Back to INAG Homepage Back to UAG-104 Contents page

The Application of Ionosondes to HF Real-Time Frequency Management in Northern Australia

Kenneth J.W. Lynn,

Communications Division, Defence Science and Technology Organisation,

PO Box 1500, Salisbury, SA 5108, Australia


The ability of networked ionosondes to provide accurate frequency management advice to HF communicators is being investigated. in northern Australia. Previous studies carried out in southern Australia have shown that the direct conversion of vertical ionograms to equivalent oblique ionograms was an accurate and effective means of providing Maximum Useable Frequencies for communication links up to 1000 km in length. The effect of spatial decorrelation as distances between control and reflection points increased was also investigated. The present program has extended the range of paths studied and demonstrated that oblique as well as vertical ionosondes can be used to provide the control point information for non-colocated links up to 2460 km in length. The importance of correcting for ionospheric gradients as the separation between the reflection and control point increases is confirmed and a method for so doing indicated.


Trials carried out in southern Australia (Lynn and Malcolm, 1991) demonstrated that vertical ionograms could be accurately converted to equivalent oblique ionograms in real-time for the frequency management of HF circuits (up to 1000 km in length) with reflection points within some 300 km of the ionosonde. Decorrelation of foF2 with increasing control/reflection point separation was found to consist of two terms, one being associated with the short term time variability (<4 hours) of the ionosphere and the other being related to large scale geographic gradients in the ionosphere including the effects of the terminator (for recent European measurements of foF2 correlation and gradients see Soicher et al, 1993). In early 1992, this work was extended by demonstrating that a vertical ionosonde near the path mid-point could provide accurate values of MUF for a 2590 km path from Darwin to Melbourne (Lynn and Kelly, 1993).

The present paper describes work in progress involving a network of oblique ionosondes with transmitters in S.E. Asia for reception in Australia (known as LLISP - the Low Latitude Ionospheric Sounding Program; Clarke et al,1993; Wright et al, 1993) as well as transmitters within northern Australia. The network within northern Australia was set up as a temporary measure specifically to develop Real Time Frequency Management (RTFM) techniques for HF communications in anticipation of the major ionosonde deployment which is expected in 1995 in support of the Jindalee Operational Radar Network (JORN). Extended ionospheric sounding is the backbone of frequency management and range calibration for an OTH radar such as JORN but has seen little use in the management of HF communications. It is the aim of the present project to develop the latter application. The main objective of LLISP is to test and improve the modelling of the low latitude ionosphere. However it is hoped to expand the application of RTFM into the equatorial area as experience in such operations is developed. Initial results obtained to date in northern Australia relevant to the RTFM of HF communications are described.


LLISP oblique chirp ionosonde transmitters are currently situated in Saipan, the Phillipines, Papua New Guinea and Cocos Island with receivers at Townsville, Darwin and Derby (see Figure 1(a)). These receivers also work in conjunction with chirp ionosonde transmitters at Townsville, Darwin and Tennant Creek in support of the north Australian RTFM program. In addition, there is a quasi-vertical ionosonde in operation at Darwin. A pulsed ionosonde (4B/IPS42 hybrid) and chirp receiver have also operated intermittently at Tindal. The ionosonde network in northern Australia is shown in Figure 1(b). All oblique ionosondes were designed and built in DSTO and feature low power operation (some tens of watts on long paths), flexible computer control of operating parameters, fast sweep speeds (e.g. 500 kHz/sec) and a maximum sounding frequency in excess of 64 MHz (required over long equatorial paths at high sunspot number). All ionosondes operate from atomic frequency standards to minimise relative drift.

Figure 1. Sites and oblique ionosonde paths (a) LLISP (b) northern Australia.

A new vertical pulsed ionosonde (the IPS 71) has been developed by KEL Aerospace to DSTO specifications and employing DSTO concepts to obtain true high-resolution Doppler ionograms. The IPS 71 typically sweeps 171 frequencies in 3.5 minutes obtaining Doppler measurements over a bandwidth of +2.5 Hz to a Doppler resolution of 0.039 Hz for every 6 km range bin from 75 to 842 km at every frequency. The IPS71, now in operation in Adelaide, will be deployed in northern Australia to further investigate the dynamics of the ionosphere.


The technique pursued here is to directly convert the reference ionogram to an equivalent oblique ionogram for the required range by a well known method (Davies, 1990). The technique as used is based on a curved earth and flat ionosphere model. A bounded error develops as the flat ionosphere approximation fails, typically when the algorithm is used to convert a vertical to an oblique ionogram at ranges greater than about 1000 km. A correction for this error has long been known as the k factor. A simple linear dependence on distance for the k factor has been assumed here suitable for correcting the F layer peak value but more sophisticated models can be developed to make the k correction a function of height and sensitive to the ionogram profile. The basic conversion algorithms have been generalised and consolidated to operate from either vertical or oblique ionograms for conversion to either greater or lesser range (Lynn,1992).

The method is ideal for operating with a mixed network of vertical and oblique ionosondes since it always produces a result, is computationally quick enough for real-time operation even on simple computers and uses all the data inherent in an ionogram (automatically correcting for height variations, for example). However the method is applicable only to the o-ray and does not take into account the effect of the earth's magnetic field which may introduce directional asymmetry. As with all methods which fully utilise oblique ionograms, the ionosonde must provide an absolute measurement of time delay. To date, this has been accomplished by calibrating time delay over the oblique ionosonde path using Sporadic E as a known reference height (Lynn and Malcolm, 1992), however GPS absolute timing has now been added to all sites .

Another technique which could be directly substituted for that used here (though with much greater complexity) is to invert the ionogram and ray trace through the resultant electron density profile (Chen et al, 1992) to again obtain the equivalent oblique ionogram. For other approaches to providing real or near-real time frequency management see Reilly et al (1991), Goodman,1992.

The main scientific problem to be overcome in using networked ionosondes to control HF communication links is to determine the distance from a sounder reflection point for which the data remains valid i.e. the correlation distances of ionospheric parameters and the manner in which decorrelation occurs. Much statistical work bearing on this point has been done in the past to indicate that significant decorrelation for the F region occurs in the range 300 to 1000 km from the sounder reflection point (Rush, 1976; Rush and Edwards,1976). The present work has concentrated on determining the real-time errors which can occur at a 15 minute update rate and the manner in which the ionosphere spatially decorrelates. This has been done by comparing computed and observed MUF over a variety of oblique paths.

Figure 2. Observed MUF for Darwin to Tindal (281 km) and the spatial extrapolation based on the Darwin vertical values of foF2.


The quasi-vertical ionosonde at Darwin was initially taken as the reference for spatial extrapolation. Figure 2 shows a comparison between the o-ray F region MUF at Darwin and that observed over a 281 km path to Tindal (no range conversion is required at this distance). The agreement is within the errors of measurement. Figure 3(a) provides a similar comparison between the converted Darwin ionogram MUF and that actually observed over a 874 km path from Darwin to Tennant Creek. Considerable decorrelation is now evident. Examination shows that two factors are at work. The short-term (<4 hr) variations continue to correlate well at the two sites but significant biases develop each day such that the predicted MUF is an overestimation. Darwin is at the edge of the equatorial anomaly region and considerable north-south gradients in foF2 develop. Accordingly the spatial predictions were corrected by multiplying each calculated value by the ratio of foF2 at the oblique path reflection point and at Darwin using values given by the ionospheric prediction program ASAPS.

Figure 3 Observed MUF for Tennant Creek to Darwin (874 km) and the spatial extrapolation from Darwin with (a) gradient corrected (b) no gradient correction.

The result is shown in Figure 3(b). Here most of the decorrelation has disappeared confirming that the foF2 gradient was the main problem. A gradient in the height of the F layer could also be expected to occur but at this range appears to have made little contribution to the decorrelation. The question of the relative importance of gradients in foF2 and in height with increasing conversion range will be the subject of future investigation.

The results of Figure 3 are consistent with those previously obtained elsewhere in Australia (Lynn and Malcolm, 1992, Lynn 1992) in which vertical ionosondes were used as the reference for ionogram conversion. One of the main points in using the direct ionogram conversion technique was the expectation that both vertical and oblique ionosondes could be used. To test this, spatial extrapolations of expected MUF for a 2460 km path from Townsville to Derby were made using an oblique ionosonde operating from Tennant Creek to Darwin (874 km) as the reference in a period which included the major ionospheric storm

Figure 4 (a) The 300 km radius of high correlation expected around the Tennant Creek to Darwin control point relative to the Townsville to Derby mid-point. (b) Comparison of the observed MUF for Townsville to Derby (2459 km) with that computed from the Tennant Creek to Darwin oblique ionosonde (gradient corrected).

on 11 May 1992. Figure 4(a) shows the relative location of paths, reflection points and the region of expected high correlation about the ionosonde control point. The comparison between the calculated and observed o-ray MUFs for the 2460 km path is made in Figure 4(b) which also includes a minor foF2 gradient correction.

The computed MUF tracks accurately through the storm indicating that the ionogram conversion is automatically correcting for the major increase in ionospheric height which occurred. Thus the expectations of the technique were confirmed. The use of oblique ionosonde paths may be preferable in managing long one-hop circuits since any error in the k factor algorithm will be reduced because of the smaller disparity in length between the referenced and managed paths.

Real-Time Gradient Correction

The results to date have used gradient corrections taken from a standard propagation prediction program (ASAPS) to extend the distance for which control point data is accurate. A true real-time management system would derive its own gradient measurements thus allowing for the more extreme variations in large scale gradients which occur e.g. during ionospheric storms. At such times, predicted gradients can be greatly in error (Lynn and Kelly, 1993). Such corrections become possible within an area of networked ionosondes. Once again the problem arises as to how to combine data from both vertical and oblique ionosondes. The technique described here can be brought to bear by converting all ionograms to a common range (e.g. that of the circuit being managed). The gradient correction at the managed reflection reflection point then becomes an interpolation problem within the network of reference-point ionograms and is applicable to any ionogram parameter i.e. frequency or group range. Bradley and Dick (1993) have recently investigated methods for interpolating ionospheric data within a vertical ionosonde network. These methods would be equally applicable in our present application once all ionograms have been converted to a common range.

In a first attempt at normalising the ionograms, zero range was taken to verify that long path oblique ionograms (e.g. Townsville-Derby, 2460 km) could usefully be converted to the equivalent vertical. This did prove to be the case for the F region MUF for which good critical frequencies were derived from the high angle rays. Much of the E region structure was smeared by the process indicating that the fundamental problem with inverting long range ionograms is that structure at lower heights is not adequately resolved by current ionosondes because of the extreme time-delay compression which is present.


The present paper provides an update on a continuing project to develop and deploy practical real-time frequency management systems for HF communications both within and external to Australia. A technique has been demonstrated which will allow a mixed network of vertical and oblique ionosondes to provide frequency management to circuits of arbitrary one-hop length with reflection points within the area covered by the ionosonde network. Future work will aim at practical experience with users and the testing of the real-time ionosperic gradient-correction methodology.


Bradley P.A. and M.I.Dick, Use of real-time ionosonde data to improve regional ionospheric models for HF Over-the-Horizon radar and passive single site target location, This Conference Proceedings, 1994.

Chen J., Bennett and P.L. Dyson , Synthesis of oblique from vertical ionograms using quasi-parabolic segment models of the ionosphere, J.Atmos.Terr.Physics, 54, 23-331, 1992

Clarke R.H., Fyfe D.F., Kettler D.I., Lynn K.J.W., Malcolm W.P., Sprey B.M., Taylor D.P. and C.S.Wright, An ionospheric sounding program to investigate HF and low VHF propagation at low latitudes, TENERP Conf. Proceedings, June 22-24, NPS, Monterey, CA, 1993.

Davies K. , Ionospheric Radio, IEE Electromagnetic Wave Series 31, Peter Peregrinus Ltd., IEE, London, 1990.

Goodman J.M., HF Communications - Science and Technology, Van Nostrand Rheinhold, New York, 1992.

Lynn K.J.W., Ionogram range conversion, Proc. of the South Pacific STEP Workshop, Melbourne, February, 1992.

Lynn K.J.W. and T.Kelly, A new real time frequency management technique based on vertical sounding demonstrated over a 2598 km circuit, Proc.of the 7th International Ionospheric Effects Symposium, Alexandria, VA, 4-6 May, 1993

Lynn K.J.W. and W.P. Malcolm, The spatial variability of foF2 and its relevance to HF real time frequency management, Proc. of the South Pacific STEP Workshop, Melbourne, February, 1992.

Reilly M.H., Rhoads F.J., Goodman J.M. and M.Singh, Updated climatological model predictions of ionospheric and HF propagation parameters, Radio Sci., 26, 1017-1024, 1991.

Rush C.M., An ionospheric observation network for use in short-term propagation predictions, Telecomm. J., 43, 544-549, 1976.

Rush C.M. and W.R.Edwards , An automated mapping technique for representing the hourly behavior of the ionosphere, Radio Sci., 11, 931-937, 1976

Soicher H., Gorman F.J., Tsedilina E.E and O.V.Weitsman, Ionospheric foF2 values and their gradients at European longitudes, Proc.of the 7th International Ionospheric Effects Symposium, Alexandria, VA, 4-6 May, 1993.

Wright C.S., Kettler D.I., Trudinger P.L., Malcolm W.P., Sprey B., Taylor D.and R.H.Clarke., The LLISP data base of low-latitude oblique-incidence ionograms, TENERP Conf. Proceedings, June 22-24, NPS, Monterey, CA, 1993.

Back to INAG Homepage Back to UAG-104 Contents page