© I.G. Wright et al., Published by EDP Sciences 2023

1 Introduction

The ionosphere is a region of Earth’s upper atmosphere (~50–1000 km in altitude) that is characterized by a relatively high density of free ions and electrons created, for the most part, by solar photoionization. The ionosphere can also be described as a magnetized, weakly ionized plasma (Stubbe & Hagfors, 1997).

Studies of the ionosphere are motivated by fundamental questions related to how its thermal structure, composition, and density respond to variations in the sun and the lower atmosphere (Burns et al., 2007; Nishioka et al., 2013; Komjathy et al., 2016). The study of the Earth’s ionosphere is also motivated by its impact on radio signals used by various civilian and military applications. The ionosphere is a diffractive medium whose permittivity is heavily controlled by electron density. Radio waves of different frequencies propagate at different speeds through the ionospheric plasma. Global navigation satellite systems (GNSS) such as GPS rely on estimating the travel time of trans-ionospheric radio signals for determining location. Therefore, spatiotemporal variations in the ionospheric delay represent a challenge for these systems.

In addition to refraction, ionospheric plasma also affects radio signals through diffraction. Irregularities in the ionospheric plasma density are often observed particularly at low and high latitudes (Fejer & Kelley, 1980). The scale size of these irregularities varies from centimeters up to hundreds of kilometers (Kelley et al., 1982). These irregularities cause the diffraction of radio waves and, as a result, variations in the amplitude and/or phase of signals detected by a stationary or moving receiver (Yeh & Liu, 1982). These variations in amplitude and/or phase of radio signals caused by ionospheric irregularities are referred to as ionospheric scintillation (Kintner et al., 2007). Ionospheric scintillation is known to affect telecommunications (Basu et al., 1988), L-band SAR remote sensing (Carrano et al., 2012), and the performance of GNSS receivers (Zhang & Morton, 2009). Variations in the ionospheric plasma capable of impacting technological systems are an important component of space weather (Bhattacharyya, 2022).

It has been recognized that fundamental and applied studies of the Earth’s ionosphere would greatly benefit from distributed arrays of small instruments (DASI) or sensors (National Research Council, 2006). There are, however, obvious challenges associated with the deployment of these sensors. For instance, in many cases, observations are needed in remote locations where electricity and the Internet are not readily available. Additionally, support for the operation and maintenance of the sensors can be limited.

At the same time, there has been a growing interest by citizens in participating in scientific studies which could help the distribution of ionospheric sensors and advances in the understanding of space weather (Knipp, 2015). An obstacle citizen science initiatives face is that sensors typically require the hosts to provide Internet and power. These dependencies may bring concerns about cost and privacy for hosts. Additionally, the sensors must be easy to install and require minimal maintenance by the citizen scientist. Therefore, ionospheric observations would greatly benefit from autonomous sensors that are easy to install and maintain.

While distributed sensor observations face challenges such as those listed above, they also offer opportunities for integrating research and education and increasing literacy about space weather. Here, we present the results of a project-based learning (Bell, 2010) initiative aimed at broadening the breadth of skills of undergraduate physics students at The University of Texas at Dallas (UTD) and themed around space weather. More specifically, we present the results of a student-driven and instructor-facilitated project that led to the design, development, and tests of an autonomous platform for ionospheric observations.

This report is organized as follows: In Section 2 we describe the design of the platform and provide information about the low-cost sensor (ScintPi 3.0) used for measurements of ionospheric total electron content (TEC) and scintillation. In Section 3 we provide information about the prototype platform built as part of the project and about its deployment in the field. In Section 4 we present and discuss the results of measurements made with the platform. Finally, in Section 5 we provide our main conclusions and plans for the future.

2 Instrumentation: design considerations

As mentioned earlier, the work and results described in this report were obtained as part of a project led by undergraduate students. The project had as main objectives: a) to create new experimental capabilities for fundamental and applied observations related to space weather and b) to provide educational and professional training opportunities to undergraduate physics students.

A team of three undergraduate physics students was given a discretionary research budget and was responsible for the proposal, research and design, assembly, and testing of their platform. As part of the proposal, students listed the main research tasks and developed a tentative schedule that was evaluated by the instructor. After the proposal phase, tasks were distributed, and students met regularly to report progress. The students also met regularly with a supervising instructor and graduate student to report team progress and discuss potential findings or challenges.

2.1 ScintPi 3.0

The design of the platform takes advantage of ScintPi 3.0, a small, low-power, and low-cost ionospheric sensor developed by Gomez Socola & Rodrigues (2022). ScintPi 3.0 can measure L-Band ionospheric scintillation and Total electron content (TEC) using signals from Global Navigation Satellite System (GNSS) such as GPS. A copy of ScintPi 3.0 is shown in Figure 1 and has been built using a Raspberry Pi single-board computer and an off-the-shelf dual-frequency GNSS receiver.

Figure 1 ScintPi 3.0: The low-cost ionospheric scintillation and total electron content (TEC) monitor used in this study.

Leading commercial scintillation and TEC monitors can cost several thousands of dollars, but Gomez Socola & Rodrigues (2022) list the cost of parts for ScintPi 3.0 below US$600.00. Because of its lower cost, ScintPi 3.0 presents an opportunity to deploy more receivers in spaced array studies and citizen science initiatives; however, in both distributed array studies and citizen science initiatives, ScintPi 3.0 requires a constant source of electricity and Internet to verify the status of the receiver and/or transfer data. Therefore, applications involving ScintPi 3.0 would benefit from a platform that mitigates these external dependencies.

2.1.1 ScintPi 3.0 measurements

ScintPi 3.0 is capable of making high sampling rate (up to 25 Hz) measurements of amplitude scintillation in GNSS signals at two frequencies (~1.2 and ~1.6 GHz). These signals are transmitted by the following GNSS: GPS, Galileo, GLONASS, BeiDou, and SBAS. Scintillation severity is measured by the traditional S ₄ index, which is commonly used in fundamental and applied studies of scintillation (Yeh & Liu, 1982):$$S_4=\sqrt{\frac{\langle I^2\rangle -{\langle I\rangle }^2}{{\langle I\rangle }^2}}$$(1)where I represent the amplitude of the GNSS signal and the angle brackets represent the ensemble average. In practice samples measured over 1 min is used to estimate S ₄. S ₄ can be described as the standard deviation of signal amplitude normalized by its average.

ScintPi 3.0 also provides TEC estimates using measurements of pseudo-ranges and phases from GNSS signals (Jakowski, 1996). The resulting TEC observations are referred to as code TEC (TEC _ρ ) and phase TEC (TEC _ϕ ). Phase information provides precise, though relative, TEC. Code information, on the other hand, provides noisy but absolute TEC estimates. Therefore, code TEC can be used to adjust the magnitude of phase TEC (Klobuchar, 1996).

2.2 Autonomous platform

In the design of the platform, two specific goals were defined: 1) to develop a platform capable of running ScintPi 3.0 without external power and 2) to develop a platform capable of remote connectivity without host-provided Internet. We consider connectivity as the ability of the system to report that the system is operating and that measurements are being collected.

2.2.1 Overall design

With respect to powering the system, different modes were considered. First, a battery-only powered mode was taken into consideration. In this case, the system is powered solely by a discharging battery for an extended period of several days. This mode would allow short campaigns of observations using, for instance, locally spaced monitors. The second powering mode taken into consideration would use solar energy. This mode would consist of a solar panel charging a battery that powers the system. Solar was determined to be the mode of recharging given the ubiquity of sunlight and the relatively low cost of photovoltaic panels. Both powering modes would be regulated by an inexpensive pulse-width modified solar charge controller designed to prevent battery overcharging, regulate voltage, and prevent reverse current flow.

With respect to connectivity, options for remote monitoring of the status of the system were considered. Although ScintPi 3.0 can run independently of Internet access, connectivity with the receiver allows a researcher to ensure that their system is making continuous measurements in remote deployments or critical applications. An LTE cellular modem was chosen as the method of supplying the system with Internet access because of the international growth in cellular coverage and the availability of cellular support for the Raspberry Pi provided by several companies. Currently, ScintPi 3.0 targets high-rate measurements, and the daily data files are somewhat large (at least 1.5 GB). While transmitting the raw data or at least scintillation indices are desired, we limited connectivity to providing the status of the system. This is, in part, due to an expectation that some locations only have older mobile networks or that some locations can only offer newer, faster mobile networks at a relatively high cost. Therefore, the design considered that files would be written on a high-capacity SD drive and retrieved after a sufficient period of observation. The connectivity, however, would allow users to confirm data acquisition and to perform basic software updates or troubleshooting if needed via shell access. Figure 2 summarizes the overall design of the system. The proposed autonomous system allows two powering modes (battery only and solar) and connectivity to a central server at UTD via an LTE cellular connection.

Figure 2 Diagram illustrating the overall design of the autonomous platform. We note that the solar panel would not be needed in the case of short-term (days) observations where a fully charged battery would be sufficient.

2.2.2 Battery

The system, including ScintPi 3.0, external GNSS antenna, and the cellular modem, consumes about 3.0 W corresponding to 72 Wh over the course of a day. In our design, we took into consideration a Depth of Discharge (DoD) of 80% for the battery, i.e. 80% of the total rated capacity. Additionally, the design also considered that the system would have to operate for at least 7 days allowing regular weekly inspections and recharges.

Therefore, for a nominal voltage of 12 V and an active period of 7 days, a battery with a capacity of 55 Ah was chosen where:$$\left(0.8\cdot {\mathrm{C}}_{\mathrm{Battery}}\right)\cdot 12\mathrm{V}=72\frac{\mathrm{Wh}}{\mathrm{day}}\cdot 7\mathrm{days} \Rightarrow {\mathrm{C}}_{\mathrm{Battery}}=52.5\mathrm{Ah}$$(2)where C _Battery is the charge capacity of the battery (in Ah), and the 0.8 multiplying factor comes from a DoD of 80%.

Although lithium-based batteries have advantages with respect to size, weight, and lifetime, a sealed led acid battery (SLA) was chosen based on cost. At the time, it was found that a 55 Ah battery costs ~150 USD less than a lithium battery. The SLA battery was connected to a Pulse Width Modified (PWM) charge controller. While Maximum Power Point Tracking (MPPT) charge controllers offer more charging efficiency than PWM controllers, especially in larger solar arrays, a PWM controller was chosen because of the relatively small load size and the fact that they cost less than MPPT controllers by about 70 USD or more. The controller regulated the charging of the battery by the solar panel and provided power to ScintPi 3.0.

2.2.3 Solar panel

To determine the size of the solar panel required by the platform, we considered the daily power consumption of the autonomous system and relevant parameters related to weather, dust, positioning, and season. Considering four peak sun hours (the lowest monthly average for Texas) and a conservative 50% efficiency factor, a solar panel would need to provide 36 W where:$$(0.5\cdot P_{\mathrm{Solar}})\cdot 4\frac{\mathrm{hours}}{\mathrm{day}}=72\frac{\mathrm{Wh}}{\mathrm{day}}\Rightarrow P_{\mathrm{Solar}}=36\mathrm{ }\mathrm{W}$$(3)where P _Solar is the nominal power provided by the solar panel and the 0.5 multiplying factor represents efficiency. A monocrystalline solar panel was chosen over a polycrystalline solar panel due to higher efficiency and performance and over a thin film due to a more rigid structure. The battery considered for this mode was the same 55 Ah battery described for the battery-only mode in the previous section. The battery and solar panel were connected in parallel with the charge controller to prevent overcharging and reverse current flow. Being deployed in the northern hemisphere, the solar panel was deployed facing south at an optimal angle of 28° off-zenith as recommended by the manufacturer.

2.2.4 Remote connectivity

For connectivity, we selected a 4G/LTE cellular modem module provided by Sixfab (https://sixfab.com). While several USB 4G/LTE modems were considered (e.g. Huawei, Qualcomm), this module was chosen based on its plug-and-play Raspberry Pi support and relatively low cost. Alternative 4G/LTE modems would need to demonstrate software compatibility with the Debian-based Raspberry Pi OS which may vary depending on the model. Sixfab provides several native subscriber identity module (SIM) card data plans. The lowest price option was selected. This plan provides 500 MB of data download and uploads per month at 9 USD. Additionally, the Sixfab module provides software for remote shell access, allowing researchers to connect to the system. This would enable remote troubleshooting and monitoring.

Given that the raw data would not be transmitted, status updates would need to include information not only related to whether the system was online but also metadata related to the acquisition of new data. For that purpose, Python scripts were developed that transmit battery capacity measurements and information about the data files to a server at UTD every hour. A watchdog script was also developed to issue an email warning if an update has not been recently uploaded, or if the status files indicated that the system was not currently writing data.

While transmitting the full contents of the data files was not possible with the selected data plan, real-time data files containing satellite positional information were transmitted to a server to prototype the capability of real-time data streaming.

3 Prototype and deployment

A prototype of the autonomous platform was built and deployed in an area adjacent to the UTD main campus. The system operated in this location for tests between April 2022 to November 2022.

Figure 3a shows the electrical components of the autonomous platform prototyped in the lab. The picture shows an open version (without a case) of ScintPi 3.0, the solar charge controller, the battery, and the cellular modem. The picture also shows a small circuit that consisted of a voltage divider and an analog-to-digital converter (ADC) that was built to read battery voltage (related to charge). Figure 3b shows the exterior of the autonomous platform prototype deployed in the field location. Components are properly secured within a waterproof junction box. The main structure of the platform consists of a standard antenna mount on which the following components are attached: GNSS antenna, solar photovoltaic panel, and the waterproof junction box containing the electronics seen in Figure 3a.

Figure 3 (a) An open view of the electronic components that are part of the autonomous platform as it was prototyped in the lab. (b) View of the platform deployed for field tests near UTD.

To determine the performance of the autonomous platform, we analyzed measurements of battery voltages which serve as an indicator of battery charge. Battery voltages for SLA batteries are roughly proportional to the percentage of charge the battery contains (UPG, 2022). Therefore, voltage measurements in parallel with ScintPi 3.0 observations can demonstrate that the system was collecting data, and the platform was actively supplying power to the sensor. Any discontinuities in the voltage observations (and ScintPi 3.0 data) would have indicated that the system stopped observations. We deployed the system in the experimental setup described above, and the status updates enabled by the LTE connection allowed near real-time tracking of the experiment.

To test the efficacy of the battery-powered mode, a fully charged battery was connected to the platform, and the system was allowed to operate until failure. To test the solar-powered mode, a fully discharged battery and a solar panel were connected to the platform, and the system was allowed to operate continuously until failure. We did not find any interruption in the measurements after two months of operating in the solar-powered mode and the system was brought back to the lab.

4 Results and discussion

4.1 Observations using battery only

The first results we present and discuss are related to our analyses of how long the 55 Ah SLA battery could power the autonomous platform and the extent to which low battery voltages could impact the quality of observations. Figure 4 summarizes the performance of the autonomous platform as the battery decreased from full capacity until the charge controller disconnected the device.

Figure 4 (a) Temporal variation of battery voltage (charge). (b) S 4 index for all the 1.6 GHz GNSS signals transmitted by satellites with an elevation greater than 25°. (c) Absolute vertical TEC derived from code and phase measurements provided by the platform and from the MIT Madrigal TEC maps. (d) Linear correlation analyses between phase TEC values measured by the platform and those provided by the MIT Madrigal maps. Note that 1 TECU corresponds to 1 × 1016 electrons/m2 or 0.16 m delay error in the GPS L1 = 1.575 GHz signal.

Figure 4a shows the terminal-to-terminal voltage of the battery over a period of nine days from 18:46 UT on 10-03-22 to 9:02 UT on 10-13-22. In total, the autonomous platform powered by the battery ran for 232 h. At the 0 hours mark (18:46 10-03-22 UTC), the battery was fully charged, corresponding to a voltage of 13.3 V. At the 232 h mark (9:02 10-13-22 UTC), the voltage measured 11.3 V, which, although not fully discharged, corresponds to the low voltage cutoff threshold for the charge controller. At that point, power was no longer provided through the charge controller to the ScintPi 3.0, and the observations were interrupted. Figure 4 shows that the choice of the battery allows for measurements beyond the target of 7 days. Almost 10 days of continuous operation was possible.

Figure 4b shows scintillation measurements for the 1.6 GHz signal made by the system while being powered by the battery. The subplot shows the S ₄ values for all satellites in view above an elevation mask of 25°. The platform was deployed near a building and to a parking lot and, therefore, some multipath effects were expected and observed. Nevertheless, the S ₄ values predominantly remained below 0.2 which is not surprising for a mid-latitude observation site. Ionospheric scintillation is known to occur frequently and at low and high latitudes, but cases of mid-latitude scintillation are expected to occur much less often (Aarons, 1982).

Figure 4c shows the absolute vertical code and phase TEC observations made by the system. Similar to the procedure described by Socola Gomez & Rodrigues (2022), satellite differential code biases (DCBs) provided by NASA’s Crustal Dynamics Data Information System (CDDIS) were used to correct TEC estimates. Vertical code TEC (VTEC) is computed using a mapping function (Carrano & Groves, 2006) for all the slant code TEC measurements made by the system for satellites with an elevation greater than 25°. The receiver bias is estimated by matching minimum VTEC values reported by the platform to minimum VTEC values reported by Madrigal VTEC maps (Rideout & Coster, 2006) for the closest location over the period of comparison. The Madrigal VTEC maps are generated using a large number of global measurements of TEC made by a distributed network of geodetic receivers and are widely used in ionospheric TEC studies (Rideout & Coster, 2006). The Madrigal VTEC maps selected have a resolution of 1 degree in latitude and longitude. Figure 4c also shows the Madrigal VTEC for the grid point located closest to the platform. The subplots show that VTEC variations measured by the platform match extremely well the VTEC variations provided by Madrigal providing additional evidence that the battery-powered system worked as expected. For completeness, Figure 4d provides results of correlation analyses between the platform phase VTEC values (corrected by code TEC) and Madrigal VTEC for the time series shown in Figure 4c. The results confirm the excellent agreement between the two sources of TEC values despite different temporal and spatial resolutions. The analyses show a coefficient of linear correlation (Pearson) R = 0.97 and a bias value β = −1.38 TECU. The bias value refers to the mean deviation of the platform values with respect to Madrigal, that is, β = 〈VTEC_Platform − VTEC_Madrigal〉, where the angle brackets represent averaging.

The results in Figure 4 demonstrate how well the battery-powered system works and that it meets the requirements of the proposed design. This powering mode would not be adequate for long-term observations, but this mode may be more suited for experiments using multiple sensors deployed locally. For instance, the system is already planned to be used in spaced-receiver scintillation experiments (e.g., Kil et al., 2000) where receivers spaced by distances of a few 100s of meters to a few km are needed. These experiments also require monitors to be placed at exact locations making the availability of adequate sites and electricity less likely. For instance, investigations of the motion of low latitude irregularities associated require the placement of receivers in the magnetic zonal (East–West) direction.

4.2 Observations using battery and solar panel

Figure 5 now summarizes the performance of the autonomous platform after being deployed on April 10, 2022, at 18:55 UT (13:55 LT) with a completely discharged battery connected to the charge controller and equipped with a solar panel.

Figure 5 (a) Temporal variation of battery voltage (charge). The lines in gray show abnormal voltages during daytime hours when the panel was also charging the battery. (b) S 4 index for all the 1.6 GHz GNSS signals transmitted by satellites with elevation greater than 25°. (c) Absolute vertical TEC derived from code and phase measurements provided by the platform and from the MIT Madrigal TEC maps. Note that 1 TECU corresponds to 1 × 1016 electrons/m2 or 0.16 m delay error in the GPS L1 = 1.575 GHz signal.

Figure 5a shows the temporal variation of the battery voltage. The discharged battery was connected to the controller at the 0 h mark. After 26 h (20:49 UT on 04-11-22), the battery was sufficiently charged by the solar panel and the charge controller reconnected the sensor to the battery. A fully charged battery would output a nominal voltage of 13.3 V. We must point out that the analog-to-digital converter used to measure the battery voltage is connected in parallel with the solar panel and battery. Therefore, any values above 13.3 V in Figure 5a are not representative of battery charge. They are artifacts caused by the solar panel providing charge to the battery during daytime hours. To emphasize the measurements that reflect battery capacity, voltage measurements made at night (between 20:00 and 05:00 LT) are plotted in black, and voltage measurements during the day are plotted in grey. At ~90% charge, the battery would output a voltage of 12.9 V. At 0% charge the battery would output a voltage of 11.3 V or less.

Figure 5b shows again that S ₄ values are mostly under 0.2 with a few cases of multipath observed from time to time. The platform was placed in the same location as the tests with battery-only and described in the previous section. While the platform operated continuously for 2 months until it was brought back to the lab, only 18 days (432 h) of the experiment were shown so that the reader can better visualize the measurements.

Figure 5c shows the VTEC measurements in the same format as Figure 4c. Again, the platform VTEC observations match extremely well the Madrigal VTEC for the entire period. There are, however, a few interesting variations and differences between the platform and Madrigal VTEC in this set of observations that were not present in the battery-only experiment shown in Figure 4. More specifically, on the fourth day (between hours 72 and 98 corresponding to April 14, 2022) a large increase in VTEC is observed by the platform as well as by the Madrigal TEC map. The platform shows a maximum VTEC of about 75 TECU while Madrigal shows a maximum of about 60 TECU. We believe that the difference is most likely caused by the fact that the Madrigal VTEC value for a given grid point is computed using measurements made by various GNSS receivers around that grid location. Therefore, given that the TEC maps have a temporal resolution of 5 min and spatial resolution of 1° (in latitude and longitude), their VTEC values would represent spatiotemporal averages for the grid locations. The comparison between the platform and VTEC map values shown here highlights the benefits of measurements provided by a single station as it could be more representative of the VTEC around the location being observed.

4.3 TEC enhancement on April 14, 2022

While outside the scope of this report, we provide some insight into the interesting event of TEC enhancement that was observed on April 14, 2022, during the tests of the autonomous platform (see Fig. 5).

Figure 6 provides additional context for the observed TEC enhancement of April 14 by showing Madrigal VTEC maps over the US for a few specific hours, 15:00, 19:00, and 23:00 UT on April 13, 14, and 15. The red star marks the location of the autonomous platform. Figure 6 shows that substantial TEC enhancements occurred over the US on April 14 compared to the previous and following days.

Figure 6 MIT Haystack’s Madrigal vertical TEC (VTEC) maps over the United States on April 13, 14 and April 15, 2022 (columns) at 15:00, 19:00, and 23:00 UT. The location where the autonomous platform was deployed is indicated by a red star. The maps show that an enhanced TEC occurred over the US on April 14. The enhancement is more clearly seen on the map for 19:00 UT.

Enhancements in TEC with respect to typical values at mid-latitudes have been observed in the past (Schunk & Sojka, 1996). These enhancements have been referred to as positive ionospheric storms and are commonly associated with geomagnetic storms which can be described as disturbances in the Earth’s magnetosphere that result from the input of energy from the sun through the solar wind. This input of energy into the magnetosphere can affect the Earth’s thermosphere and ionosphere. For instance, magnetospheric particles can precipitate and magnetospheric electric fields can map to the high-latitude ionosphere. This can lead to enhanced ionospheric currents and Joule heating which in turn affect the composition and global circulation of neutral winds in the upper atmosphere as well as the behavior of ionospheric electric fields over a wide range of latitudes and longitudes. Changes in composition, wind circulation, and ionospheric E  ×  B drifts affect production and recombination rates leading to enhancements in ionospheric F-region densities and TEC. Unfortunately, the thermosphere and ionosphere respond differently to different geomagnetic storms, and the relative contribution of various processes involved are difficult to quantify due, in great part, to a lack of distributed observations, and motivation for the present work. Nevertheless, geomagnetic activity and the occurrence and magnitude of geomagnetic storms are commonly quantified by geomagnetic indices (Menvielle et al., 2011).

Figures 7a and 7b show again scintillation and VTEC observations, respectively, made by the autonomous platform prototype between April 13 and 15, 2022. This is an expanded view of the observations presented in Figure 5 that better show the enhancement in VTEC observed on April 14, 2022. The observations also better show that L-Band scintillation did not accompany this event. While scintillations are not frequently observed in the southern US, cases have been observed and reported (Rodrigues et al., 2021). Figures 7c and 7d show the Planetary K (Kp) and the Disturbance Storm-Time (Dst) indices for the period of interest. These indices utilize ground-based magnetometer measurements at mid (Kp) and low latitudes (Dst) and are commonly used to quantify geomagnetic storms. Figures 7c and 7d show significant increases in the Kp index and a reduction in the Dst index on April 14 which confirm that the observed enhancement in TEC was indeed associated with a geomagnetic storm whose effects were detected over a wide range of latitudes. More specifically, Kp rose to a value of 6 between 15 and 18 UTC, corresponding to a moderate geomagnetic storm following the NOAA scale for geomagnetic storms (NOAA, 2023).

Figure 7 An expanded view of the observations between April 13 and 15, 2022. (a) S 4 index for all the 1.6 GHz GNSS signals transmitted by satellites with elevation greater than 25°. (b) Absolute vertical TEC derived from code and phase measurements provided by the platform and from the MIT Madrigal TEC maps. (c) Planetary K (Kp) index. (d) Disturbance Storm-Time (Dst) index. The yellow shade on the panels shows the region of enhanced TEC which coincides with a region of enhanced Kp and low Dst values.

5 Concluding remarks

We presented and discussed the results of a student-led project with the goals of a) creating new experimental capabilities for fundamental and applied observations related to space weather and b) providing educational and professional training opportunities to undergraduate physics students. Below we provide concluding remarks related to these two goals.

5.1 On creating new experimental capabilities for space weather observations

With respect to the first goal, we presented the design, construction, deployment, and measurements made by an autonomous platform for ionospheric observations. The platform uses low-power, low-cost ionospheric scintillation and a TEC monitor (ScintPi 3.0) for observations. The platform allows observations to be made at locations where electrical power and the Internet are not readily available.

We investigated two modes of powering the autonomous platform: a solar-powered mode and a battery-only mode. In the battery-only mode, the autonomous platform ran continuously from a fully charged SLA battery for 232 h while making measurements. The advantages of this powering mode are a lower cost and an easier installation for short-term (~7 days or less) campaigns. A researcher can deploy these smaller, lightweight platforms, and retrieve the systems several days later or replace the batteries. Even as the battery approaches full discharge, the system remains online and its observations unperturbed.

In the solar-powered mode, tests showed continuous power and operation in a mid-latitude deployment. Adequate measurements were made over more than 2 months of operation when the station was brought back from the field to the lab. The quality of the measurements was attested by the ability of the system to measure an event of TEC enhancement associated with a geomagnetic storm. The absolute VTEC values measured by the platform were in excellent agreement with VTEC independently provided by the MIT Haystack VTEC maps. Across the long-term behavior of the autonomous system, the depth of charge never fell below ~85%, even during rainy or cloudy weather. We also notice that the system can restart itself from a depleted state in 26 h and reaches full charge in 96 h.

The solar-powered mode would be intended for long-term or permanent campaigns. For future work, we note that in the solar-powered mode, the platform utilizes the same battery as the battery-only mode; however, given that the system is recharged daily, the size of the battery can be greatly reduced. Moreover, a lightweight and smaller lithium battery may be employed to reduce the total system size greatly. This would the system to be more easily installed in a wider range of sites such as a street or fence poles and with smaller mounts.

For connectivity, we used Sixfab’s 4G/LTE cellular modem module. This module was chosen based on its relatively low cost and its Raspberry Pi support. Additionally, the Sixfab module provides software for remote shell access, allowing researchers to connect to the system as it is powered. Throughout the field tests, the prototype successfully transmitted regular status files and enabled remote access. Further work may be done to expand the capabilities of the platform’s cellular transmission. For instance, it may be possible to equip the device with an unlimited data SIM card. Adopting an unlimited data plan would support the entire transmission of the raw data files from a receiver to the processing server. The solar-battery-powered mode unlimited data via cellular transmission would enable a true autonomous design, wherein a researcher would only need to travel to the device for repair in the event of a malfunction.

In addition to research, the autonomous platform can also be used for public outreach and citizen science initiatives. For instance, there has been interest in citizen scientists to host ScintPi 3.0 but with some concern about providing access to the Internet. The platform would overcome this challenge. We also envision using measurements made by the platform to increase literacy about global navigation satellite systems and space weather.

5.2 On educational and professional training

With respect to educational and professional training, we present the development of the autonomous platform derived from project-based learning (Bell, 2010) initiative aimed at broadening the breadth of skills of undergraduate physics students at The University of Texas at Dallas (UTD) and themed around space weather.

The project provided students with opportunities to learn and gain experience with electronic circuits, computer programming, data visualization, and hardware/software interface. For instance, the project required students to design and implement an analog-to-digital converter (ADC) for measurements of battery discharge. The project also required students to design and assemble the hardware securing the antenna, solar panel, and housing the electronics. Additionally, the project required students to write several scripts responsible for managing the data files, communicating with the data server at UTD, interfacing with the cellular modem, interfacing with the ADC, etc. Project tasks also required students to learn more about data visualization and space weather.

Additionally, the project provided students with an opportunity to develop soft skills such as teamwork, communication, time management, creativity, and decision-making. For instance, students met regularly to distribute tasks among team members, make decisions related to the design of the platform, report progress, find solutions to design problems and discuss results. The students were also encouraged to make technical presentations of their projects and results.

Finally, several factors indicate the success of this project with respect to engagement and education. For instance, a poster presentation of the project by the students was recognized with an undergraduate award. Additionally, the senior undergraduate student that was involved in the project is now pursuing a Ph.D. related to space weather at low latitudes using observations made by autonomous platforms.

Acknowledgments

Work at UTD has been supported by the Eugene McDermott Fellowship, by NSF Award AGS-1554926 and by NSF’s Graduate Research Fellowship Program (GFRP). This material is based upon work supported by the National Science Foundation Graduate Research Fellowship Program under Grant No. (2136516). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. The editor thanks an anonymous reviewer for his/her assistance in evaluating this paper.

Data availability

The scintillation and TEC measurements presented in this work are available in a Zenodo repository (https://zenodo.org/record/7829808#.ZFPW2y-B35i).

Kp data is provided by NOAA Space Weather Prediction Center (https://www.swpc.noaa.gov). Dst data is provided by the World Data Center for Geomagnetism, Kyoto (https://wdc.kugi.kyoto-u.ac.jp).

TEC map data can be obtained from the Madrigal database (http://madrigal.haystack.mit.edu). GPS TEC data products and access through the Madrigal distributed data system are provided to the community by the Massachusetts Institute of Technology under support from US National Science Foundation grant AGS-1952737. Data for the TEC processing is provided from the following organizations: UNAVCO, Scripps Orbit and Permanent Array Center, Institut Geographique National, France, International GNSS Service, The Crustal Dynamics Data Information System (CDDIS), National Geodetic Survey, Instituto Brasileiro de Geografia e Estatística, RAMSAC CORS of Instituto Geográfico Nacional de la República Argentina, Arecibo Observatory, Low-Latitude Ionospheric Sensor Network (LISN), Topcon Positioning Systems, Inc., Canadian High Arctic Ionospheric Network, Institute of Geology and Geophysics, Chinese Academy of Sciences, China Meteorology Administration, Centro di Ricerche Sismologiche, Système d’Observation du Niveau des Eaux Littorales (SONEL), RENAG: REseau NAtional GPS permanent, GeoNet - the official source of geological hazard information for New Zealand, GNSS Reference Networks, Finnish Meteorological Institute, SWEPOS – Sweden, Hartebeesthoek Radio Astronomy Observatory, TrigNet Web Application, South Africa, Australian Space Weather Services, RETE INTEGRATA NAZIONALE GPS, Estonian Land Board, Virginia Tech Center for Space Science and Engineering Research, and Korea Astronomy and Space Science Institute.

References

All Figures

	Figure 1 ScintPi 3.0: The low-cost ionospheric scintillation and total electron content (TEC) monitor used in this study.
In the text

	Figure 2 Diagram illustrating the overall design of the autonomous platform. We note that the solar panel would not be needed in the case of short-term (days) observations where a fully charged battery would be sufficient.
In the text

	Figure 3 (a) An open view of the electronic components that are part of the autonomous platform as it was prototyped in the lab. (b) View of the platform deployed for field tests near UTD.
In the text

Figure 4 (a) Temporal variation of battery voltage (charge). (b) S 4 index for all the 1.6 GHz GNSS signals transmitted by satellites with an elevation greater than 25°. (c) Absolute vertical TEC derived from code and phase measurements provided by the platform and from the MIT Madrigal TEC maps. (d) Linear correlation analyses between phase TEC values measured by the platform and those provided by the MIT Madrigal maps. Note that 1 TECU corresponds to 1 × 1016 electrons/m2 or 0.16 m delay error in the GPS L1 = 1.575 GHz signal.

In the text

Figure 5 (a) Temporal variation of battery voltage (charge). The lines in gray show abnormal voltages during daytime hours when the panel was also charging the battery. (b) S 4 index for all the 1.6 GHz GNSS signals transmitted by satellites with elevation greater than 25°. (c) Absolute vertical TEC derived from code and phase measurements provided by the platform and from the MIT Madrigal TEC maps. Note that 1 TECU corresponds to 1 × 1016 electrons/m2 or 0.16 m delay error in the GPS L1 = 1.575 GHz signal.

In the text

	Figure 6 MIT Haystack’s Madrigal vertical TEC (VTEC) maps over the United States on April 13, 14 and April 15, 2022 (columns) at 15:00, 19:00, and 23:00 UT. The location where the autonomous platform was deployed is indicated by a red star. The maps show that an enhanced TEC occurred over the US on April 14. The enhancement is more clearly seen on the map for 19:00 UT.
In the text

Figure 7 An expanded view of the observations between April 13 and 15, 2022. (a) S 4 index for all the 1.6 GHz GNSS signals transmitted by satellites with elevation greater than 25°. (b) Absolute vertical TEC derived from code and phase measurements provided by the platform and from the MIT Madrigal TEC maps. (c) Planetary K (Kp) index. (d) Disturbance Storm-Time (Dst) index. The yellow shade on the panels shows the region of enhanced TEC which coincides with a region of enhanced Kp and low Dst values.

In the text