance fluorescence” of Na, formed as a result of the dissociation of NaCl molecules (of marine origin) in the lower layers of the atmosphere, and regard it as weighty evidence in favor of the presence of sodium at altitudes below 60 km. According to their hypothesis, NaCl located at altitudes of 20–40 km dissociates under the action of solar radiation \((\lambda < 2400\ \text{Å})\), and the appearance of the absorption line is due to the quenching of the resonance fluorescence of the liberated sodium atoms by nitrogen atoms. As the lower layers of the atmosphere are immersed in the Earth’s shadow and as a result of the cessation, in consequence of this, of the dissociation process, the free Na atoms rapidly disappear, recombining with chlorine, oxygen, etc. atoms. Since quenching processes can play a noticeable role only at altitudes not exceeding approximately 40 km, when the Earth’s shadow reaches these altitudes only the higher layers of the atmosphere begin to take part in forming the intensity of the \(D\)-line; there, the effect of reversal no longer takes place, and the sodium \(D\)-line appears as an emission line. This explanation, however, seems to us unlikely. I. A. Khvostikov\(^5\) succeeded in exciting the resonance fluorescence of sodium in the ground-

…layers of the atmosphere, up to altitudes on the order of 1000 m; moreover this fluorescence, proving the presence of Na at these altitudes, was detected precisely by the increase in the scattering ability of the air in the region of the sodium \(D\)-line. It is not difficult to see that this result, referring, it is true, to lower layers of the atmosphere, directly contradicts the authors’ basic assumption.

More probable, in our view, is another explanation, connected with the difference in the masses of air participating in the formation of the brightness of the twilight sky in different parts of the spectrum. Indeed, the presence of sodium in the lower layers of the atmosphere, proved by the experiments of I. A. Khvostikov, must cause a certain (and, judging from the data given in \(^{5}\), considerable) increase in the extinction coefficient in the region of the sodium \(D\)-line in comparison with neighboring spectral regions. As a result, there must arise a difference also in the effective height of the earth’s shadow, i.e. in the height of the lower boundary of the scattering volume for the \(D\)-line itself and for neighboring parts of the spectrum (compare, for example, \(^{4}\)). Since the density of the atmosphere decreases rapidly with height, owing to the indicated “dispersion” of the effective height of the earth’s shadow there must also arise a difference in the intensity of the light scattered by the atmosphere within and outside the limits of the \(D\)-line. This difference will be the greater, the greater the density gradient; and since the latter rapidly decreases with height, the difference in intensity must decrease as the Sun sinks below the horizon. On the other hand, this difference must vanish also at small angles of depression, owing to the decrease of the shielding thickness of the lower atmospheric layers (see, for example, \(^{6}\)). The indicated effect of a change in the relative brightness of the twilight sky in different parts of the spectrum as a result of the “dispersion” of the effective heights of the earth’s shadow, caused by differences in the extinction coefficients, may reach a noticeable magnitude and is apparently to a considerable extent responsible for the color of the twilight sky and its changes in the course of twilight at solar depressions of less than \(10^\circ\).

From this point of view, the reversal of the \(D\)-line of atmospheric sodium observed by the authors may be only apparent and connected with a decrease in the intensity of the background of scattered sunlight within the line in comparison with the background outside it.

In favor of the fact that here we are dealing with an effect of this kind is also the presence, not noted by the authors, of a characteristic bending of curves 5 and 6 (Fig. 1) in the wavelength region 5860–5880 Å, which disappears both at larger and at smaller zenith distances.

In any case, the direct determination of the intensity of the \(D\)-line as the difference between the intensities inside and outside the corresponding spectral interval, as has been done up to now and as the authors do, without a detailed analysis of all the factors influencing the formation of the brightness of the twilight sky, may lead to serious errors.

Thus, one of the essential results of the paper under review, it seems to us, is the clarification of the real necessity, in interpreting data on the twilight luminescence of sodium, of a detailed consideration of the whole picture of the formation of the brightness of the twilight sky.

The indicated phenomenon of “dispersion” of the effective heights of the earth’s shadow is much easier to take into account for the zenith than for other regions of the sky. Moreover, in that region of the sky where the measurements were made, multiple scattering, which greatly complicates the picture, must play an essential role. Therefore one can only regret that the very interesting and important results obtained by the authors do not refer to the zenith and, consequently, can hardly be made the basis for a detailed theoretical examination on the basis of the modern theory of twilight.

G. Rosenberg

References

1. V. I. Chernyaev and M. F. Vuks, DAN 14, 77 (1937).
2. I. A. Khvostikov, The Glow of the Night Sky, Publishing House of the Academy of Sciences of the USSR, 1948.
3. S. F. Rodionov and E. N. Pavlova, DAN 67, 251 (1949).
4. J. Bricard and A. Kastler, Ann. Geophysique 1, 1 (1944).
5. I. A. Khvostikov, UFN 36, 372 (1948).
6. N. M. Shtaude, Izv. AN SSSR, Ser. Geofiz. i Geogr. 11, 349 (1947).