WDS'10 Proceedings of Contributed Papers, Part II, 194–198, 2010.
ISBN 978-80-7378-140-8 © MATFYZPRESS
Study of Chance for Good DDA Drift Velocity Estimation
for Ionospheric F-region Drift Measurements
D. Kouba
Charles University Prague, Faculty of Mathematics and Physics, Prague, Czech Republic,
Institute of Atmospheric Physics, Academy of Sciences of the Czech Republic, Czech Republic.
P. Koucka Knizova
Institute of Atmospheric Physics, Academy of Sciences of the Czech Republic, Czech Republic.
Abstract. Estimation of the plasma drift velocity measured by Digisonde depends on
the number of reflection points and their distribution. In the paper we divide and analyse
plasma drift measurements according to the number of reflection points. A detailed
study of Digisonde drift measurement quality has not been published yet. Two extreme
groups are selected for further detail analysis, the first one with less than 100 reflection
points and the second with more than 800 points. Within the data in these groups we
detect annular and diurnal variability. Measurements containing low number of the
reflection points occur mainly around equinoxes and during day-time. On the contrary,
maximum occurrence of the measurements with more than 800 points is in winter and
summer. The lowest chance to register extreme number of the reflections is during
afternoon and around sunset. A very special group of measurements consisting of
measurements with high number of the reflection points and a bipolar pattern of a
SKYmap has maximum occurrence during winter nights.
Digisonde drift measurement and velocity estimation
Digisonde plasma drift measurement is based on reflection of the electromagnetic wave from the
ionosphere (frequency of the transmitted pulse is equal to the plasma frequency at the reflection point).
The receiving antenna field observes vertical and oblique echoes arriving from the reflection points
locations. The oblique echoes occur due to ionospheric irregularities disturbing the electron density
contours.
The Digisonde software uses spectral analysis to distinguish individual echoes with the different
Doppler frequency shifts. The multielement antenna interferometry can determine the source location
(incidence angle, azimuth of arrival) for the identified echoes (Reinisch et al., 1998).
SKYmaps show a graphical representation of the reflection points. Reflection points are plotted
on a horizontal plane (the vertical echoes are at the SKYmap centre). Corresponding values of the
Doppler frequency shifts are usually distinguished by the different colors of the plotted symbols.
Identified source locations can be used to calculate the bulk velocity of the ionospheric plasma
structure over the sounder location. In an ideal situation, when the plasma moves with a single bulk
velocity (in that case the pattern of SKYmap points is bipolar), we can determine a velocity vector by
a least squares fit of the SKYmap points. This technique is commonly referred to as the ‘Digisonde
Drift Analysis’ (DDA method) and the resulting velocity is called the ‘drift velocity’ (see (Reinisch et
al., 1998) for more details).
It is necessary to emphasize here, the fact that drift velocity estimation based on DDA method has
a significant assumptions. First, the DDA method estimates a unique velocity vector across the region
of a measurement. It means that all the registered reflection points are used in the drift velocity vector
fitting. In a real ionosphere, the drift velocity can be described using a 3D vector field. Usually it is
possible to approximate the local situation over the sounder by a unique velocity vector. Unique (or
similar) drift velocity vectors across the measured region produce SKYmap with the bipolar pattern.
Area with a positive points passing across “zero” points to a negative area. Assuming a predominant
horizontal velocity component the reflection points in some direction from a sounder location have a
negative Doppler shift. Amplitude of the Doppler shift depends on a distance between the reflection
point and the sounder location for specific azimuth. Sounder detects larger Doppler shift from more
distant reflection points. For reflection points close to a vertical direction the Doppler shift is getting
194
KOUBA AND KOUCKA KNIZOVA: STUDY OF CHANCE…
near to zero. Measured SKYmaps with a bipolar pattern and a sufficient number of the reflection
points is easy to interpret and the DDA method estimates drift velocity accurately in this case.
The DDA method meets problem with velocity estimation in such cases when the SKYmaps have
a non-bipolar character. There are both a “pure positive” aside from a “pure negative” clouds on the
SKYmap and a motion in the ionosphere over the sounder is difficult to interpret. In these cases the
DDA method estimation fails. It is impossible to approximate velocity in the area over the sounder as
just an unique velocity vector here. Hence, the situation is more complicated and requires 2D or 3D
approach.
The second important assumption of the DDA method is a detection of the reflection points in the
oblique directions. For a perfectly smooth, horizontally stratified ionosphere and a given sounding
frequency only one such reflection point exists, responsible for a single vertical echo. The off-vertical
echoes occur due to the ionospheric irregularities disturbing the density contours. Theoretically, three
spatially separated reflection points are sufficient for the velocity vector calculation. Practically, the
DDA method fits the velocity vector for all measured reflection points. The velocity estimation is
more accurate and robust for SKYmaps with the reflection points which are distributed in a wide area.
SKYmaps with the reflection points cumulated only in a small cloud vertically over the sounder
produce inaccurate estimations. Therefore, for a robust result of the drift measurement, it is an optimal
to observe the disturbed ionosphere with undulated boundaries or cloudy structures. Such a case
guaranties plenty of reflections from a wide oblique area. Several works interpreted the drift data
where a velocity estimation was sometimes based on a few reflections (Belehaki et al., 2006).
Moreover, absence of any reflection points on the SKYmap was sometimes incorrectly interpreted as a
null velocity. However, the drift measurement can be interpreted as a null velocity only in a case of
Registering reflections from the different directions with a Doppler shift values close to zero. Absence
of any reflections can indicate the fact that the actual ionospheric plasma stratification or structure is
inconvenient for our type of the drift measurement and we are not able to obtain any information about
plasma motion.
For the registration of a sufficient amount of the reflection points there are also other important
factors like a local geological and a geographical conditions in the locality around antenna field,
measurement setting and a precise construction of the antenna field. Small number of the reflection
points or their absence can be also result of an incorrect choice of sounding frequency window or of a
blanketing by sporadic E layer. Usually, the drift measurement in F-region uses preceding ionograms
(foF2) to set the sounding frequency window. Optimally, the drift measurement should follow immediately after the ionogram sounding in order to eliminate any ionospheric changes as much as possible.
When the ionogram autoscaling software detects an incorrect critical frequency foF2, the drift
measurement might be performed with an improper sounding frequencies. The ionogram autoscaling
process is a complex problem and several working teams worldwide have been working on that task
for a long time (Galkin at al., 2008; Pezzopane and Scotto, 2007; Igi et al., 1992 among many others).
Examples of the ionograms (Figure 1) illustrate situation when Es layer stratification occurs.
Sporadic E is a thin layer with very high ionization sometimes exceeding maximum ionization in the F
layer. Such a layer may allow electromagnetic waves to pass through and later reflect from higher
layers or reflects all electromagnetic waves and completely blanket the above laying ionosphere. On
the left panel there is one of the most usually seen daytime summer ionogram. The reflection traces
from all E, F1, F2 and sporadic E layers are observed according to handbook of ionogram
interpretation (Wakai et al., 1987). In this case autoscaling process correctly detects critical frequency
and there is a good chance to perform drift measurement successfully. On the middle panel, there is
ionogram with traces of E, F, Es and Es second order reflections. Es layer partially blankets F layer
trace. Hence the F layer reflection trace is seen only at its upper part. Reflection from the F layer is
seen only in approximately 0.5 MHz frequency window (4.7-5.2MHz). Sounding frequencies for drift
measurement calculated automatically from foF2 might be likely affected by Es. Moreover, Es layer
varies quickly and can significantly evolve in time between ionogram and drift measurement.
Although autoscaling process detects the foF2 correctly, probability of a good F-region drift
measurement is minimal in such situation. On the right panel, there is an ionogram with the E, Es and
the multiple orders Es reflection traces. The F layer trace is completely blanketed there. Naturally,
foF2 can not be detected.
195
KOUBA AND KOUCKA KNIZOVA: STUDY OF CHANCE…
Figure 1. Example of ionograms (graph of the virtual height of the ionosphere plotted against
sounding frequency) with sporadic E occurrence. Only vertical ordinary reflections are presented.
In the situation when autoscaling process does not detect the foF2, the drift measurement is
performed in a fixed frequency window 3-3.5 MHz (Standard Digisonde drift setting). However, then
the drift measurement will fail. Although F layer exists in ionosphere, due to Es blanketing there is no
chance to detect any reflections from the F layer. Es layer complicates ionospheric drift measurement
in midlatitudes mainly during summer.
Observation in the station Pruhonice
In January 2004, a KEL Aerospace ionosonde was substituted by a new digital sounder Digisonde
DPS-4 with four cross-loop receiving antennas in the Pruhonice Observatory. Since then, the
Digisonde provides routine drift measurements for both E and F region in addition to a regular
ionogram sounding. Probably, thanks to a quality of the antenna field design, good terrain conditions
and proper measurement setting many reflection points were registered in most of the drift
measurements. Therefore we have a chance to study velocity estimation process in detail. We noticed
that not all the measured reflection points are related to the ionospheric drift. We introduced a new
quality control method included in the procedure of evaluation of plasma drifts. Our method consists
of a three-step selection of detected reflection points and application of the standard DDA algorithm
on the corrected SKYmaps. This selection method guarantees a better quality of obtained drift
velocities and improves further interpretation of plasma motion (Kouba et al., 2008).
We applied and tested an application of the selection method on all drift data collected in 2006.
We manually checked all the F, E, and Es drift measurements, totally of about several tens thousands
of records. We obtained unique high-quality dataset of ionospheric drift measurements registered
during a period of exceptionally low geomagnetic and solar activity.
Our experience shows that a measured drift data display often expressively different characters.
Within the measurement, there are trouble-free measurements with thousand of the registered
reflection points with the bipolar pattern of Doppler shift. However, there exist measurements from the
same day when only a few reflection points registered. In such case it is problematic to interpret the
result as a drift velocity (velocity estimation error is huge).
It is interesting to study together not only estimated drift velocity, but also number of reflection
points registered during drift measurements. Such an analysis shows that the probability of a
successful drift measurement can be dependent on a daytime or a season. It can be related to the state
of the ionosphere which is inconvenient for our measurement (absence of irregularities, horizontal
stratification) and it is necessary to interpret these measurements carefully.
Number of reflection points
During one drift measurement from a few to several thousands reflection points are registered.
Typical number of the reflection points obtained during one measurement is several hundred. For our
study, we selected two extreme groups of measurements. In the first group there are measurements
with less then 100 registered reflection points. Interpretation of this kind of measurements is often
problematic and velocity estimation typically leads to a higher standard deviation.
196
KOUBA AND KOUCKA KNIZOVA: STUDY OF CHANCE…
Figure 2. Diurnal and seasonal F-region drift measurement with less then 100 registered reflection
points occurrence.
Figure 2 shows diurnal and seasonal variations of the F-drift measurements with low number of
reflection points. Maximum occurrence of these measurements is around equinoxes. In the diurnal histogram, there is evident that the most probably we obtain low-quality measurement during day-time.
The second extreme group contains measurements with more than 800 registered reflection
points. When a huge number of reflections is registered, reflection points typically cover a wide area
over the sounder. Spatial distribution of the reflection points is an important property for further
velocity estimation. When a plenty of reflections in a wide area are registered there is a good chance
for correct and accurate interpretation and velocity estimation. It is possible to apply directly DDA
method on the drift measurements with a bipolar pattern. In case of measurements of the reflection
points with non-bipolar pattern it is better to limit the DDA estimation to a narrower area over the
sounder. In case of complicated SKYmap pattern the drift velocity is estimated for a smaller area over
the station and there is a better chance for fulfilling the assumption of a unique velocity vector (see
(Kouba et al., 2008) for more details). It is also possible to study registered reflections in more details
here. When there are plenty of registered reflections in a wide area, it is possible to reconstruct drift
velocity vector field over the sounder in 2D (or 3D) approximation. Of course, such a possibility exists
only for the exceptionally good measurement only and unfortunately such a situation is rare. However,
for a special situations (TID passing, gravity waves detection, geomagnetic storms) it can lead to
interesting results.
Figure 3 shows diurnal and seasonal variations of F-drift measurements with high number of
reflection points. Maximum occurrence of these measurements is in winter and summer. The lowest
chance to register extreme number of reflections is during afternoon and around sunset.
A very special group of measurements consists of measurements with high number of reflection
points and bipolar pattern of SKYmap. The DDA method produces the most reliable estimation for
this group of measurements. Estimated drift velocity is accurate in all components and measurement
can be easily interpreted. Figure 4 shows seasonal and diurnal distribution of measurements with
highest number of reflection points and bipolar pattern of SKYmap. Probability of registration of this
class of measurement is highest during winter nights.
Figure 3. Diurnal and seasonal F-region drift measurement with more than 800 registered reflection
points occurrence.
197
KOUBA AND KOUCKA KNIZOVA: STUDY OF CHANCE…
Figure 4. Diurnal and seasonal occurrence of F-region drift measurement with more then 800
registered reflection points and bipolar pattern of SKYmap.
Conclusion
Estimation of the plasma drift velocity measured by the Digisonde depends also on the number of
reflection points and their space distribution. Dividing the drift measurements according to the number
of reflection points we can select two extreme groups with low (up to 100 reflection points) and with
extremely high (more than 800) number of reflection points. From histograms, it is evident that such
measurements are not randomly distributed but show significant diurnal and annual variability.
Measurements containing a low number of reflection points occur mainly around equinoxes and
during day-time. On the contrary, maximum occurrence of the measurements with more than 800
points is in winter and summer. The lowest chance to register extreme number of reflections is during
afternoon and around sunset. A very special group of measurements consisting of measurements with
high number of reflection points and bipolar pattern of SKYmap has a maximum occurrence during
winter nights.
Comparison of measurements at other station can bring a new complex idea about ionosphere
dynamics and variability in the future.
Acknowledgments. This work was supported by the Grant Agency of the Academy of Sciences of the
Czech Republic (project IAA300420704), by the Czech Grant Agency, project No. 205/06/1267 and
205/06/1619 and European Union project COST 296 (MIERS).
References
Belehaki A.,Tsagouri I., Michalareas G., Herekakis T., F region drift observations from Athens Digisonde,
Radio Science, 41, 2006.
Galkin IA, Khmyrov GM, Kozlov AV, et al., The ARTIST 5, Symposium/11th International Digisonde Forum,
APR 27-MAY 03, 2007 Lowell, MA, conference proceedings, pp. 150-159, 2008.
Igi S., Minakoshi H. and Yoshida M., Automatic ionogram processing system. 2. Automatic Ionogram Scaling.
Journal of the Communication Laboratory 39 2 , pp. 367–379, 1992.
Kouba, D., J. Boška, I. A. Galkin, O. Santolík, and P. Šauli, Ionospheric drift measurements: Skymap points
selection, Radio Scence., 43, 2008.
Pezzopane M., Scotto C., Automatic scaling of critical frequency foF2 and MUF(3000) F2: A comparison
between Autoscala and ARTIST 4.5 on Rome data, Radio Science, 42, 2007.
Reinisch, B.W., J.L. Scali and D.M. Haines, Ionospheric drift measurements with ionosondes, Annali di
Geofisica, 41(5-6), 695-70, 1998.
Wakai, N., Ohyama, H., Koizumi, T., Manual of Ionogram Scaling, 3rd Version, Radio Research Laboratory
Ministry of Posts and Telecommunications, Japan, 119pp., 1987.
198