The maintenance of the night ionosphere

Journalof Atmospheric andTerrestrinl Physics,1968,Vol. 30,pp. l%i-18%.

The maintenance

PcrgamonPress. Printediu NorthernIreland

of the night ionosphere

J. E. TITHERIDGE Radio

Research

(Received

Centre,

University

23 February;

of Auckland,

in revised form

Pu’ew Zealand

12 April

1968)

Abstract-The electron content of the ionosphere was recorded continuously at latitudes of 34”s and 42% for a period of 18 months. The results arc combined with ionosonde measurements to determine the mean changes in the night-time ionosphere at latitudes of 10 to 60”. The effective loss coefficient /3is constant at about, 4.10m5 see-l in summer, and 3.10V5 see-l in winter, throughout, most of the night. This agrees with the values observed at sunset when allowance is made for changes in the height and temperature of the P-layer. A nocturnal source of ionization is observed at all seasons. It acts for about 3 hr near midnight, at geomagnetic latitudes of 15 to 40” in summer and 25 to 50” in winter. Tho total influx is about 2.1016 electrons/m2, with an annual variation of 50 per cent and a solar cycle variation of 150 per cent. The influx increases by 50 per cent during periods of high magnetic activity. Calculations of the diffusion of ionization from tho exposure, assuming that there is no production or loss of protons by charge exchange, predict a nocturnal flux of 2.1016 electrons/m2 into the ionosphere at latitudes between 20 and 45O. This influx will extend to higher latitudes in winter variation

than in summer. It will have an annual of about 150 per cent.

variation

of 50 per cent and a solar cycle

1. INTRODUCTION HAS long been realised that the electron density in the ionosphere does not decrease throughout the night in the way predicted by simple theory. When the production of ionization ceases at sunset, the combined effects of recombination and diffusion should cause the electron density-height profile to approximate that of an DUNGEY, 1956; MARTYN, 1956). This stable K-Chapman layer (DUNCAN, 1956; distribution will be obtained about 1 hr after sunset. The ionized layer should then decay without change of shape, so that the effective loss coefficient @’ is the same at all heights. The value of ,6’ can be estimated from the changes which occur in the critical frequency of the F-layer after sunset. At medium latitudes this indicates a loss coefficient of about 4*10-5 set-l in summer, and 12*10-5 see-l in winter (RISHBETH, 1964). More recently it has been possible to make continuous observations of the t,otal electron content of the ionosphere, using signals from a geostationary satellite. ‘The changes occurring after sunset give an integrated loss coefficient of 5*10-5 see--l in summer and 12.10-5 see-l in winter at medium latitudes (TITHERIDGE, 1966). This agrees closely with the values determined for the peak of the layer, giving strong support to the idea of a stable layer decaying with the same effective loss coefficient at all heights. If this rate of decay persisted unchanged throughout the night, the ion density would be reduced by a factor of about 30 in 10 hr. However, the observed reduction, in peak density and in total content, is generally by a factor of 2 to 5. Some additional process must therefore be operating to maintain the night-time F-layer. The possible processes fall into four basic groups. IT

?

1557

1858

J. E. TITKERIDGE

(a) A reduction in the recombination rate. (b) Ionization diffusing down from the exosphere (or, in winter, ionization diffusing along the lines of force from the summer hemisphere). (c) Production of ionization during the night, possibly by corpuscular ra,diation. (d) Horizontal movements of ionization. A reduction in the recombination rate is most easily obtained by an increase in the height of the layer. This is considered in Section 2, where it is estimated that the effective night-time loss rate in equinox and winter will be about one quarter of the sunset value. This change cannot, however, account for the increases in electron content which are commonly observed during the night in winter (DA ROSA and SMITH, 1967). A night-time source of ionization, through process (b) or (c) above, is therefore required. Horizontal movements of ionization, across the magnetic field, are restricted to the circulation of complete tubes of force and can not readily produce a widespread increase in electron content. Increases in electron content, near midnight, normally occur only in winter. This suggests a seasonal change in the strength of the night-time source of ionization. The effect of an external source could, however, be masked by the higher densities and more rapid recombination in summer. To separate these effects, hourly values of total content were analysed assuming an exponential decay of ionization (TITHERIDGE, 196%~). This showed that a night-time source was present throughout the year at medium latitudes, although its strength may be reduced near the equinoxes. The present paper examines in more detail the seasonal and latitudinal variations in the strength and time of occurrence of the night-time source of ionization, and the changes with solar and magnetic activity. Continuous records of the total electron content of the ionosphere, obtained over a period of 18 months at latitudes of 34”s and 42”S, are used to determine the mean variation of electron content from sunset to sunrise in each season (Section 3). The results show a reasonable agreement with the night-time recombination coefficients calculated in Section 2. However most of the results clearly demonstrate the existence of a large source of ionization, acting for a few hours near midnight. The amount and duration of the In Section 4 the same analysis is applied to critical influx are readily obtained. frequency measurements, over a range of latitudes, to determine the extent of the night-time source and its seasonal and sunspot cycle variations. Finally these results are used to evaluate the relative importance of the different possible sources. 2. THE REDUCTION OF THE Loss

COEFFICIENT AFTER SUNSET

In the FZ-region of the ionosphere, ionization is removed by the process of dissociative recombination. This proceeds at a rate which is limited primarily by the number of nitrogen (or, possibly, oxygen) molecules available. The loss coefficient p at any height is therefore proportional to the N, density at that height. A reduction in the apparent loss rate during the night can be explained most readily by a decrease in the amount of molecular nitrogen in the F-region. The effects of changes in the temperature of the upper atmosphere have been considered by HAREIS and PREISTER (1962). Their calculations indicate a decrease in temperature

The maintenance

1859

of the night ionosphere

of about 10 per cent from just after sunset to early morning, at low and medium. levels of solar activity. This produces an increase in the density of nitrogen molecules at heights between about 120 and 180 km, with a decrease at greater heights At a height of 300 km, the calculated decrease is by a factor of about l-4. The loss coefficient @ at a fixed height in the F-region should therefore decrease by this amount. The direct effect of temperature changes on the reaction rates has been considered by THOMASand NORTON (1966). Theoretical calculations, and the agreement between ionospheric observations and laboratory measurements, suggest that the rate coefficients do not vary rapidly with the neutral gas temperature. Changes in the reaction rate, due to the small decrease in temperature during the night, are therefore unlikely to exceed 10 per cent. Much larger changes in the effective loss coefficient will be produced by changes in the height of the layer. Vertical movements of 100 km or more are not uncommon and are probably produced by the effect of horizontal winds in the neutral atmosphere (KING et al., 1967). The possibility of height changes caused by electrostatic forces has also been suggested by several workers, but present knowledge of these forces is inadequate for any realistic calculations. The changes actually occurring in the height of the FZ-layer can, however, be determined from ionosonde records. WRIGHVJ 1962) gives the heights obtained from a full analysis of records taken over a period of 1 yr, at a chain of stations in the northern hemisphere. These results were used to determine the mean height of the peak of the F-layer at 1 hr after sunset, and again at 2 a.m., and the results are shown in Fig. 1. There do not appear to be any large changes in the shape of the FZ-layer during the night, so these height changes represent an overall vertical movement of the ionosphere. The extent of this movement varies considerably with latitude, but the increase is always much less in summer. The mean height increase is 8 km in summer, 33 km in equinox and 29 km in winter. To estimate the average effect at medium latitudes it will therefore bc assumed that the height does not change appreciably in summer, while it increases by 30 km in equinox and winter. For an atmospheric temperature of 850”K, the scale height of molecular nitrogen is 30 km. The night-time increase in the height of the F-layer w-ill therefore cause a reduction in /3 by a factor of 2.7 in equinox and winter. Combining this with the reduction of l-4 in the molecular density at a fixed height, gives a net reduction in /? of about l-4 times in summer, and 4 times in equinox and winter. The values of /? measured near sunset (Section 1) therefore give the approximate night-time loss co&cients listed in Table 1. These deca,y rates are compared with the observed variations in the following section. Table 1. The effective loss coefficient of the P2-layer, Summer

Equinox

in 1O-5 s~c-~ Winter

Sunset (measured)

5

12

12

Night

4

3

3

(estimated)

J. E. TITHERIDGE

1860

i_‘,.)s”;sP

i

k E 0 ._

x400(b)

;

I

I

I

-

/ Equinox

/

.$380-

:

/

/

-\N, .

.

/

cd

_-

--

360340-

6

I

/\

36cT(c)

/\ I

I Winter

34

/

/

/

/

/

\

\

\

\

+’

’

10

I 20

/

/

1

,I

32T 300

/

/

I

30

40 Latitude,

1 50

degrees

Fig. 1. The mean height of the peak of the P-layer at 1 hr after sunset (solid lines) and at 2 a.m. (broken lines) at different latitudes in the northern hemisphere 3. THE NIGHT-TIME CHANGESIN TOTAL CONTENT 3.1 Experimental measurements. Continuous records of the electron content of the ionosphere were obtained by recording the polarization angle of the 137 MC/S signal from the geostationary satellite Syncom 3 (TITHERIDGE, 1966). The results have a long-term accuracy of a few per cent, and changes in electron content are dis.played with an accuracy of about 0.2 per cent. Measurements have been obtained at Auckland (lat. 37*O”S, long. 175aO’E) since June 1965, and at Invercargill (46*4’S, 168.3”E) since August 1965. These give the electron content of the ionosphere at latitudes of about 34”S and 42”s respectively. Syncom 3 was positioned over the Equator, at a point almost due north of New Zealand. The ray paths to Auckland and Invercargill are shown in Fig. 1. The path to Auckland is parallel to the magnetic field, to within a few degrees, at all heights between 200 and 2500 km. Most of the night-time ionizat’ion falls within this height range. The observations at Auckland therefore give a direct and reasonably accurate measure of the total electron content of the tube of force which crosses the Equator at a height of about 6400 km. The observations at Invercargill give, less accurately, the electron content of the tube of force crossing the Equator at a height of about 1100 km. Since ionization can move freely only in a direction parallel to the magnetic field, the measurements are almost unaffected by changes in the height, shape or thickness of the F-layer.

The maintenance

Fig.

1861

of the night ionosphere

2. The

cargill

dotted lines show the ray paths joining Auckland (A) and Inver(I) to the satellite Syncom 3. The continuous curves are dipole field lines for different values of Mcllwain’s parameter L.

00 18

20

22

00

02 Hours,

06

04 Local

Time

Fig. 3. The changes in total electron content of the ionosphere between sunset and sunrise at Auckland. Each curve is the average of 6 months observations. The broken lines were calculated using a fixed loss coefficient of 4*10e5 see-l in summer, and 3.10-s see-l in winter.

3.2 Mean variations for each season. Total electron content measurements from June 1965 to December 1966 have been combined to obtain the mean seasonal variations shown in Figs. 3 and 4. Each curve is the average of continuous records, from sunset to sunrise, on about 180 nights. The broken lines in Figs. 3 and 4 show the exponential decay of electron content calculated using the estimated night-time values of loss coefficient from Table 1.

J. E. TITHERIDQE

1862

16-

INVERCARGILL 42’S

(@=

47’)

6-

I

I

I

I

I

I

I

18

20

22

00

02

04

06

HOU?S,

Local

Tlmc

Fig. 4. The mean variation of total electron content from sunset to sunrise at Invercargill.

In summer, this provides a good fit to the observed variation in the 3 hr after sunset. The observed variation is also approximately exponential in the 3 hr before sunrise. Near midnight, however, some additional process acts to raise the values of total content from one decay curve to another. In equinox and winter the effective loss coefficient at sunset is about, 12*10V5 set-l. The loss rate decreases steadily after sunset, approaching the night-time value of 3*10-5 see-l (corresponding to the broken lines in Figs. 3 and 4:) a few hours before midnight. The observed variation for about 4 hr before layer sunrise also follows closely an exponential decay law with a loss coefficient of 3*10e5 s~c-~. It therefore appears that the loss coefficients in Table 1, estimated from the changes in the height of the P-layer, provide a reasonable approximation to the actual recombination rate during most of the night. A change of only 30 per cent in the assumed values of p gives a much poorer fit to the observations in all cases, so that the experimental curves define the value of ,5 to within 620 per cent. The changes in electron content near midnight cannot be explained by a simple loss mechanism. An injection of ionization is required, acting for a few hours near midnight, to raise the values of total content from one exponential decay curve to a second curve at a higher level. Because of the slow recombination of ion.ization in winter, this additional ionization is sufficient to cause an increase in total content.

The maintenance

1863

of the night ionosphere

In summer and equinox the larger electron densities cause a more rapid recombination. Night-time increases in total content are therefore not observed. Figures 3 and 4 show, however, that there is no real difference between summer and winter, and a night-time source of ionization is required at all seasons. This confirms the result obtained earlier by fitting a model to the values of total content measured on. individual nights (TITHERIDGE, 1968a). The present results also indicate that the night-time influx of ionization is reduced at the equinoxes. The loss of ionization by recombination will proceed throughout the night, at a rate which is proportional to the total content. The total amount of the additional. ionization injected near midnight can therefore be determined by the construction illustrated in Fig. 3. A vertical line (bc) is drawn between the two extrapolated decay curves, so that the areas A and B are equal. If the electron content varied along the path abed, the total amount of ionization lost by recombination would then be the same as for the actual variation along the curve ad. The total production must also be the same, since the end values are unchanged. The total amount of ionization produced by the external source is therefore given by the extent (I) of the vertical line bc. The values of I for each curve, and the approximate times during which the external source was acting, are listed in Table 2. Using the values of /l from Table 1, the maximum error in I is about O-3. 10la. However, the results also depend to some extent on the assumed values of /?. The loss rates required to fit the experimental curves are defined to within 20 per cent, giving a maximum overall error of 30-50 per cent in I. This error could be further increased if the loss coefficients change appreciably near midnight. Variations in p will also alter the apparent time of the influx by up to 30 min. Table 2. The total amount of ionization produced near midnight (in 1016 electrons/m2) and the local time during which most of the production occurred Latitude 34OS Ionization Duration Summer Equinox Winter

3.3 1.0 3.8

22.00-00.00 23.30-01.20 22.00-02.00

Latitude Ionization 1.0 0.8 2.2

42OS Duration 22.30-00.15 23.30-01.00 23.00-02.00

3.3 Changes with magnetic activity. The effect of magnetic activity on the mean variation of electron content during the night is shown in Figs. 5 and 6. The broken lines give the average of all days (from June 1965 to December 1966) when the mean value of K, between noon and midnight was 3 or more. The continuous lines give the average of all other days. The greater overall slope of the broken lines indicates an increase in the loss coefficient ,6 during the periods of high magnetic activity. This supports the result obtained earlier (TITHERIDGE and ANDREWS, 1967; TITHERIDGE, 1968a). The same type of irregular variation is observed near midnight under quiet and disturbed conditions. The curves before and after this irregularity were approximated by an exponential decay with a fixed loss coefficient p. To obtain a good fit,

186-l

J. E. TITHERIDGE

I 18

I

20

I

22

I

00

I 02 Hours,

I

I

04 Local

06 Time

Fig. 5. The variation of electron content from sunset to sunrise at Auckland. The continuous lines show the average variation when the mean value of K, (between noon and midnight) was less than 3. The broken lines are for K, equal to 3 or more.

L

I

18

I

20

I

22

I

00

I

02 HOUI-5,

I

I

04 Local

06 Time

Fig. 6. The mean variation of electron content at Invercargill tinuous lines) and K, > 3 (broken lines).

for ICO c 3 con-

The maintenance

lS6ii

of the night ionosphere

,!? had to be 30 per cent greater than the night-time values used previously (Table 1). The total amount of additional ionization required near midnight was then determined from the vertical separation of the exponential decay curves, and the resuhs are shown in Table 3. Table 3. The night-time loss coefficient (@, in 10e5 set-l) and the amount of ionization injected near midnight (I, in 1016 electrons/m2) under quiet and disturbed conditions Auckland (34”d) Summer Equinox 1Vinter R,

< 3

p I

K,>3P I

4 3.3 5 6.0

3 1.0 4 1.4

3 3.8 4 4.6

Invercargill (42”s) Equinox TVinter Summer 4 1.0 5 1.6

3 0.8 4 1.5

3 2.2 4 3.1

The night-time influx of ionization shows a definite dependence on magnetic activity, at all seasons. Under disturbed conditions (K, = 3 to 5) the total amount of the influx is always greater than for quiet conditions. The increase varies from 21 to 57 per cent, with a mean increase of 50 per cent. The time of the influx is generally unaltered, except for summer at Auckland when it appears to start muclt earlier under disturbed conditions. 4. THE EXTENT OF THE

NIGHT-TIME

SOURCE OF IONIZATION

4.1 Changes with latitude. Measurements of the critical frequency of the ionosphere at a number of stations were used to determine the range of latitudes over which the night-time source of ionization is present. Observations from January 1965 to December 1966 were combined to give 8 months for each season. At each hour, the 8 monthly median values of f,FZ were averaged and used to calculate the mean electron density at the peak of the F-layer. The results are shown in Fig. 7. Each set of three curves, on the same horizontal line, gives the change from sunset to sunrise in summer, equinox and winter at a given station. The broken lines for summer and winter in Fig. 7 show the exponential variations corresponding to the night-time loss coefficients of 4.10~” and 3*10-5 see-1 from Table 1. For equinox conditions, the loss coefficients had to be increased to 5.10-5 Such a large value does not agree with the total content see-l to fit the observations. measurements in Figs. 3 and 4. The difference is caused by an increase in the slab thickness of the ionosphere during the night in the equinoxes, so that the peak density decreases faster than the total content. At Port Moresby, the loss coeB’-cient also had to be increased in winter to fit the observations. This change could be predicted from Fig. 1, which shows that the night-time increase in the height of t’he F-layer (and the resulting decrease in p) does not occur at low latitudes. The night-time changes shown in Fig. 7 support the results obtained from the tot,al content measurements. Thus at 29’S there is a definite influx of ionization just after midnight in summer, and a more sustained influx for about two hours tither side of midnight in winter. This agrees well with the total content variations at 34’S (Fig. 3). Similarly the changes in peak density at Godley Head (449) shou

1866

J. E. TITHERIDGE

EQUINOX PORT

MORESBY

AAOUL

CAMPBELL

I.

I.

__

2-

Hours,

Local

Time

Fig. 7. The variation in the electron density at the peak of the F-layer, from sunset to sunrise. Each curve is the mean result from 8 months during 1965-1966 (except at Port Moresby, where each curve includes 6 months). C$is the geomagnetic latitude. The broken lines show the exponential decay obtained with a fixed loss coefficient /3. The value of 8, in lop5 see-l, is given by each curve.

the same seasonal variation in the occurrence and amplitude of the night-time influx as the total content measurements at 42”s (Fig. 4). This confirms that (in summer and winter) the ionization is decaying at approximately the same rate at all heights, so that the peak density plots of Fig. 7 provide a reliable indication of the overall changes in total content. At a latitude of 297‘3, Fig. 7 shows that the total amount of ionization injected near midnight is about the same at all seasons. (Note that the vertical scale is doubled for winter in Fig. 7.) At other latitudes, however, there is a large seasonal

The mrtintenance of the night ionosphere

1867

variation. An appreciable influx occurs only in summer near 2O”S, and only in winter near 45’s. At latitudes above 50”s there is no night-time influx of ionization. The time of the influx is approximately the same at all latitudes. The source of ionization acts from about 21.00 to 00.00 in summer, and 23.00 to 02.00 in winter. This is in reasonable agreement with the total content results of Table 2. At t,he equinoxes, the influx indicated by Fig. 7 occurs about 1 hr earlier than that given by the total content measurements. This is probably because the slab thickness of the ionosphere increases after midnight. The total content results will be more accurate, since they give the overall changes in a fixed tube of force and are comparThe latitudes and times at which there atively unaffected by vertical movements. is an appreciable influx of ionization during the night, as shown by Figs. 3, 4 and 7; are summarized in Table 4. Table 4. The approximate

range in geomagnetic and in local time, over which the night-time of ionization is important Local

Summer Equinox Winter

latitude, source

Time

Latitude

(hr)

15-409 20-45”s 25-50”s

21.30~00.00 22.30~01.00 22.30-02.00

4.2 Seasonal and anma. changes. A large influx of ionization occurs only in summer at latitudes of about 2O”S, and only in winter at 45S. To see whether this is a true seasonal effect, the mean variation in the electron density at the peak of the F-layer was determined at these two latitudes in the northern hemisphere (Fig. 8). The results show a definite seasonal change. Thus in summer, the nighttime influx of ionization is much stronger at 21”N than at 40”N. In winter, however: the night-time influx is much greater at 40”N. There is therefore a real seasonal variation in the position of the source, which moves closer to the Equator in summer in both hemispheres. At both latitudes in Fig. 8, the amount, of the night-time influx of ionization is greater in summer, and less in winter, than at corresponding latitudes in the Southern Hemisphere. There is therefore a world-wide annual variation in the strength of The total amount of the night-time influx, in the night-time source of ionization. both hemispheres, is 50 per cent greater near December than near June. This a,grees with the annual variation in the density of the daytime F-layer. Whistler measurements have shown a similar change in the ion density in the exosphere. which is greater in December than in June by about 20-50 per cent (CARPENTER., 1962) to 100 per cent (SMITH, 1961). 4.3 Sunspot cycle changes. The previous results all apply to conditions in the ascending phase of the sunspot cycle, at a mean sunspot number R of about 30. Corresponding plots for Rarotonga and Godley Head, at sunspot minimum and sunspot maximum, are given in Fig. 9. The broken lines show, as before, the changes which would be produced by an attachment-like loss process with a fixed loss coefficient /?. The night-time loss coefficients used previously (Table 1) give reasonable

1868

J’. E. TITHERIDGE

SUMMER

WINTER 21”

N 4-

._ 4-

Fig. 8. Changes in the electron density at the peak of the P-layer at Maui (21 “N) and at Boulder (40”N). Each curve is the mean result from 4 months during 1966.

WINTER

agreement throughout the sunspot cycle except at Rarotonga, where the loss coefficient is increased to 5*10-5 see-l at sunspot maximum in winter. Figure 9 shows that the night-time influx of ionization follows the same general pattern throughout the solar cycle. The influx occurs primarily during the summer at 2l”S, and during the winter at 44%. The magnitude of the influx is, however, increased by a factor of 2.5 (f0.5) at sunspot maximum. This apparent change, obtained from the peak density plots in Fig. 9, could be affected by changes in the slab thickness of the ionosphere. Preliminary results at Auckland suggest, however, that the increasing temperatures near sunspot maximum are offset by an increase in the mean ionic mass in the topside ionosphere, so that there is no large ch.ange in slab thickness. If a linear variation with sunspot number is assumed, the strength of the nighttime source is therefore proportional to 1 + 0.01 R. This variation is less than t’hat observed in the mean density of the daytime I/‘-layer, which is proportional to 1 + 0.02 R. The changes in the electron density in the magnetosphere are much smaller. Results given by CARPENTER (1962) for 1958 and 1961 indicate a mean variation proportional to 1 + 0.004 R. The solar cycle changes in the magnitude of the night-time influx are therefore midway between the changes observed at the peak of the ionosphere and the changes occurring at a height of several earth radii. 5. THE NIGHT-TIME SOURCE OF IONIZATION It, has been shown that, even when allowance is made for a reduction in the loss coefficient after sunset, the night-time changes in the electron content of the midlatitude ionosphere cannot be explained by simple recombination. Some source

The maintenance

of the night ionosphere

1869

Fig. 9. Night-time changes in the electron density at the peak of the ionosphere near sunspot maximum (1959) and sunspot minimum (1964). Each curve gives the mean result from 4 months observations at Rarotonga (21%) or Godley Head (44%).

of ionization is required. This is most evident in winter when, because of the slower recombination, the external source commonly causes an increase in electron content. When allowance is made for the normal loss processes it is found, however, that an influx of ionization occurs at all seasons. The total influx is typically about 2.1016 electrons/m2 (Table 2). It acts for about 3 hr near midnight, so that the effective production rate is about 2~10~~electrons/mz/sec. The influx occurs at geomagnetic latitudes of 15 to 40” in summer, moving to 23 to 50” in winter. The magnitude of the influx shows an annual variation, being about 50 per cent greater in December than in June. It is also increased by about 50 per cent during periods of high magnetic activity, and shows a solar cycle variation of about 2.5: 1. In the following sections the possible sources of ionization are considered in relation to these requirements. 6. CORPUWULAR IOXIZATION There is little evidence for any night-time source of electromagnetic radiation, of sufficient strength to produce an appreciable amount of ionization. The presence of a corpuscular source is, however, suggested by the following observations. 6.1 The electron te,mperature in the night-Gne F-region. In the absence of a nighttime source of energy, the electron temperature T, and the ion temperature Ti should equalize within a few minutes. Backscatter measurements show, however, that T, remains somewhat greater than Ti through the night (EVANS, 1965a). The heat flux required to maintain the observed difference was estimated to be about 4*10P5 ergs/m3/sec. For a vertical scale of 250 km, this gives a total heat input of about 10 ergs/m2/seo.

1870

J-. E. TITHERIDGE

Measurements of electron temperature by the satellite Ariel 1 (WILLMORE, 1964) also showed the presence of a source of energy during the night, with a total heat input of about 2 ergs/m2/sec. Results from Explorer 17 (BRACE and REDDY, 1965) indicate a night-time heat input of about 10 ergs/m2/sec, agreeing with the backscatter estimate. It was suggested that this heat could be produced by a flux of soft electrons (about 100 eV) with a total energy of about 100 ergs/m2/sec. NATHAN (1966) calculated the heating effects of trapped electrons with an energy of a few keV. An energy flux of about lo3 ergs/m2/sec was required by the Ariel measurements, and five times this by the Explorer 17 results. Heating occurred mainly through the ionization of atmospheric atoms, the secondary electrons produced having a mean energy of about 7 eV or lo-l1 ergs. For a heat input of 10 ergs/m2/sec this implies a production of 1012 electrons/m2/sec in the upper ionosphere. This agrees well with the prod’uct,ion of 2*1012 electrons/m*,‘sec required to explain the changes in total electron content. 6.2 Nightglow observations. Recent observations show t,hat N,+ radiation (3914 A) is a normal feature of the night airglow at medium latitudes. Since the lifetime of N,+ is 10 see or less throughout the atmosphere, this requires a production of ionization during the night. Measurements of N,+ radiation reviewed by DALGARKO (1964) indicate a zenith intensity of some tens of Rayleighs at medium latitudes. YANO (1966) observed about 10 Rayleighs. A rocket measurement at midnight showed a zenith intensity of 5R at heights above 85 km, with a flnther 5R below this height (O’BRIEN et al., 1965). The glow below 85 km was attributed to cosmic radiation, and the ionization produced will disappear rapidly. The 6R observed above 55 km corresponds to a production rate of 3.1012 ions/m2/sec, and a heat, input of 100 ergs/m2/sec. The source of t,his energy is not known. It is consistent with the flux of soft electrons suggested by BRACE and REDDY (1965) to explain t’he night-time heating. PRAG et al. (1966) considered nightglow excitation by trapped protons (about 4 keV). This can explain the maintenance of the night ionosphere below 200 km, but has little effect, above 250 km. 6.3 Relation to the present results. The observed heating effects in the night-time ionosphere do not prove the existence of a source of ionization, since the heat could be conducted down from the exosphere (EVANS, 1965b). The N,+ radiation does, however, require the production of 3.1012 ions/m2/sec in the lower F-region. An extension of this production to greater heights could produce the 2*1012electrons/m2,i set required in the upper P-region. The correspondence would be strengt,hened if it could be shown that the N,+ radiation occurs mainly during a few hours near midnight. Present measurements, however, are obtained by integration otter most of the night. Xteasurements near the South Atlantic anomaly indicate that trapped radiation causes an increase in the total elect,ron content of the ionosphere in this area (&NDONCA, 1964). Smaller effects might therefore be expected at other places. The N,+ glow and the night-time heating effects can both be explained by the effects of trapped electrons with an energy of a few keV (NATHAN, 1966). It is shown above that these particles will also produce about 10.I2 elect,rons/m2/sec in the F-region. This is sufficient to explain most of the changes in the night-time ionosphere. The latitude range (15-50’ geomagnetic) of t,hese changes, the sunspot cycle T-ariation

The maintenance of the night ionosphere

1871

of 2.5 : 1 and the increase of 50 per cent during periods of increased magnetic activity are not inconsistent with this source of ionization. The diurnal and seasonal changes in the night-time production of ionization are more difficult to explain. Figures 3 and 4 show that production occurs for only a This might result from electrostatic fields lowering the few hours near midnight. mirror points of the trapped electrons at this time. Similarly, the seasonal changes could be caused by a longitudinal electric field, which raises the mirror points in one hemisphere and lowers them in the other (CATCHPOOLE, 1967). However the heating effects are observed throughout the night, and are not concentrated near midnight in the same way as the electron content effects. This represents a major difficulty in attempts to explain the night-time production of ionization in terms of corpuscular radiation. 7. DIFFUSION OF IONIZATION FROM THE EXOSPHERE Coupling between the ionosphere and the exosphere can be used to explain both the heating of the night-time ionosphere and the changes in electron content. As a heat source, it has been estimated that the exosphere could supply the 2 ergs/m2/sec required by the Ariel measurements, but not the 10 ergs/m2/sec required by the Explorer 17 results (GEISLER and BOWHILL, 1965; EVANS 1965b). The electron content of the exosphere is sufficient to provide the night-time influx of ionization at latitudes above 40” (EVANS, 1965a; DA ROSA and SMITH, 1967). The transfer of ionization between the ionosphere and exosphere is, however, restricted by the slow rate of diffusion of protons through oxygen ions (HANSON and ORTENBURGER, 1961). Under steady state conditions, the vertical movement of ionization is severely limited by this diffusive barrier, at a height of 500 to 1200 km. HANSON and PATTERSON (1963, 1964) and GEISLER (1967) showed that the upwards flow of ionization during the day would be less than loll electrons/m2/sec. For charge conservation the same limit must apply to the downwards flow at night. Movement of ionization across the diffusive barrier therefore cannot produce the required night-time influx of 2.1012 electrons/m2/sec. Vertical movement of ionization can still occur, however, through changes in the height of the diffusive barrier. This effect has been studied assuming that the barrier is completely effective (TITHERIDGE, 1968b). In a given tube of force, a fixed number of protons will then float up and down on top of the heavier 0+ ions. At medium latitudes, a diurnal variation of 2: 1 in the electron density of the ionosphere causes a change in the height of the diffusive barrier from about 1000 km near noon, to 500 km at midnight. This agrees with recent observations. The resulting downwards flow of ionization across the 1000 km level, between noon and midnight, is shown by the broken line in Fig. 10. The horizontal scale in Fig. 10 gives the latitude of the dipole field line at a height of 400 km, corresponding approximately to the centre of the ionosphere. The amount of the influx is reduced at low latitudes, because of the reduced electron content of t,he tubes of force. For geomagnetic latitudes below 16” the lines of force do not rise above 1000 km, so the influx across this level is zero. The influx is also reduced at latitudes above 45”, because of the small diurnal change in the temperature and movement of composition of t’lle exosphere at high latitudes. A large downmrds

1872

J. E.

Geomagnetic

TITHERIDGE

Latitude

(at

400km)

The solid line shows the calculated Fig. 10. Diurnal changes in the exosphere. electron content of the tube of force, extending from 1000 km to the Equat’or, at night. The broken line shows the amount by which this tube content decreases from day to night. The crosses and circles indicate the observed flux into the night ionosphere at different latitudes.

ionization therefore occurs only at latitudes between 20 and 45”. This is precisely t,he latitude range over which a night-time source of ionization is required (Table 4). The circles in Fig. 10 show the amount by which the total electron content of the ionosphere is increased, in a few hours near midnight, at geomagnetic latitudes of 38”s and 47OS. They give the average of the summer, equinox and winter influxes from Table 2. The crosses show the corresponding increases obtained from the peak density plots of Fig. 7, assuming an average value of 240 km for the slab thickness of the ionosphere. These observed influxes are about twice the calculated value, shown by the broken line in Fig. 10. The discrepancy could result from errors in the model used to calculate the night-time influx. This model was based on Explorer 22 measurements of the electron density and temperature at a height of 1000 km. It agrees with whistler and satellite measurements above 1000 km. Downwards extrapolation of the model does, however, give ionospheric densities which are too small by a factor of about 2. If the assumed exospheric densities are doubled, to join correctly with ionosphere profiles, the calculated night-time influx across the 1000 km level will also be doubled, giving good agreement with the observations. The theoretical curves in Fig. 10 assume that there is no diurnal change in the total number of protons in a given tube of force. The increase in the number of protons below 1000 km at night is balanced by a small decrease in the proton density at the top of the field line, where the volume of the tube of force is very large. This small decrease does not appreciably affect the total number of ions in a vertical column of unit cross-section. The increased number of protons at low heights will therefore cause an increase in measured values of the total content of the ionosphern. There will also be an increase in the density at the peak of the ionosphere, since tlrc

The maintenance of the night ionosphere

lSi3

0+ ions are compressed into a smaller volume. This effect can be estimated independently as follows. Calculations of the decay of the night-time F-layer generally consider the changes occurring in a vertical column of ionization. Horizontal movements, and the effects The electron density profile then apof the Earth’s magnetic field, are neglected. proaches that of an u-Chapman layer, and decays with the same rate at all heights. At heights above 500 km, this decay is produced by a steady downwards flow to replace ionization lost by recombination in the lower F-region. The electron content of the ionosphere, at medium latitudes, is typically about 12.1016 electrons/m” during the day, decreasing to 6.10 l6 electrons/m2 at midnight (TITHERIDGE, 1966). About one third of this ionization is above 500 km. The downwards flow of ionization across the 500 km level, between sunset and midnight, should therefore be about 2+1016electrons im2. In practice, it has been observed that the height of the 0+ to H+ transition level, at’ medium lat,itudes, decreases from about 1000 km during the day to 500 km at night. This agrees with theoretical calculations based on the assumption that the diffusive barrier is completely effective, so that there is no production or loss of protons by charge exchange (TITHERIDGE, 1968a). All the 0+ ions above 500 km during the day must therefore diffuse down to heights below 500 km at night. This gives a total flux of 4.10X6 ions/m2 across the 500 km level; twice the value required for diffusive equilibrium. Changes in the exosphere therefore produce an addit,ional influx of about 2.1016 ions/m2 into the night ionosphere. This calculated value agrees well with the observed influxes (Table 2). 8. CONCLUSIONS The total electron content of the ionosphere has been recorded continuously, at latitudes of 34’S and 42’S, since June 1965. Records up to the end of 1966 were used to produce accurate mean curves showing the changes in electron content from sunset to sunrise in each season. Similar curves, showing the variation in the density at the peak of the ionosphere, were obtained from ionosonde measurements at 1atit)udes of 10-60”. The effective loss coefficient B was found to decrease for several hours after sunset in equinox and winter. From 9 p.m. to sunrise, however, the value of /3 was constant at about 4.10P5 set-l in summer, and 3*10-5 set-l in equinox and winter. This is appreciably less than the values of 5*10-5 in summer, and 12~10-~ in equinox and winter, obtained from the rate of change of electron content at sunset. The difference can be fully explained by the changes which occur in the height and temperature of the F-layer, reducing the number of nitrogen molecules at the peak of the layer after sunset. The increase in height after sunset, and the resulting decrease in p, is slightly less at low latitudes. There is little change in the night-time values of /l over t)he solar cycle, except for an increase to 5*10-5 set-i in the winter values at low latitudes. In all seasons, the night-time loss coefficient increased by about 30 per cent during periods of high magnetic activity. The mean curves also show clearly the presence of a nocturnal source of ionization. This acts from about 21.30 to 00.00 hr L.T. in summer, and 22.30 to 02.00 hrinwinter. During this time, it raises the values of t’otal cont’ent from one exponential decay

1874

J. E. TITHERIDGE

curve to a second curve at a higher level. At medium latitudes, the total amount of ionization produced by the source is about 2~10~~electrons/m2. The influx is slightly greater in summer and winter Ohan at the equinoxes. It is confined to a limited range of latitudes, from about 15 to 40” geomagnetic in summer and 25 to 50” in winter. An analysis of Northern Hemisphere measurements shows that this change is a true seasonal effect, with the night-time source of ionization moving closer to the Equator in summer in both hemispheres. There is a world-wide annual variation in the strength of the source. The total amount of the night-time influx, in both hemispheres, is 50 per cent greater near December than near June. The same latitudinal and seasonal variat,ions are observed throughout the solar cycle, but the magnitude of the night-time influx is about 2.5 times greater at sunspot maximum (R = 160) than at sunspot minimum (R = 15). The night-time influx of ionization also shows a definite dependence on magnetic activity, at all seasons. Under disturbed conditions (K, = 3 to 5) t’he total amount of the influx is increased by about 50 per cent. The nocturnal ionization could be produced by high energy particles. Some evidence for this is provided by air glow observations, which require a production of ionization at heights below 250 km, and electron temperature measurement’s which indicate an input of heat throughout t’he ionosphere at’ night. NATEIAN (1966) showed that both these observations could result from a flux of t’rapped electrons with an energy of a few keV. Calculations show that this flux would also produce about 1012 electrons/m2/sec in the F-region. This is almost, sufficient to explain the changes in the night ionosphere. The heating effects are, however, observed throughout the night whereas the influx of ionization is required only near midnight. The heating can also be explained more readily by conduction from the exosphere. The night-time influx could also be produced by ionization diffusing down from the exosphere. If the diffusive barrier is completely effective, so that there is no diurnal change in the total number of protons in a given tube of force, the height of t’he O+ to H+ transition level at medium latitudes will decrease from about 1000 km during the day to 500 km at night. This change agrees with recent observations. Calculations show that it will produce a flow of ionization, across the 1000 km level, sufficient to increase measured values of electron content by 1-2.1016 electrons, m2. The calculated influx is also confined to latitudes between 20 and 45” geomagnetic, agreeing exactly with the observed effects. The decrease in the height of the O+ to H+ transition level causes an influx of ionization across the 500 km level, of 2.1016 electrons/m2. This is just t’he a’mount required to explain the observed changes in the ionosphere. The influx is reduced at latitudes above about 45’ because of the smaller diurnal changes in the exosphere. At high latitudes the electron temperature remains high throughout the night. The Of to H+ transition level is well above 1,000 km, and its movement has little effect on the ionosphere. During the winter night, however, the atmospheric temperatures will be lower and the transition level may fall sufficiently, at latitudes of 40-W, to affect the ionosphere. The nocturnal influx would then extend to higher latitudes in winter, as observed. The large difference in the sunset times at the two ends of a mid-latitude field line will also produce a flow of ionization across the Eq.uator, increasing the effects in the winter hemisphere.

The maintenance

of the night

1875

ionosphere

The flow of ionization from the exosphere will increase gradually after sunset. and will cease before ground sunrise. The major part of the influx could therefore be concentrated into a few hours near midnight, as required by the observations. The annual change of 50 per cent in the electron density of the exosphere will cause the observed change in the strength of the nocturnal influx. The solar cycle variation of about 2.5: 1 in the nocturnal influx is also in agreement with probable changes in t’he density of the lower exosphere. It therefore appears that, if the basic assumption of proton conservation is accepted, diffusion of ionization from the exosphere will produce most of the observed changes in t,he night,-time ionosphere. 9cknowZedgentent~This Sational Aeronautics

work was carried out under Research and Space Administration, Washington,

Grant D.C.

No. SsG-64-60

from

tllc.

REFERENCES BRACE L.H.~~~REDDYB.M. CARPENTER C_4TCHPOOLE l)ALGARNO

D.L. J.R. 9.

R. A. DVNGEY J. TV. EVANS J. V. Evaxs J. V. GEISLEZ J. E. GEISLER .J.E. and BOWHILL S. A. H.4r;soxW.B.and ORTENBURGER I.B. HAKSOXW.B.~~~ PATTERSON T.N.L. HASSONTV.B.~~~PATTERSON T.N.L. HARIUS I. and PREISTER W. KINGJ.W.,KOHL H,~~~PRATTR. MARTYN D. F. MlWDOXcA F.

DVSCAN

SATHAN 1~.V. S. K. O'BRIEN B. ,J., ALLC-~~F. R. and GOLDWIRE H.C. PRAG A. B., &rORSE F. 9. and McNE_& R. J. J~ISHBETH

H.

Rosa A.V. and SMITH F.L. SXITH R. L. T~ronf~sL. and SORTON R. B. TITHERIDC~EJ. E. TITHERIDGE J.E. TITHERIDGE J. E. TITHERIDGEJ.E.~~~ANDREWSM.K. ~~~~~~~~~~~ A. P. \YKIGIIT,J.IV. 1-.zxoK.

DA

1965 1962 1967 1964 1956 1956 1965a 1965b 1967 1965 1961 1963 1964 1962 1967 1956 1965 1966 1965 1966 1964 1967 1961 1966 1966 1968a 196Sb

1967 1964 1962

J. Geophys. Res. 70, 5793. J. Ccophys. Res. 67, 3345. J. L4tmosph. Terr. Phys. 29, 819. _4nnls. Geophys. 20, 65. Aust. J. Phys. 9, 436. J. Atmosph. Terr. Phys. 9, 90. J. Geophys. Res. 70, 4331. J. Geophys. Res. 70, 4365. J. GeopJzys. Res. 72, 81. J. Atmosph. Terr. Phys. 27, 457. J. Geophys. Res. 66, 1425. Planet. Space Sci. 11, 1035. .Planet. Space Sci. 12, 979. J. Atmosph. Sci. 19, 4. ,I. Atmos. Terr. Phys. 29, 1529. Sust. J. Phys. 9, 161. Spare Research T’. p. 687. Sorth-Holland. Amsterdam. Plrcn~t. #puce Sci. 14, 717. J. Gcophys. Res. 70, 161. J. Geophys. Res. 71, 3141. J. Stmosph. Terr. Phys. 26, 657. J. Geophys. Res. 72, 1829. J. Geophys. Res. 66, 3709. J. Gecphys. Res. 71, 227. J. Atmosph. Terr. Phys. 28, 1135. J. Geophys. Res. 73, 2985. ,I. Atmosph. Terr. Phys. 30, 1845. Planet. Space Sci. 15, 1157. hoc. Ro?J. Sot. A281, 140. J. Res. Satn. Bw. Stand. 66D, 29i.