|Back to INAG Homepage||Back to UAG-104 Contents page|
TOWARDS A SEMI-AUTOMATIC IONOGRAM SCALING PROGRAM TO PROVIDE MODEL PARAMETERS RATHER THAN TRADITIONAL SCALING PARAMETERS
Allon W. V. Poole
Hermann Ohlthaver Institute for Aeronomy, Rhodes University, Grahamstown 6140, Republic of South Africa.
Phone: (SA code) 461-318460; Fax (SA code) 461-25049;
Christopher C. Mercer
VODACOM, P. O. Box 335, Pretoria 0001, Rebublic of South Africa.
At ionosonde stations where the raw ionograms are produced in digital form rather than on film, traditional manual scaling methods have given way to automatic computer based scaling algorithms. However, the output data still adhere to the scaling rules and parameters that arose historically with the evolution of the ionosonde and were appropriate for the manual scaling of film based ionograms. We have found that the standard ionogram scaling parameters are not well suited to the requirements of real time ionospheric modelling. We have therefore embarked upon the development of a PC-based program that will fit an ionogram synthesised from a simplified real height model to the ionogram by least squares and then report the parameters of this model rather than the standard URSI scaling parameters. We hope to build up an archive of these data, and produce monthly medians that will be more directly representative of the ionosphere.
At the Hermann Ohlthaver Institute for Aeronomy our ionograms are captured digitally and can be displayed on a computer screen. We have developed a computer program to allow the manual scaling of 14 traditional ionospheric parameters (foF2, h'F2, foF1, h'F1, foE, h'E, foEs, h'Es, fbEs, Es types, fmin, MUF3000, M3000F2, fxI) by moving a cursor around the screen. These data are then stored, and monthly bulletins are produced that report the data as well as monthly medians in the traditional manner. These medians are an attempt to describe the average behaviour of the ionosphere for retrospective and for archival use.
Unfortunately, these parameters are not an efficient description of the ionosphere for communication purposes. This is because they describe the ionogram and not the ionosphere. They also contain descriptive and qualifying letters that require human interpretation to be useful and are thus not optimally designed for computers. It has become apparent that these parameters are of limited application to our main goal of developing a useful temporal and spatial ionospheric model for the South African region.
For a similar number of parameters, a simple ionospheric model can be described. If a model is chosen that is accessible to analytic (rather than iterative or numerical) raytracing, it has greater potential for communications applications and other purposes such as single site location (SSL). It thus seems reasonable to describe the average behaviour of the model ionosphere rather than the ionograms and to store this information for archival purposes.
The model must therefore fulfil two important criteria:
(i) it must be analytical, as discussed above.
(ii) it must be described by parameters that can be meaningfully averaged over some time period, eg. a month.
In compliance with this requirement, we have chosen the Quasi-Parabolic (QP) layer as the basic building block of our model. Each layer is described by three parameters. Hill (1979) gives the standard form for the "normal" QP layer:
is the electron density
is the geocentric distance
is the value of r at the layer peak
is the value of N at the layer peak
is the smaller value of r when N = 0 (bottomside)
is the layer thickness
We then concatenate three such "normal" QP layers with two "inverted" QP layers to form a Multi-Quasi-Parabolic (MQP) model [Chen et al (1990)]. The three normal layers represent the E, F1 and F2 regions of the ionosphere. The form of the inverted layers is the same as that for the normal layers except that the signs of A, B and C are reversed.
The three QP-layer parameters rm, Nm and ym specify the height of maximum ionisation, the maximum ionisation and the layer semi-thickness respectively. Thus monthly median values of these quantities will specify a median layer in an easily understood and uncomplicated way. It is proposed that archived, median values of these quantities will provide a reasonable basis from which each model layer can be reconstructed.
Fitting the model to the ionogram
One standard way of doing fitting the model to the ionogram is to convert the ionogram to an N-h profile by means of some inversion program such as POLAN. Q-P layers are then fitted to this profile by, for example, least squares. We propose to iteratively adjust the QP layers by numerical, non-linear optimisation methods until the corresponding ionogram best fits the input ionogram. This task is simplified and speeded up by the fact that the model is analytic, allowing for the speedy calculation of the ionogram that would be associated with that model. This "model" ionogram is then compared with the "data" ionogram and the error signal used to iterate the model until the best fit is obtained. A comparison of the standard method and our proposed method is shown in Figure 1.
A synoptic, analytic model is required to describe the ionosphere in a manner that can be easily averaged for archival and prediction purposes. The MQP model is well suited to this purpose because it can be ray traced analytically at both vertical and oblique incidence. HOIA is writing an interactive PC-based program that will fit this model to digital ionograms without needing to convert to an N-h profile first.
Hill R, Radio Science, 14, p885, 1979
Chen J, Bennett J A and Dyson P L, J. atmos .and terrestrial Phys., 52, p277, 1990.