AbstractCorrelation distances in the ionosphere are important for ionospheric data assimilation models that must balance modelled parameters against real observations. Many communication systems that transmit radio signals in the HF band rely on such models in order to operate accurately. However, defining the correlation distance is difficult because of the considerable variations present in the ionospheric regions and the lack of dense observations to measure them. In such cases, models have to entrust that an observation can be extrapolated away from its source, which if wrong can introduce errors in the ionospheric representation.
In this research, the correlation distance in the ionosphere is examined from the perspective of two instruments: the ionosonde and the HF ground-to-ground sounder. Both provide observations of the bottomside ionosphere through time delay and in the latter case also by angle of arrival.
Initially the research focussed on the analysis of HF angle- of- arrival (AoA) observations to characterise and understand the effects of travelling ionospheric disturbances (TIDs) from a dedicated campaign, part of the USA High-Frequency Geolocation and Characterisation Program (HFGeo). It was found that nowcasting the TIDs was critically important for the application of HF geolocation. Having established this, the foundation questions for the rest of the research were to find the ground density of instrumentation required for ionospheric modelling – hence finding the correlation distances.
The analysis turned to Europe where a network of suitable ionosondes existed with a long-term dataset that would allow an evaluation over multiple years. The initial analysis of the similarity between the ionosonde data from different stations observed inconsistencies between using Pearson’s correlation coefficient and root-mean-square error (RMSE). A new approach was implemented, called Taylor diagrams, that is capable of summarising multiple statistical parameters onto a single graph in order to produce a similarity measure. The research further studied the diurnal cycle phase offset in the foF2 measurements from the European ionosondes and phase synchronisation method was developed that that increased the East-West correlation distance twofold.
The research then allowed the new correlation method to be used on the HFGeo data from the USA during calm and TID affected ionospheres. It was shown that the correlation distance was dependent not only on the severity of the ionospheric conditions but also on the height of the reflection layer and the sampling frequency of the ionosonde. The chapter showed that during TIDs the ionosonde correlation distance can be reduced below 147±17 km and during very strong disturbances – to 47±15 km. However, in a manner analogous to the diurnal cycle in the foF2 measurements, the results indicated that most of the decorrelation was due to the mismatch of the phase front of the passing TIDs across the region.
In response to the need to track and model the phase front of TIDs, a technique called Feature Based Dynamic Time Warping (FDTW) was applied to vertically and horizontally synchronise the phase offset between ionosonde density profiles. This technique was demonstrated to operate successfully in tracking TIDs across the region of interest.
The research has implications for the density of observations needed of the bottomside ionosphere for mapping the ionosphere to provide support applications such as HF communications and geolocation. This in turn is important for the deployment of instrumentation and the correlation distances applied in algorithms for data assimilation.
|Date of Award||24 Jun 2020|
|Supervisor||Cathryn Mitchell (Supervisor) & Biagio Forte (Supervisor)|