the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
The deployment of a geomagnetic variometer station as auxiliary instrumentation for the study of Unidentified Aerial Phenomena
Abstract. Witness reports of Unidentified Aerial Phenomena (UAP) occasionally associate UAP sightings with local electromagnetic interferences, such as spinning magnetic compasses onboard aircraft or sudden malfunctions of mechanical vehicles. These reports have motivated the incorporation of a magnetometer into the instrumentation suite of the Galileo Project (GP), a Harvard-led scientific collaboration whose aim is to collect and analyze multi-sensor data that collectively could help elucidate the nature of UAP. The goal of the GP magnetometry investigation is to identify magnetic anomalies that cannot be readily explained in terms of a natural or human-made origin, and analyze these jointly with the data collected from the other modalities. These include an ensemble of visible and infrared cameras, a broadband acoustic system and a weather-monitoring system. Here, we present GP’s first geomagnetic variometer station, deployed at the GP observatory in Colorado, USA. We describe the calibration and deployment of the instrumentation, which consists of a vector magnetometer and its data acquisition system, and the collection and processing of the data. Moreover, we present and discuss examples of the magnetic field data obtained over a period of 6 months, including data recorded during the May 2024 G5 extreme geomagnetic storm. We find that the data meet and even surpass the requirements laid out in GP’s Science Traceability Matrix. Key to the evaluation of our data is the proximity of the variometer station to the USGS magnetic observatory in Boulder, Colorado. By comparing the two sets of data, we find that they are of similar quality. Having established the proper functioning of the first GP variometer station, we will use it as the model for variometer stations at future GP observatories.
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-3431', Anonymous Referee #1, 22 Jul 2025
-
AC1: 'Reply on RC1', Foteini Vervelidou, 03 Oct 2025
We thank the Reviewer for reading and commenting on our manuscript. We provide our replies below each comment, shown in italics.
This is a very well written manuscript. Although I will be surprised if the magnetometer system is ever used to provide insightful data regarding UFOs, the data themselves will be valuable for a wide variety of applications. I have a few questions, which I hope the authors will address.
We thank the reviewer for the positive feedback about the manuscript, and we are glad to hear that our data can be of use to the community.
- I note that the air-conditioning system at the BOU observatory was not working during calibration. Since the authors are set up near that observatory, are there plans to revisit the BOU observatory and investigate system response over a wider range of temperatures?
The specific site discussed in this manuscript is no longer operational. As we mention in Section 6 of the manuscript, the magnetometer remained deployed only until September 26, 2024.
Given the temperature stability achieved by means of the insulation measures we took and the fact that we are using the magnetometer as a variometer and not for absolute measurements, we do not anticipate the need for calibrating our magnetometer over a wider range of temperatures. Nevertheless, we will keep monitoring the temperature variations and if need be we will seek ways to calibrate it at the closest available facility.
- Are there plans to make the data available to the scientific community in the style of (say) Intermagnet or SuperMag? This is important. Other investigators will make use of the data, and, in the end, the project represented in this manuscript will find indirect support from the wider scientific community.
Yes, all our data are meant to be publicly shared. We have already uploaded at Zenodo all the raw magnetic field and temperature data collected over the days discussed in the manuscript, and we are happy to share the rest of our data with anyone interested. Currently, due to limited resources, we cannot make all data directly available. By default, we process the data we consider interesting based on a variety of factors. For example, in the case of this manuscript, we considered interesting the data that were representative of the performance of the magnetometer. But we are happy to allocate resources to respond to specific data requests.
Citation: https://doi.org/10.5194/egusphere-2025-3431-AC1
-
AC1: 'Reply on RC1', Foteini Vervelidou, 03 Oct 2025
-
RC2: 'Comment on egusphere-2025-3431', Anonymous Referee #2, 01 Sep 2025
The paper by Vervelidou et al. on “The deployment of geomagnetic variometer station as auxiliary instrumentation for the study of unidentified aerial phenomena” was an interesting read for someone who does not do research on UAPs. The introduction is clear and gives the required background. The following sections describes the experimental setup and gives examples of data sets with an accompanying discussion. The authors give a clear presentation of the experiments, including all the challenges that occur in experimental work in the laboratory or in the field, such as sensor failures, broken cooling systems, and elks interfering with the experimental setup.
The paper is easy to follow. I am very close to recommending a publish-as-is, but I have just a few minor issues that I would like to see addressed.
Page 5: The discussion of the magnetometer / ADC interface is a bit short, e.g. it is nice to be told the noise floor of the magnetometer, but if it is poorly interfaced to a insufficient ADC, this noise floor is never reached. Please elaborate a bit here. How far is the experimental noise floor from the theoretical noise floor?
Figure 4: The data are sorted by temperature, but it might be beneficial to add a small inset to the figure with the raw data as a time series.
Figure 5: What is MD? I suggest to also add the distance between the magnetometer and the camera as this is not directly clear from the picture.
Figure 10 could be discussed in slightly more detail. Clearly, the instrument is very well insulated as seen by the small overall change in temperature, but the plotted data curve is very “thick”. I suggest adding an inset with a zoom on a few seconds of data to make the data quality and any structure in it more visible. Please also add a comment about the “thickness” of the plot. Is it explained by the electronic properties of the sensor or are external noise sources in play?
Figure 14: I suggest adding insets with zooms on the spikes, so the shape of the spikes can be seen.
Page 21: The discussion of the origins of spikes is a bit short. Are there other potential sources of spikes in magnetometer data, e.g., could cosmic rays be the culprit? Is spikes like this this a common feature for magnetic observatories? On page 6 a 160 m distant road is mentioned. Is there any correlation to traffic on this road or can any other noise be attributed to traffic?
Citation: https://doi.org/10.5194/egusphere-2025-3431-RC2 -
AC2: 'Reply on RC2', Foteini Vervelidou, 03 Oct 2025
We thank the reviewer for the encouraging feedback and the detailed comments that help us improve the manuscript. We provide our responses below each comment, which is given in italics.
- Page 5: The discussion of the magnetometer / ADC interface is a bit short, e.g. it is nice to be told the noise floor of the magnetometer, but if it is poorly interfaced to a insufficient ADC, this noise floor is never reached. Please elaborate a bit here. How far is the experimental noise floor from the theoretical noise floor?
We have added information about the specs of the ADC in the revised manuscript. More precisely, we addressed this comment by doing the following:
Section 2
We now term NI-9239 the Data Acquisition Module (DAQ) and explain that it contains an ADC rather than being just an ADC. We provide the following specifications for the DAQ and its ADC: “The DAQ contains a 24-bit delta-sigma analog-to-digital converter and has an input range of +/- 10 V. The DAQ allows for sampling rates, fs, that range from 1.613 to 50 kHz, has an alias-free bandwidth of 0.453 x fs, and an input-referred noise of 70 μVrms or equivalently 700 pT. Assuming that the input-referred noise corresponds to the highest possible sampling rate, the expected noise in our raw measurements, given that we sample at the lowest possible fs, is approximately 120 pT. This makes the noise floor of the DAQ the main contributor to the expected noise of our setup.”
To allow the reader to follow the conversion from V to T, we now specify the sensitivity of the magnetometer: “The measurement range of the magnetometer is +/- 100 μT, its sensitivity is 1 V per 10 μT,….”
We also provide the noise floor of the PSU: “The PSU has a noise floor of < 5pT/√Hz at 1 Hz.”
Section 6.1
We revised extensively this section. Now we present also our raw data, and we provide an estimate of the experimental noise level, both for the raw and the low-pass filtered data:
“Figure 9 shows the raw measurements (left column) and the 1-sec low-pass filtered data (right column) obtained during May 9th, 2024, a magnetically quiet day (Kp index < 3). In the panels of the right column, the black solid line corresponds to our magnetic field measurements, while the red dashed line corresponds to the magnetic field data recorded at BOU. The baseline of our 1-sec data has been adjusted to that of BOU (hence the difference in the baseline between our raw and 1-sec data). The X, Y, and Z magnetic field components are shown, in the top, middle and bottom row, respectively. The RMS noise of our raw magnetic field measurements is ≈ 20 nT for the X and Y component, and ≈ 50 nT for the Z component. These values, which are typical for our raw measurements on any given day, are two orders of magnitude higher than the theoretical noise floor according to the specifications of the magnetometer and the DAQ, presented in section 2. The potential source of this noise is discussed in the Discussion section. Given the experimental noise of our raw measurements, the expected RMS noise of the 1-sec data is approximately 1-2 nT per magnetic field component, which is indeed what we observe. In addition to that, we observe that the Z component of our raw measurements (Figure 9e) exhibits a ≈ 40s long spike with an amplitude of ≈80 nT. As will be discussed in section 6.4 and in the Discussion section, such spikes occur in our raw measurements usually a few times per week, at irregular intervals.”
The revised Figure 9 is given in the Supplement of our reply.
The potential source of the experimental noise is discussed in the Discussion section, whose first paragraph now starts as follows:
“As noted in section 6.1, the experimental noise in our magnetic field recordings is two orders of magnitude larger than the theoretical noise floor, according to the specs of the magnetometer, the PSU1 and the DAQ. This theoretical noise floor, however, does not account for the 10 meters cable that we use to connect our magnetometer to the PSU1. Given our experience with the external temperature and humidity sensor (see Appendix A), we consider this to be the most plausible explanation for this discrepancy. Nevertheless, this experimental noise, which for our 1-sec data corresponds to 1-2 nT per magnetic field component, is still largely within our performance requirements. These were laid out in the GP Science Traceability Matrix (Watters et al., 2023), which states that the magnetometer should have a "resolution of order ∼nT to resolve diurnal geomagnetic field variations". The diurnal variation, ...”
- Figure 4: The data are sorted by temperature, but it might be beneficial to add a small inset to the figure with the raw data as a time series.
Thanks to this comment we realized two things: we had plotted the data as a function of increasing temperature, while temperature decreased with time, and moreover that we were calling “raw vector data” the “uncalibrated, 1-sec vector data”.
We corrected the term wherever it appears in the text and we added a panel in Figure 4 with the raw uncalibrated data as a function of time. Moreover, we now show the results of the calibration as a function of temperature in its actual order (i.e., descending), so that the two panels show the data in the same order.
The revised Figure 4 is given in the Supplement of our reply.
- Figure 5: What is MD? I suggest to also add the distance between the magnetometer and the camera as this is not directly clear from the picture.
We have included the information about the distance in the figure (9.4 m), and we have included the initials PTZ and MD in the caption. In particular, MD corresponds to the all-sky camera arrays.
The revised Figure 5 is given in the Supplement of our reply.
- Figure 10 could be discussed in slightly more detail. Clearly, the instrument is very well insulated as seen by the small overall change in temperature, but the plotted data curve is very “thick”. I suggest adding an inset with a zoom on a few seconds of data to make the data quality and any structure in it more visible. Please also add a comment about the “thickness” of the plot. Is it explained by the electronic properties of the sensor or are external noise sources in play?
We implemented the suggestion of the reviewer to add to Figure 10 an inset with a 10 sec zoom. However, we considered that this inset still does not allow to discern any structure in the data. The raw data have been obtained at a sampling rate of 1612.9 Hz, and we need to go down to a resolution of a few hundred data points to start discerning any structure in the data. For this reason, we updated Figure 10 to include an inset of a 300 data points zoom-in.
Both Figure 10 with an inset of a 10 sec zoom and the revised Figure 10 with the 300 data points zoom are given in the Supplement of our reply.
Moreover, we added a comment about the thickness of the plot:
“We note that the RMS noise in the raw temperature data shown in Figure 10 is approximately 0.07 oC. Although we do not have information about the noise specifications of the integrated temperature sensor, the observed noise level is consistent with typical industrial temperature sensors.”
And we also clarify at the beginning of the respective paragraph that these are raw data obtained with a sampling rate of 1612.9 Hz:
“Figure 10 shows the raw temperature measurements, obtained with a sampling rate of 1612.9 Hz by the integrated temperature sensor of our magnetometer, during the day of May 9th, 2024. The inset shows a 300 data points zoom-in.”
- Figure 14: I suggest adding insets with zooms on the spikes, so the shape of the spikes can be seen.
We implemented the recommendation. The revised Figure 14 is given in the Supplement of our reply.
- Page 21: The discussion of the origins of spikes is a bit short. Are there other potential sources of spikes in magnetometer data, e.g., could cosmic rays be the culprit? Is spikes like this this a common feature for magnetic observatories? On page 6 a 160 m distant road is mentioned. Is there any correlation to traffic on this road or can any other noise be attributed to traffic?
To address this comment by the reviewer, we have added this paragraph in the Discussion section:
“There are many different sources that can give rise to spikes in magnetometer data, and it is possible that not all of the spikes in our data have a common source. The spikes that have a duration of less than a second are most probably not of natural origin, but longer spikes can also be of artificial origin, like digital errors. For example, we see in Figure 9e that the spike occurs just after a change in the gain. Some spikes could be due to electromagnetic interferences, which themselves have a large variety of causes. The road at 160 m distance from our site could be the source of some interference, especially whenever large vehicles like trucks were passing by. While we are not able to interpret all spikes in our raw data, we expect that some will be eliminated after we install the ferrites recommended by NI for electromagnetic compatibility compliance when using NI-9239 (i.e., our DAQ), as opposed to the generic, cost-effective ferrites we used in this deployment.”
INTERMAGNET observatories typically do not publish data at a higher resolution than 1-sec data, and these data are typically cleaned before being published.
-
AC2: 'Reply on RC2', Foteini Vervelidou, 03 Oct 2025
-
AC3: 'Reply to the Associate Editor', Foteini Vervelidou, 03 Oct 2025
We thank the Associate Editor for considering our manuscript eligible for discussion in EGUsphere, and for forwarding it for peer review. We address his comments, shown in italics, below.
- The authors write that their aim is "to identify magnetic anomalies that cannot be readily explained in terms of a natural or human-made origin". So, the attention should be paid for explanation of the observed phenomena.
We have indeed striven to interpret to the best of our ability and knowledge the data we collected at our variometer station and to describe our findings in the paper. To better clarify our approach to identifying anomalies, we have now added at the end of the Discussion section the following:
“The identification of potential anomalies in our magnetic field data will rely on the collection of magnetic field measurements over a sufficiently long time for us to gain an understanding of the typical magnetic field variations at each site, due to natural phenomena and human activity. Moreover, our magnetic field recordings will be compared with the magnetic field data of the closest available magnetic field observatories, and will be analyzed in tandem with the data collected by the multi-sensor, multi-modal instrumentation suite at the respective GP site.”
-
Please give the magnetic variations (observations) not in microTesla, but in nanoTesla.
We implemented the suggestion and we provide an example in the Supplement. We find, however, the result less easy to read than if we keep the units to be microTesla, so we prefer to keep our plots in microTesla. If, however, this poses a problem, we will revise our manuscript accordingly.
Data sets
Magnetic field and temperature data obtained at the geomagnetic variometer station of the Galileo Project in Colorado, USA The Galileo Project https://doi.org/10.5281/zenodo.15825118
Model code and software
Python script to record magnetic field and temperature data Laura Domine, Abigail White https://doi.org/10.5281/zenodo.15824706
Viewed
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,230 | 0 | 2 | 1,232 | 0 | 0 |
- HTML: 1,230
- PDF: 0
- XML: 2
- Total: 1,232
- BibTeX: 0
- EndNote: 0
Viewed (geographical distribution)
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This is a very well written manuscript. Although I will be surprised if the magnetometer system is ever used to provide insightful data regarding UFOs, the data themselves will be valuable for a wide variety of applications. I have a few questions, which I hope the authors will address.
1. I note that the air-conditioning system at the BOU observatory was not working during calibration. Since the authors are set up near that observatory, are there plans to revisit the BOU observatory and investigate system response over a wider range of temperatures?
2. Are there plans to make the data available to the scientific community in the style of (say) Intermagnet or SuperMag? This is important. Other investigators will make use of the data, and, in the end, the project represented in this manuscript will find indirect support from the wider scientific community.