Amplitude scintillation measurements
We start by presenting examples and results of our evaluation amplitude scintillation measurements. Figures 3, 4 show examples of typical amplitude scintillation measurements made by ScintPi 3.0 and comparisons with collocated PolaRx5S records considering GNSS L1 and L2 signals, respectively. As mentioned earlier, these measurements were made at the Presidente Prudente site, in the southeastern Brazilian region.
Figure 3 shows (red curves) the amplitude scintillation index (S4) for the L1 (~ 1.6 GHz) signals measured by ScintPi 3.0 on Dec. 27, 2021. More specifically, it shows the temporal variation of the S4 values for each satellite tracked throughout the day. Each panel corresponds to the records from one satellite, and its constellation and identifier number are indicated in the upper left portion of the panel. Each panel also shows the elevation (blue curve) of the satellite. Only data for satellites with elevation > 20° are shown. Finally, S4 data from PolaRx5S are also shown in each panel for comparison purposes.
Figure 4 shows the S4 values for the L2 (~ 1.2 GHz) signals measured by ScintPi 3.0 (green curves) and Septentrio on Dec. 27, 2021. The format is the same as that of Fig. 3. Figures 3, 4 serve to illustrate that the S4 values provided by ScinPi 3.0 follow very closely the those provided by the Septentrio monitor. Careful inspection of the measurements shown in Figs. 3, 4 shows that increases in S4 values only occur during nighttime hours, mostly between approximately 00:00 UT (21:00 LT) and 04:00 UT (01:00 LT). This agrees with the expectation of scintillation associated with equatorial plasma bubbles (EPBs) reaching low latitudes, as is the case of PPR. A closer look in the data and additional details will be provided in the following sections.
For completeness, we also show expanded views of L1 and L2 scintillation measurements in Fig. 5. It shows a closer look at a few examples of scintillations measured by ScintPi 3.0 and PolaRx5S. Figure 5 allows one to better visualize the good agreement between the S4 values provided by the two receivers. Similar to ScintPi 1.0 (Rodrigues and Moraes, 2019) low S4 values (S4 < ~ 0.2) provided by ScintPi 3.0 are more variable than those provided by PolaRx5S. This is because of the lower resolution of signal strength values provided by the u-blox receivers compared to PolaRx5S.
Next, to better quantify the agreement between the S4 values provided by ScintPi 3.0 and those output by the Septentrio monitor, we provide the results presented in Fig. 6. Figure 6 shows scatter plots of the ScintPi S4 versus Septentrio S4 for L1 (red) and L2 (green) and for the two days of measurements (rows). Figure 6 also summarizes our comparisons of the measurements showing, for each day and signal, the coefficient of linear correlation (\(r\)), the average value (\(\mu\)) and the standard deviation (\(\sigma\)) of ScintPi S4 with respect to Septentrio S4 values. Therefore, \(\mu\) and \(\sigma\) represent the average and standard deviation, respectively, of the differences in S4 values provided by ScintPi 3.0 with respect to S4 values provided by the PolaRx5S. Only data from satellites with elevation angle greater than 20° are shown, and only data for which scintillation is observed (Septentrio S4 > 0.2) are considered in the analyses. The results show that the ScintPi S4 values are highly correlated with the Septentrio values with \(r\) ranging from 0.968 to 0.981. The results also show that the deviation of ScintPi 3.0 values with respect to Septentrio are minimal, with the average values ranging between 0.020 and 0.039 and the standard deviation ranging between 0.042 and 0.070.
Total electron content (TEC) measurements
We now turn our attention to TEC measurements made by ScintPi 3.0. In the following sections, we present and discuss code and phase TEC measurements made by ScintPi 3.0 and a comparison of these measurements with similar observations made by the collocated Septentrio monitor. Example results include TEC depletions detected by our ScintPi 3.0 and associated with scintillation events, and estimates of the temporal variation of vertical TEC curves.
Examples of code and phase TEC measurements
Figure 7 shows a summary of line-of-sight (slant) TEC measurements made by ScintPi 3.0 on Dec. 27, 2021. It shows, more specifically, the local variation of the TEC measurements obtained from differential pseudo-ranges (code TEC, \(TE{C}_{\rho }\)) in gray as well as from differential phase (phase TEC, \(TE{C}_{\phi }\)) measurements in red. The observations are shown for all satellites capable of transmitting two signals (L1 and L2) and tracked by ScintPi 3.0 throughout the day. The constellation and satellite identifier numbers are indicated in each panel. For reference, the satellite elevation angle (blue curve) is also indicated in each panel.
As mentioned earlier, code TEC is a noisy but absolute measurement. Phase TEC, on the other hand, is a relative measurement because of the intrinsic ambiguities associated with carrier phase tracking. Phase TEC measurements, however, are more precise than code TEC measurements (Jakowski, 1996). Figure 7 serves to show that both phase and code TEC can be estimated from ScintPi 3.0 measurements. For better visualization purposes, phase TEC curves were leveled to code curves to illustrate the agreement between the two estimates. Figure 7 also exemplifies the fact that data from a large number of satellites (61 in this example) can be obtained with ScintPi 3.0. This allows for a wide coverage of the sky by a single station compared to GPS-only receivers.
Comparison with Septentrio TEC measurements
We now take a closer look at the TEC measurements provided by ScintPi 3.0 and compare them with measurements provided by the collocated Septentrio PolaRx5S monitor. The left-hand side panels in Fig. 8 show expanded views of code (gray) and phase TEC (red) example measurements made by ScintPi 3.0 on Dec. 27, 2021. Information about constellation and satellite identifier numbers are exhibited in the respective panels. For comparison, PolaRx5S measurements for the same satellites and signals are shown in the middle panels. Finally, for a more direct evaluation of ScintPi 3.0 observations, phase TEC measurements from ScintPi 3.0 and PolaRx5S are shown in the right-hand side panels. Figure 8 shows that, for some signals, code TEC provided by Septentrio is noticeably less noisy than the code TEC provided by ScintPi 3.0. This is typically the case of signals from Galileo. The same difference is not as noticeable for signals from GPS, GLONASS and BeiDou. Note that BeiDou has fewer satellites than other constellations and measurements were not available for the time window in Fig. 8. Differences in code TEC for ScintPi 3.0 and PolaRx5S are the result of different receiver hardware and signal processing. More importantly, Fig. 8 shows that the phase TEC provided by ScintPi 3.0 is in excellent agreement with the phase TEC provided by PolaRx5S. ScintPi 3.0 measurements can capture small amplitude (fraction of TECU), long period (~ 1 h) variations as seen in the examples. ScintPi phase TEC measurements can also capture the large gradients associated with TEC depletions such as that detected by GAL27 between about 03:00 and 04:00 UT. Finally, because ScintPi 3.0 can make high-rate (up to 20 Hz) measurements of TEC, the system can be used to study small-scale irregularities.
While our initial goal was to produce sensors capable of providing information about the occurrence of amplitude fading cases in L-Band signals and about the occurrence of small-scale irregularities in TEC, we realized that ionospheric users might also be interested in the time variation of the vertical TEC for various applications.
We now present results that illustrate the potential of ScintPi 3.0 measurements on providing the temporal variation of absolute vertical TEC (VTEC). Again, for better evaluation purposes our ScintPi 3.0 results are presented alongside similar estimates produced using collocated Septentrio measurements in PPR. The estimation of absolute VTEC curves involves leveling the more precise but relative phase TEC to the absolute but noisy code TEC curves, taking into consideration the satellite differential code biases (DCBs) as well as the receiver DCBs, and applying a mapping function that converts slant TEC measurements to VTEC estimates. The procedure we followed is similar to that described by Carrano and Groves (2006), except that we simply set the receiver bias as the value that causes the minimum observed VTEC near local sunrise to be zero for simplicity. Carrano and Groves (2006) propose and apply a more detailed and accurate approach to determine the receiver bias. We point out that we are not interested in the accuracy of the VTEC method at this time but, instead, how ScintPi 3.0 results compared to those from PolaRx5S. The satellite differential code biases (DCB) were obtained from the Crustal Dynamics Data Information System (CDDIS). Conversion of slant TEC to VTEC was done using the Single Layer Model (SLM) mapping function (Klobuchar, 1987; Carrano and Groves, 2006):
$$VTEC = TEC cos\left[ {arcsin\left( {\frac{{R_{E} }}{{\left( {R_{E} + h} \right)}}{\text{cos}}\left( E \right)} \right)} \right]$$
(5)
where \(E\) is the elevation angle of the GNSS satellite with respect to receiver, \(h\) = 350 km is the thin shell height, and RE is the Earth's mean radius
Figure 9 shows an example of our results. The left-hand side panels show the VTEC derived from ScintPi 3.0 measurements of GPS (top) and GALILEO (bottom) signals. The right-hand side panels show the VTEC derived from the Septentrio signals. The most striking result is that the derived VTEC curves are virtually the same for signals from both monitors providing evidence of the usefulness of ScintPi 3.0 measurements in studies of the VTEC. Additionally, the measurements also show the same ionospheric behavior derived from measurements from distinct GNSS constellations, confirming the quality of the VTEC estimates provided by ScintPi 3.0. The GPS and GALILEO curves show maximum VTEC estimates of approximately 45 TECU around 1800 UT (~ 1500 LT) and minimum VTEC around 0800 UT (~ 0500 LT). The VTEC curves also show larger TEC variability during nighttime (2000 UT–0800 UT) which would be expected from the large latitudinal gradients in ionospheric density created by the pre-reversal enhancement (PRE) of the zonal electric field around sunset hours and by large electron density variations associated with equatorial plasma bubbles (Valladares et al. 2001). The results in Fig. 9 are also in good agreement with independent results presented by Okoh et al. (2021). Their estimates of TEC using u-blox F9P and comparisons with other non-collocated data sources led them to suggest that the receiver could be adequate for TEC studies.
Example of application
One of the main motivations of our efforts towards the development of low-cost sensors was to increase the accessibility to measurements that would allow the monitoring as well as fundamental and applied studies of ionospheric irregularities and scintillation. Here, we show measurements that demonstrate the potential of ScintPi 3.0 to distributed monitoring and measurements of ionospheric TEC perturbations and scintillation associated with equatorial plasma bubbles. More specifically, we now present simultaneous measurements made by three ScintPi 3.0 monitors deployed in South America, at the Jicamarca Radio Observatory (JRO), Presidente Prudente (PPR) and Campina Grande (CG). Figure 10 shows the spatial and temporal distribution of the scintillation measurements made by these monitors. It shows the S4 index on L1 signals (all constellations) as a function of ionospheric pierce point (IPP) coordinates. The color scale indicates the S4 value on the L1 signals.
The measurements in Fig. 10 serve to show that a low-cost deployment such as that provided by ScintPi 3.0 (and 2.0) can allow the monitoring and detection of scintillation and irregularities over a wide region. The example shows clear occurrence of scintillation in Campina Grande (CG) and Presidente Prudente (PPR). No obvious enhancements in L1 S4 could be observed at Jicamarca (JRO) and it was unclear from the scintillation observations alone that irregularities occurred. Closer inspection of S4 along with TEC measurements can provide additional information as it will be shown later in this section.
The example in Fig. 10 also shows that moderate scintillation occurs first in CG, around 23:00 UT. Moderate and intense scintillation starts to occur around 01:00 UT in PPR. The observations reflect the fact that CG is located closer to the magnetic equator and to the east of PPR. The sunset terminator crosses the magnetic longitude sector of CG first, triggering the EPBs earlier. These EPBs reach CG quickly since the site is located near the equator, causing the first scintillation events.
The fact that PPR is located to the west of and at a higher magnetic latitude compared to CG can explain the occurrence of scintillation at a later time. Conditions for EPB generation (sunset terminator) would take longer to occur in the longitude sector of PPR. It would also take longer for EPBs to develop vertically, map along magnetic field lines and reach the latitude of PPR compared to CG. Additionally, one can see that strong scintillation events occur more often in PPR. This can be explained by the location of the site. The severity of amplitude scintillation increases with the amplitude of the ionospheric density perturbations (Basu et al. 1976). PPR is located near the nominal location of the southern peak of the equatorial ionization anomaly where background densities are larger resulting in greater amplitudes of density perturbations. Those conditions are often invoked to explain the occurrence of stronger scintillation events near the equatorial anomaly peaks (de Paula et al. 2003b).
Figure 10 does not show moderate or strong scintillation events at Jicamarca. This can be caused by the absence of EPBs in the Western sector of South America or by the fact that ionospheric F-region densities near the equator are typically low leading to only weak or absent scintillation at L-Band signals. The TEC measurements provided by ScintPi 3.0 can provide additional information about the occurrence of EPBs.
Figure 11 shows an example of simultaneous measurements of TEC and scintillation made by the ScintPi 3.0 monitors at JRO (left panels), CG (middle panels) and PPR (right panels). The example serves to show signatures of ionospheric plasma depletions over the three sites. The depletions can be seen as large amplitude (several TECU) variations in TEC. The depletions are commonly accompanied by increases in scintillation activity (S4 values) and short-time, small-amplitude variations in TEC. At JRO, in particular, it can be seen that scintillation did occur but only weakly between 01:00 and 02:00 UT associated with a TEC depletion and irregularities. Examples of EPBs over CG and PPR can be seen at 03:00–04:00 UT and 00:00–02:00 UT, respectively, with more noticeable TEC depletions and steeper S4 increases. Therefore, the three monitors provide evidence of EPBs occurring throughout the American sector. Additionally, the measurements show that scintillation in the L2 signals (black curves) are stronger than the scintillations in the L1 signals (red curves). This is a result of amplitude scintillation scaling with frequency (Fremouw et al., 1978; Van Dierendonck et al. 1993; Carrano et al. 2014; Jiao and Morton, 2015).
The examples of measurements in Figs. 10, 11 serve to show that ScintPi 3.0 can aid studies of ionospheric irregularities and scintillation requiring distributed observations. While we do not envision ScintPi 3.0 fully replacing commercial receivers employed by arrays of distributed sensors such as the Low-latitude Ionosphere Sensor Network—LISN (Valladares and Chau, 2012) they can help to aid and expand these arrays.
One direct benefit of ScintPi 3.0 is the reduced cost of maintenance and replacement compared to commercial monitors. A defective commercial monitor might need to be sent back to manufacturer creating expenses related to shipping and repair and extended downtime. ScintPi 3.0 can simply be fully replaced reducing costs and downtime.