Go to National Library of New Zealand Te Puna Mātauranga o Aotearoa
Volume 77, 1948-49
This text is also available in PDF
(2 MB) Opens in new window
– 91 –

A Radiosonde Method For Potential-Gradient Measurements In The Atmosphere.*

Summary.—A radiosonde method for obtaining records of potential gradient is described. The method is a simple modification of the Bureau of Standards radio-meteorograph The factors involved in the calibration of the instrument, a typical record obtained, and the further extension of the work is discussed.

Introduction.—The distribution of land, sea and population in New Zealand—in particular in the Auckland district—makes it imperative to rely on radiosonde methods for the recording of potential-gradient distribution in the upper air, since the chances of recovering any records taken within an ascending instrument are extremely remote. Various types of radio-meteorographs or radiosondes have been and are still being developed by various countries(1). In this country, the best-known type of radiosonde transmitting and receiving equipment is that evolved by the Bureau of Standards, and, as it happens, the transmitter appears to be the most readily convertible one for our purpose.

The Bureau of Standard Radio-meteorograph.—In the Bureau of Standard radiosonde(2) the meteorological elements control the audio-frequency modulation of the transmitter's carrier frequency by means of resistance variation in the grid-circuit time constant (R-C combination) of the modulator valve. Disregarding the dashed lines and circles in the diagram, Fig. 1 represents the circuit of the transmitter as used at present.

An aneroid-barometer unit moves a switch arm over a contact assembly, in which every 5th resp. 15th contact is arranged to cause the emission of reference signals, i.e., such with a fixed audio-frequency modulation of 190 resp. 192 c/s. The remaining contacts when reached by the arm of the baroswitch activate a relay, connecting a hygrometer element in parallel to the grid-condenser of the modulator. When the baroswitch arm glides over an insulated spacer a temperature-sensitive resistance takes the place of the humidity-element. It should be noted that the contact strips are narrower than the insulated spacers—a feature of which is made special use, as will be seen later—thus giving with the normal meteorograph temperature indication of longer duration than those relating to humidity. The contact sequence is calibrated against pressure and since the temperature at each level is known, the record reveals accurately the height of the sonde at each particular instant.

[Footnote] * Paper submitted to the Sixth Science Congress, Wellington, 20th-23rd May, 1947.

– 92 –
Picture icon

Fig. I.

The audio-frequency modulation appearing at the receiver output is converted electronically into D.C. currents proportional to frequency and automatically recorded.

Prior Methods of Atmospheric Potential-gradient Measurements.—It is outside the scope of this paper to deal with measurements of the potential gradient on, or close to, the earth's surface. We are rather concerned with the distribution of the gradient at different heights above the earth and particularly with the disturbed field in the presence of thunderclouds.

One of the few methods serving this purpose, though not employing radio-transmissions, was evolved by Simpson and Scrase(3) who obtain a record within the instrument during the ascent and thus rely on recovery. The potential gradient is recorded on polarity paper, which gives the sign, and the instrument depends for its action on point-discharge currents initiated by the field. For if two points, connected by a conductor, are exposed in a strong electric field, with the axis of the conductor parallel to the field, point-discharge will commence, i.e., current will begin to flow from one point to the other, provided the magnitude of the field is above a minimum value (starting potential). The direction of the current is governed by the direction of the field, positive electricity entering the point directed towards the positive potential in an attempt to equalise the potentials at the levels of the respective points. The magnitude of the gradient dV/dX determines the width of the trace obtained on the polarity paper, giving a gradient of 10 volts per centimetre as the smallest potential gradient which produced a legible record with the arrangement used (20 m.-length of wire). In general, as will be seen, the sensitivity of point-discharge current measurements depends on the length of the collector wires, the number of the points used at each end and the overall sensitivity of the recording arrangement.

Mecklenburg and Lautner(4) devised a particularly small and stable electrometer attached to a parachute and suitable for throwing from an aeroplane. But here again recovery of the apparatus was essential.

Wenk(5), on the other hand, designed a sonde in which the discharging time of a condenser is made a function of the applied potential, the time being longer than a reference value for positive and shorter for negative gradients. By interrupting transmission periods to correspond to these time intervals, both

– 93 –

sense and magnitude of the potential gradient is conveyed to the receiving station.

The Modifications to the Bureau of Standard Radiosonde.—The main idea adopted in our modification of the Bureau of Standard Radiosonde to make possible potential gradient recording is just the opposite to Wenk's principle. Using the point-discharge method a positive gradient causes a current to flow through the grid leak, to which our collector wires with one or several points at their ends are attached, in such a direction as to assist the discharge of the grid condenser in parallel to the grid leak and vice versa. Hence in influencing the duration of grid-condenser discharge our modulation frequency is increased for positive gradients whilst a negative gradient, which leads to a reversal of the current direction, will decrease the modulation frequency from a pre-adjusted reference value. The modulation appears in the R.F. section of the sonde in the same way as in the original apparatus, namely by interrupting the R.F. oscillation during the period when the modulator is oscillating. Thus the duration of the R.F. interruptions is sensibly constant and only the number of those occurring will be perceived as the modulation frequency by the unaltered receiving equipment. This represents a further difference to Wenk's principle.

Preliminary tests made it seem advisable to separate collector and radiator functions, using polystyrene-impregnated cork spacers between, the γ/2-dipole and the collector wires. Strongest signals were emitted for a collector wire length of (2n-1) γ/2, n = 5 being chosen in the later ascents (corresponding to approximately 20 m. length). The relay of the original sonde was retained in the early ascents to disconnect the grid of the modulator valve from the collector wires and to insert a fixed-time constant at the reference contacts.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Calibration.—For a given strength of field, F, the point-discharge current, after the “threshold-gradient,” M, for a given length of collector wires has been exceeded, depends in the main on the number of points, n, exposed at each end of the collector wires, in fact it has been found that with the spacing between the points used, it is directly proportional to n. This is easily understood, since each point may be assumed to act individually for striking, but once discharge has started on each they act collectively. Based on Whipple and Scrase(6) the current follows the law i0 = n · a (F2 - M2) μA V/cm V/cm

where a is a constant depending mainly on the exposure of the point, i.e., their relative distance and shape. (For large values of F, M may be neglected.) The constant “a” and “M” is determined in the laboratory in an arrangement as indicated in Fig. 2 and thus the relationship between F and i can be verified. The vertical lines in the diagram represent tautly stretched wire screens, the outer ones being 4 ft. square whilst the inner ones measure 5 ft. square with small insulated lead-throughs in various positions. The horizontal line repre-

Picture icon

Fig. 2

– 94 –

sents the collector wire which is terminated in one or several points. The uniformity of the field is ascertained by repeating the measurement in different positions, and the distance between the outer and the earthed centre screens is altered until, for the same gradient, V, the current is sensibly constant. The earthed centre screens guard against leakage currents and serve to minimise the insulation problem.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

The relationship i0 = na (F2 - M2), however, only holds near the ground. At greater heights the effect of the decrease in atmospheric pressure must be allowed for, since with the same gradient, i increases as the pressure diminishes. According to Tamm(7), who measured the discharge current between point and plate under various conditions, the current change may be expressed approximately by i/i0 = (P0/P1)1.6

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

where i = current at pressure p, i0 = current at pressure p0 (ground level say), and if we introduce an exponential law between height and pressure of the usual form h = 8 log0 P0/P where h in km and the mean air temperature taken as 273°K, we obtain i = n a (F2-M2) e0.2h

Finally we have to determine the relation between current and modulation frequency and knowing p or h from the calibration of the baroswitch supplied with each sonde, F can be evaluated as a function of height.

The graph of modulation frequency may be determined in several ways—


By replacing i through the grid leak due to F by an equivalent current supplied by a high-impedance source;


by replacing the IR drop by an equivalent grid-bias battery;


by direct calibration of modulation frequency against F.

The preliminary tests were carried out and some experimental results were obtained in 1945 by R. Belin, M.Sc., as part of his Honours thesis under the supervision of the author.

Picture icon

Fig 3

– 95 –

Checking the relationship between i and F the experimental points follow roughly the empirical law stated above, and plotting the modulation frequency against bias voltage gave a linear relation (14 cycles variation per volt change for a particular transmitter) for positive voltages, whilst for negative bias the modulation frequency tends very rapidly and non-linearly to zero. This becomes also apparent in Fig. 3 which shows a direct calibration of L dV./dX, against modulation for a collector wire (length L) with seven points on either end. This curve gives a relation of approximately the second order for positive gradients only, since in this region the frequenecy depends linearly on the volts applied to the modulator grid.

This graph may be used to evaluate potential gradients from our records by dividing L dV/dX for a pressure-corrected modulation frequency or current by the length of the collector wire to obtain the gradient in volts/cm. However, the foregoing considerations are merely empirical, e.g., they do not take into consideration different types of ions that may be encountered in the atmosphere, particularly in clouds. They must be accepted with certain reservations until further study has provided a surer basis to our problem.

Depending on the collector-wire length, L, chosen, the sensitivity of the sonde can be readily assessed or adjusted, since the greater the distance between the points the smaller the gradient need to be in order to produce the same point-discharge current. Whilst, however, a change in length affects the minimum gradient which can be measured, the number of points only controls the current flow after the starting potential has been reached. The minimum gradient measured so far is 3 volts/cm.

Picture icon

Fig. 4.

– 96 –

Preliminary observations seemed to indicate a noticeable if not considerable influence of relative humidity on the striking potential, thus influencing particularly the evaluation for weak fields. This problem is being further investigated.

Results.—Of particular interest is the investigation of charge distribution in thunderclouds and the potential gradient under these disturbed conditions.(8) Only few records have been obtained so far by Belin and the author(9) One of these, shown in Fig. 4, has particularly interesting features and gives an idea of the sensitivity of the sonde.

The meteorological conditions as judged from the ground revealed some fracto-stratus formation below a towering cumulo-nimbus cloud.

A smoothed curve drawn through the recorded points is shown in Fig. 4. Starting from about 4 volts/cm. the field increases first to 5 volts/cm. at 600 ft. and then drops back gradually to zero up to 2,000 ft. A weak negative region follows up to 3,500 ft. This latter part corresponds to the fracto-stratus formation. On entering the cumulo-nimbus, the gradient rapidly increases beyond the recorder range, a rough estimate based on the modulation frequency observed with the monitor loudspeaker indicating a gradient of about 20 to 25 volts/cm. At about 7,000 ft. a decrease (still off the record) commenced followed at 10,000 ft. by a rapid reversal to a strongly negative field (zero modulation frequency—the absence of receiver noises indicating the presence of the carrier). At 14,000 ft. the negative gradient commenced to decrease and changed ultimately into another positive gradient of 6 volts/cm. at 14,600 ft. which at 20,000 ft. had decayed again to normal field strength following the inverse-square law relation.

In explaining the record, it appears that the cumulus cloud is negatively charged at its base with a strong positive charge centred at about 10,000 ft., a further strong negative charge appearing at 14,000 ft. The base of the fracto-stratus region appears to be positively charged as well as the upper-most region beyond 20,000 ft.

Picture icon

Fig. 5 gives a reproduction of the top portion of the actual record showing continuous small-scale fluctuations observable throughout the ascent. (Any attempt at explaining these fluctuations will have to wait until more records of this kind have been obtained.)

Further Extension of Work—It will be apparent that a logarithmic response of the sonde or some kind of range extension would be preferable in order to increase the range and still maintain good accuracy for small gradients. Whilst this can be readily done on the receiving end for positive gradients, either electronically or by a manually controlled attenuator, this method would have no influence on negative gradients as here the modulation

– 97 –

frequency soon drops to zero. Shifting the reference frequency higher up leads either to loss of sensitivity (small grid leak say) or excessive bias-voltage values. The choice of grid-condenser values is likewise limited by the oscillatory conditions of the modulator valve. The solution which is being incorporated into the sondes at present consists in causing negative gradients, which lead to zero modulation frequency, to give increased positive modulation-frequency excursion. This is achieved by combining relay and baroswitch in the arrangement as shown in Fig. 6.

Picture icon

Fig. 6

Positive gradients with a current flow downwards, give an increase in modulation frequency in the down position of the relay-switch, whilst negative fields give a positive deflection when the cathode is switched to the upper collector wire. Since the contacts over which the baroswitch glides are narrower than the insulating spacers, the switch position is easily recognised from the record. The simple alteration essential to convert the original radiosonde transmitter for potential-gradient operation are indicated by the dashed circles and lines in Fig. 1.

The grid condenser has been changed from a value of about .08 μF to .003 μF, the humidity element is not being inserted, and the temperature sensitive resistance is to be replaced by a 1 megohm carbon resistor. Finally, the collector wires ending in one or several points are connected to the relay contacts as shown.

Combining the potential-gradient record with one obtained on a different radio frequency by a second receiver from a normal sonde tied to the gradient sonde, temperature correlation and humidity correction will ultimately be possible. But whilst the radiosonde methods have the advantage of giving on immediately available record independent of recovery of the transmitter, the apparatus would become more involved if one attempted to sectionalize a cloud by releasing a greater number of sondes in relatively quick succession, say ever 10 minutes.(10) This would require a frequency separation between the individual transmitters to cover a wider band with corresponding additions to the receiving and recording apparatus. On the other hand, the need for this complication would be counteracted by the use of radar both to plot the actual path of the balloon through the cloud and by ascertaining the size of the water droplets present. These results would further benefit if supplemented by parallactic photography fixing the full extent of a cloud in space.

By using radio-active material on the points or better still stronger ionizing agents, e.g., small flames, the striking potential is lowered and, hence, the minimum gradient which can be recorded for the same length of collectors will be further reduced. Information about gradients at still higher levels up to ionospheric regions and beyond might be obtained by the ultimate use of rockets from which the sonnet attached to a parachute would be ejected.

– 98 –

Acknowledgments.—The author is greatly indebted to Dr. Barnett, Director of Meteorological Services, for his interest and generosity in making the sondes and necessary accessories available and to Professor P. W. Burbidge for his helpful suggestions.


(1) F. J. Scrase; J. Sc. Inst., 18, 119. No. 7, July, 1941.

(2) H. Diamond, W. S. Hinman, F. W. Dunmore, and E. G. Lapham; J. Res. Nat. Bur. Std., 25, 327, 1940.

(3) G. Simpson and F. J. Scrase; Proc. Roy. Soc. A, 161, 309, 1937.

(4) W. Mecklenbury and P. Lautner; Zs. f. Phys., 115, 557, 1940.

(5) P. Wenk: Naturwiss., 30, 225, 1942 (Wir. Eng. 1942, p. 315, 1913).

(6) F. J. W. Whipple and F. J. Scrase; Geo. Mem., No. 68, 1936.

(7) H. Tamm; Ann. d. Phys., 6, 259, 1901.

(8) K. Kreielsheimer; Austria. Journ. Sc., 9, 95, 1946.

(9) K. Kreielsheimer and R. Belin; Nature, 157, 227, 1946.

(10) G. Simpson and G. D. Robinson; Proc. Roy. Soc. A, 177, 281, 1944.