Solar Eclipse 2017

LF Radio Propagation Experiment

Conducted by John Magliacane, KD2BD, on August 21, 2017

Introduction

The solar eclipse that swept across the continental United States on August 21, 2017 provided a unique opportunity to study the effects the eclipse might have on radio signals propagating via the Earth's ionosphere.

The lowest layer of the ionosphere is the 'D' layer. The 'D' layer is most rapidly affected by changes in solar radiation, and it is the layer responsible for long-distance radio communications in the low-frequency (LF) radio spectrum.

The National Institute of Standards and Technology (NIST) operates radio station WWVB on a frequency of 60 kHz to within one part in 10¹⁴ from a transmitter located in Wellington, Colorado, just north of Fort Collins. Since the path of totality was predicted to cross the Great Circle path between WWVB and my home in East Central New Jersey, hardware was developed over the summer months to try to measure and record amplitude and phase perturbations that might occur to reception of WWVB over the course of the eclipse.

Hardware Development

I employ a WWVB-disciplined frequency standard of my own design as a reference for conducting precision frequency measurements. The ability to measure perturbations in the received phase of WWVB's carrier would require the use of an independent frequency standard having even greater precision and stability that would be immune to the effects of the eclipse. In order to keep the uncertainty of the WWVB carrier phase measurements to one degree or less over an eight hour period, a frequency standard having an accuracy and stability to within 1.6 parts in 10¹² would be required. A clock having this level of accuracy would be in error by less than one second in 50,700 years.

Realizing that some commercial AM radio broadcasters in the United States employ GPS-disciplined Rubidium frequency standards to control the frequency of their transmitters, a survey of every local AM radio station within ground wave range was conducted using my WWVB-based frequency measurement hardware. Through this process, radio station WFAN in New York, NY was found to be broadcasting on exactly 660 kHz with no discernible signal fading after sunset.

A carrier phase tracking receiver was developed to generate a very precise frequency reference from WFAN's 660 kHz carrier. This reference was used to phase lock the local oscillator of a quadrature phasing (image rejecting) direct conversion receiver of my own design to exactly 236 kHz, thereby allowing reception of WWVB as a 1 kHz audio tone having all the amplitude and phase characteristics of WWVB, itself.

The WFAN-derived Master Frequency Reference was also used to synthesize a 60 kHz carrier. The 60 kHz carrier was downconverted to a 1 kHz audio tone using exactly the same local oscillator and mixing processes as those operating in the WWVB receiver.

Since each mixer was driven by the same 236 kHz local oscillator, any phase instability present in the local oscillator would simply cause the phase of the 1 kHz reference tone and that of the audio from the WWVB receiver to track each other by exactly equal amounts. As such, any phase deviations observed between these two audio sources over time would be the result of changes in WWVB signal propagation at 60 kHz alone, rather than any vagaries of the hardware used to make the measurements.

Audio from the WWVB receiver and the 1 kHz audio reference tone were recorded concurrently for later processing and analysis using a homebuilt Slackware v14.2 Linux-based PC running digital audio recording software of my own design. Audio recorded during the experiment is available through the HamSCI Community at Zenodo.

Overview of the Eclipse Propagation Experiment Methodology employed by KD2BD

Some Results

Using the hardware described, peak carrier amplitude and carrier phase measurements of WWVB were conducted between 14:00 UTC and 22:00 UTC on the day of the eclipse. This eight hour period provided a substantial amount of baseline data on both sides of the eclipse. Additional baseline data was gathered for several days prior to the eclipse as well.

Anomalous propagation effects were observed beginning at 17:00:00 UTC when the phase of WWVB's carrier began to reverse its earlier upward trend. The rising trend was probably the result of steadily rising 'D' layer ionization levels throughout the day. Observations taken on days prior to the eclipse showed the same effect, with a trend in the opposite direction occurring after sunset (see The Effects of Sunset below).

WWVB Relative Carrier Phase as a Function of Time on the Day of the Eclipse
At 17:57:20 UTC, the Moon's shadow crossed the 1622 mile Great Circle path between WWVB and KD2BD along the east coast of Central New Jersey. At that time, the center of the Moon's shadow was located over south-central Nebraska at approximately 41.2 degrees North and 99.55 degrees West. While unrelated to the eclipse, the Sun produced an X-ray burst at this time that affected the regions of the ionosphere unprotected by the eclipse, and produced a "spike" in the phase plot.

The Path of Totality Crossed the RF Path Between WWVB and KD2BD at 17:57:20 UTC

Maximum carrier phase shift and maximum signal strength rise were both observed at 18:13:00 UTC when the Moon's shadow was located over North Central Missouri at approximately 38.966 degrees North and 92.666 degrees West.

It is interesting to note is that an abrupt change in carrier phase occurred well before the path of totality crossed the Great Circle path between WWVB and New Jersey. However, WWVB's carrier amplitude remained remarkably constant until the path was nearly fully eclipsed.

WWVB Relative Carrier Amplitude as a Function of Time on the Day of the Eclipse

Some Projections and Hypotheses

The phase of WWVB advanced at a rate of approximately 19 degrees per hour prior to the onset of the eclipse. This corresponds to a continuous decrease in RF path length of approximately 263.7 meters per hour, or a total of 799.4 meters between the start of data collection at 14:00 UTC, and the time the rising phase began to reverse its upward trend shortly after 17:00 UTC when a maximum phase shift of +57.6 degrees was measured.
The phase plot shows effects of the eclipse continued until approximately 19:30 UTC, for a duration of 2 hours and 30 minutes.
After the eclipse, the phase resumed its rising trend, but this time at a slightly lower rate of approximately 16.5 degrees per hour.
Ignoring the period of the eclipse, the average rate of carrier phase advancement was approximately 17.75 degrees per hour over the eight hour period studied.
If this average rate is applied throughout the eclipse period, we can project an estimated carrier phase of +80 degrees at 18:13 UTC had the effects of the eclipse not been present.
The measured carrier phase at this time was -36.0 degrees, which is 116 degrees behind the projected value. At a carrier frequency of 60 kHz, this phase retardation corresponds to a free-space RF path length increase of 1.609 km, or very close to 1 statute mile. Therefore, it appears as if the Moon's shadow produced an upward projecting concavity within the 'D' layer of one half this amount.
The phase of WWVB was clearly affected to a much greater extent than its amplitude during the eclipse. One might expect similar effects to both aspects of the signal had WWVB been received through similar amounts of surface wave and ionospheric wave propagation. Since this was not the case, the RF path between WWVB and KD2BD was likely one where the skywave path was significantly dominant. This premise is consistent with previously published NBS literature that suggests propagation at 60 kHz across a 1622 mile distance to be the result of a two-hop path.
As such, the 4 dB signal strength increase observed during the eclipse may have been due to increased 'D' layer efficiency, rather than the vector addition of skywave and surface wave paths moving toward a closer phase alignment at the receiving location during the time of the eclipse.
The narrow signal peak within the center of the broader peak might be a multipath and/or focusing effect due to 'D' layer scattering. It does not appear to be due to solar weather events because there were none occurring at that time. The "spike" on the leading edge, however, may have been due to a solar X-ray burst occurring at that time.

The Effects of Solar Weather Events

Fortunately, the Sun was relatively quiet during the eclipse, although there were several minor events that slightly affected the results illustrated here. In particular, Solar Event #3370 occurred just as the RF path was entering into full eclipse, and is believed to be responsible for "spike" in the phase plot at 17:57 UTC:

 
:Product: 20170821events.txt
:Created: 2017 Aug 24 0357 UT
:Date: 2017 08 21
# Prepared by the U.S. Dept. of Commerce, NOAA, Space Weather Prediction Center
# Please send comments and suggestions to [email protected] 
#
# Missing data: ////
# Updated every 5 minutes.
#                            Edited Events for 2017 Aug 21
#
#Event    Begin    Max       End  Obs  Q  Type  Loc/Frq   Particulars       Reg#
#-------------------------------------------------------------------------------

3370 +     1739   1757      1801  G13  5   XRA  1-8A      C3.0    1.8E-03       
3370 +     1754   1754      1754  SAG  G   RBR  410       110                   

3380       1958   1958      1958  PAL  G   RBR  245       100                   

3390 +     2012   2022      2026  G13  5   XRA  1-8A      C1.5    6.2E-04   2671
3390       2019   2021      2029  HOL  3   FLA  N09W23    SF      DSD       2671
3390 +     2019   2021      2021  PAL  G   RBR  1415      45                2671
3390 +     2019   2021      2021  PAL  G   RBR  4995      30                2671
3390 +     2021   2021      2021  PAL  G   RBR  410       30                2671
3390 +     2021   2021      2021  PAL  G   RBR  8800      20                2671
3390 +     2021   2021      2021  PAL  G   RBR  610       130               2671
3390 +     2021   2021      2021  PAL  G   RBR  2695      12                2671

Some of the Larger Solar Events that Occurred Within the Data Gathering Period on the day of the Eclipse

It is possible that Solar Event #3390 may have caused the small "spike" in signal amplitude recorded around 20:20 UTC, however, no observable change in carrier phase was detected at that time.

GOES X-Ray Flux Levels from August 19 through August 21, 2017

Solar Flares and X-Rays

More significant X-ray events occurred on the days leading up to the eclipse, and many of these produced measurable increases in WWVB signal levels. However, unlike the signal enhancement caused by the eclipse, the 3.3 dB signal level increase caused by solar event #3190 on August 20, 2017 (for example) had a significantly shorter rise time compared to its fall time:

WWVB Relative Carrier Amplitude Recorded on August 20, 2017

While the X-ray burst affected a much larger area of the ionosphere, it had a relatively minor effect on received carrier phase compared to the solar eclipse that affected a much smaller region of the ionosphere:

WWVB Relative Carrier Phase Recorded on August 20, 2017

 
:Product: 20170820events.txt
:Created: 2017 Aug 23 0357 UT
:Date: 2017 08 20
# Prepared by the U.S. Dept. of Commerce, NOAA, Space Weather Prediction Center
# Please send comments and suggestions to [email protected] 
#
# Missing data: ////
# Updated every 5 minutes.
#                            Edited Events for 2017 Aug 20
#
#Event    Begin    Max       End  Obs  Q  Type  Loc/Frq   Particulars       Reg#
#-------------------------------------------------------------------------------

3190 +     1920   1939      1949  G13  5   XRA  1-8A      C9.4    1.1E-02   2672
3190       1935   1935      1939  SAG  G   RBR  610       21                    
3190       1935   1938      1940  SAG  G   RBR  4995      36                    
3190       1935   1938      1940  SAG  G   RBR  8800      110                   
3190       1935   1938      1940  SAG  G   RBR  15400     95                    
3190       1935   1939      1939  SAG  G   RBR  2695      26                    
3190       1937   1938      1939  SAG  G   RBR  1415      35                    
3190       1938   1938      1939  SAG  G   RBR  410       76                    
3190       1938   1938      1938  SAG  G   RBR  245       30

Some details of Solar Event #3190 that produced a large increase in WWVB signal level at 19:40:00 UTC on August 20, 2017.

A full understanding and appreciation of the effects the solar eclipse had on the Earth's 'D' layer requires a consideration and elimination of any solar weather events that may have been occurring at the time. While the ionosphere directly below the eclipse may have been shielded very briefly from such events, the same cannot be said for the remaining sunlit portions of the ionosphere that were also responsible for signal propagation.

The Effects of Sunset

As a further "sanity check", the phase and amplitude of WWVB were examined during sunset periods in early October to further test the hardware used during the eclipse and to further validate the results obtained during the experiment.

WWVB Relative Carrier Phase Recorded During Sunset on October 1, 2017

In this plot, the carrier phase can be seen continuing its normal daytime rise until about an hour and a half after local sunset (22:31:09 UTC as determined through PREDICT Software) when the rise began to reverse its earlier trend.

The phase plunged over 145 degrees between 23:30:00 UTC on October 1, 2017 and 00:10:00 UTC on October 2, 2017, implicating an RF path length gain of over 2 kilometers during that 40 minute period. Overall, the sunset period contributed to an increase in RF path length of nearly 3 kilometers, which implies that that WWVB signal propagation remained within the 'D' layer of the ionosphere, and did not rise to any higher regions after sunset.

WWVB Relative Carrier Amplitude Recorded During Sunset on October 1, 2017

The carrier amplitude increased by a little over 10 dB during the sunset period, but not before undergoing several undulations in signal strength before the radio path was completely immersed in darkness. While in darkness, short-term signal strength variabilities increased noticeably, and the phase plot shows some short-term variabilities during this time period as well. Since WWVB was significantly stronger at this time, noise is not a likely cause of this variability.

References and Further Information

This page was last modified on January 22, 2021.