Weatherfax Signal On 80m

Did you know there is a HF weatherfax transmission which can be heard in the 80m band designated for amateur radio use in Australia?  And, it's not a pirate station.  This page describes my experiences with receiving images from this station.

Back to Miscellaneous Technical Page

Investigating the weatherfax signal on 80m

In Australia, the frequency range 3500 - 3700 kHz is allocated for amateur radio use, and is known as the "80m band".  The term "80m" is a historical reference to the nominal wavelength * of the frequencies within that "band", or frequency range.  (The actual wavelengths in free space for the 80m band range from 85.7 m at 3.5 MHz to 81.1 m at 3.7 MHz, so the term "80m" is very nominal indeed!)

* Using wavelength as a band designator dates to the earlier days of radio, when wavelength was more commonly used than frequency as a means of locating a signal in the radiofrequency spectrum.
  (Such archaic practices can lead outside observers to judge amateur radio operators as a reactionary and conservative lot ...)

In preparation for the 160m Trans-Tasman CW and Digital Contest held in July 2011, I decided to revisit some digital modes (eg PSK31, RTTY, etc) on HF ... after last using PSK31 some 10 years ago.  This revisitation involved building an interface unit which would provide the transmit audio, receive audio, and PTT line connections between the Personal Computer (PC) soundcard and the FT-817 transceiver.  (More will be said about this interface unit on a separate webpage sometime in the future.)  I chose MultiPSK as the software for generating and decoding digital modes.  One of the benefits of this software is that it provides a large number of modes.  (The other benefit is that it's freeware if one doesn't want the professional modes such as SITOR, etc.)

In the week after the contest, I was prowling around the 80m band one evening looking for interesting digital modes to try out.  According to the WIA band plan, the frequency range 3620 - 3640 kHz is designated as the digital modes sub-band on 80m.  (Unfortunately, it is not unusual to hear Australian amateur radio operators using SSB in that sub-band, even though much of the rest of the 80m band is available for their use.)  I found a signal centred on 3622.5 kHz which was quite audible - the FT-817's display indicated S8 (for what that's worth as a measurement result) - but also quite noisy.  The MultiPSK spectral display exhibited two strong spectral traces 800 Hz apart, as shown below - this suggested a frequency-shift keyed (FSK) or frequency modulated (FM) signal.  Stepping through the various FSK and FM digital modes provided in MultiPSK further suggested that this was a facsimile (or "fax") signal.

Fax spectrum
Screenshot of MultiPSK in BPSK31 mode for illustrating the spectral components of the fax signal.  The FT-817 transceiver is tuned to 3620.6 kHz in DIG mode with Menu #26 DIG MODE set to PSK31-U - this effectively implements a USB receiver tuned to a suppressed carrier frequency of 3620.6 kHz .  The strong spectral component at approximately 2300 Hz (indicated by the bright red trace on the right-hand side of the display) is the white sub-carrier; while the somewhat smeared spectral component at approximately 1500 Hz (indicated by the speckled red and yellow trace in the middle of the display) is the black sub-carrier.  Note that 1500 Hz and 2300 Hz are standard sub-carrier frequencies for fax; hence the tuning of the FT-817 to obtain these frequencies in the audio passband for the PC soundcard.  The mean frequency of these tones is 1900 Hz, which corresponds to a transmitted frequency of 3622.5 kHz - when referring to the transmission frequency as such, the black and white sub-carriers are expressed as -400 Hz and +400 Hz respectively.

Here is a brief sound recording of the signal.

I had no prior experience with receiving fax signals via HF radio, let alone with the MultiPSK software; so clearly some experimentation would be required in order to receive some images.  As well as frequency shift, two other important parameters for fax transmissions are the Lines Per Minute (LPM) and Index Of Cooperation (IOC).  The frequency shift was obvious: 800 Hz per the spectral traces.  MultiPSK offers LPM selections of 60, 90, and 120; and IOC selections of 288 and 576.  Experimenting with these eventually yielded the combination of 120 LPM with an IOC of 576 which produced an image that looked like part of a map - but a map of what I wasn't sure.  Remember; at this time I had no idea of what this transmission was, where it was coming from, who was sending it, or for what purpose; and the signal was quite noisy, thus rendering the received image quite distorted, smeared, and "noisy".


During the following day, some enquiries to some knowledgeable people suggested that this might be a weatherfax station.  But, a weatherfax station operating in the 80m amateur radio band?  A strong but noisy signal suggested the possibility that it was originating from some considerable distance; for example, a country which did not have that amateur radio band allocation; but, there was also the possibility that it was a not-so-strong signal emanating from not-so-far away.  Then someone advised me of a reference they found to a Japanese weatherfax station operating on 3620.6 kHz.  The pieces of this puzzle began to fall into place.

NOAA's National Weather Service Marine Forecasts webpage contains a PDF document titled "WORLDWIDE MARINE RADIOFACSIMILE BROADCAST SCHEDULES" - this lists the details of radiofacsimile station JMH, which is operated by the the Japan Meteorological Agency (JMA).  Details of the JMH transmission schedule are also held on JMA's server here.  The station transmits on frequencies of 3622.5, 7795.0, and 13988.5 kHz; using emission mode F3C ** featuring white sub-carrier at +400 Hz and black sub-carrier at -400 Hz; and at 5 kW power.  (It is not known if this is transmitter output power, ERP, etc.)

** According to ACMA's publication "Emission characteristics of radio transmissions", the emission designator F3C can be understood as shown in the table below.

Symbol
What the symbol defines
Value
Meaning
1st
Type of modulation of the main carrier F
Emission in which the main carrier is angle modulated by Frequency modulation
2nd
Nature of signal(s) modulating the main carrier 3
A single channel containing analogue information
3rd
Type of information to be transmitted C
Facsimile


The only unanswered question remaining was why is a Japanese weatherfax station operating in the Australian 80m amateur radio band?  Reference to the
Japan Amateur Radio League, Inc. (JARL) website for the Japanese amateur radio bandplan provided the answer.  Japanese radio amateurs have access to six (6) segments of radio frequency between 3500 kHz and 3805 kHz, with gaps between each segment.  One such gap covers the frequency range of 3612 - 3680 kHz; hence radiofacsimile station JMH is not operating in the amateur radio 80m band in Japan.

Back to top

Images received from JMH on 3620.6 kHz

Propagation conditions on the evening of 14 July 2011 were improved over those experienced earlier in the week when I first heard this signal.  The images shown below are two of many received that evening.  The general illegibility of these charts is a reflection of the relatively low signal to noise ratio at the receiving location.

Fax @ 1118Z, 14 JUL 2011
Weatherfax image received at 1118 UTC on 14 July 2011.  This image shows a synoptic chart for the northwest Pacific region.  The coastlines in the region can be seen as light grey lines.  Latitude curves and longitude lines are shown in dark grey / black.  Isobars (lines linking points of equal atmospheric pressure) are depicted in black, with atmospheric pressure values indicated against some of them.  The concentration of closed isobars in the bottom right quarter of the chart represents Typhoon TY 1106 (Ma-on).  At the top left-hand and bottom right-hand corners can be seen a box containing the radiofacsimile station's callsign, JMH.  For more information on understanding synoptic charts; visit the Australian Bureau of Meteorology's website, or read the NOAA National Weather Service "Ocean Prediction Center's Radiofacsimile Charts User's Guide".

Fax @ 1219Z, 14 JUL 2011
Weatherfax image received at 1219 UTC on 14 July 2011.  This image shows the predicted track for Typhoon TY 1106 (Ma-on) in the northwest Pacific region.  In contrast to the previous image; all features such as coastlines,
latitude curves, longitude lines, and the typhoon track, are shown in black.

It would be reasonable to say that these images were received at a distance outside the intended coverage area served by radiofacsimile station JMH on the transmission frequency of 3622.5 kHz.  One question could be, "What is the coverage area for these transmissions?"  I have not been able to find any information on the JMA website or other sources regarding the intended coverage area of JMH.  However, some insight into the answer to this question can be gained by examining the propagation predictions for July 2011.

In order to do this, we need to establish what received signal-to-noise ratio (SNR) is required for "good" quality radiofacsimile reception.  In the early 1990s, a group of engineers in the USA tested the feasibility of transmitting facsimile over HF radio voice-channels using existing fax machines and radio installations. Among other facsimile systems, they tested a Group 1 fax - which uses FSK, similar to the signal transmitted by JMH - with a HF radio channel simulator. They found that Group 1 fax produces documents on most simulated HF channels, although the quality of the document degrades as the impairments become severe. Their work is reported in the following paper: Facsimile Transmission Via HF Radio; Bodson, Cramer, Deutermann, & Robinson; IEEE Transactions On Vehicular Technology, Vol. 40, No. 3, August 1991; pp 515-520.  Bodson et al found that 16 dB SNR provided a "good" quality of received document.

The ASAPS software produced by IPS can be used to determine the likely maximum usable ranges for a radiofacsimile signal transmitted at 5 kW, and received with 16 dB SNR minimum.  Due to the lack of information concerning the JMH transmitter and antenna system, it is assumed that the 5 kW figure applies to the antenna input, and that the antenna is an isotropic radiator.  The receive antenna is also assumed to be isotropic (which matches the experience with the 80m inverted vee antenna at VK2IG), and the received noise level is that experienced at a rural site.  (Note that such a received noise level could also be expected for a location at sea.)   The predicted coverage areas are shown below in the two figures for 1100 and 1200 UTC in mid-July 2011.

Coverage @ 1100 UTC
Predicted coverage area for "good" quality image reception at 1100 UTC in mid-July 2011.  The locations of the transmitter and receiver are marked with red crosses.

Coverage @ 1200 UTC
Predicted coverage area for "good" quality image reception at 1200 UTC in mid-July 2011.  The locations of the transmitter and receiver are marked with red crosses.

The figure below shows the propagation conditions between the JMH transmitter site and the Gundaroo receiving site, as predicted by ASAPS.

Propagation conditions, July 2011
Predicted propagation conditions between the JMH transmitter site and the Gundaroo receiving site.

Back to top

Fax signal "bits and pieces"

Like much of my experience in amateur radio; I learnt a lot while chasing down the source and purpose of this fax signal.  For example, how does the receiver know what is the start of a transmitted image signal, and when has the transmitted image signal reached the end of the page?  What do you do to synchronise an image if you miss the start signal?  And; how do you correct the image skew - where the image doesn't sit "square" in the image pane - as seen in the weatherfax images shown above?  This section provides some answers to those questions.

Image start

For a system using an IOC of 576; the start of the transmitted image signal is indicated by a 5 s duration 300 Hz start tone, as shown in the spectral display in the figure below.

Fax start signal

Partial screenshot of MultiPSK in HF FAX mode for illustrating the fax start signal.  The "waterfall" spectral display scrolls from top to bottom, thus the components at the bottom occurred earlier in time than those above it.  The fax start signal spectral components can be seen in the lower quarter of the waterfall.  It consists of components spaced at 300 Hz intervals; falling at baseband (audio) frequencies of 400, 700, 1000, 1300, 1600, 1900, 2200, and 2500 Hz.  Above that can be seen the black sub-carrier at 1500 Hz and white sub-carrier at 2300 Hz.  In this case, only a few lines have been transmitted - these can be seen as a narrow grey/black region at the very top of the black image pane (located in the lower portion of the figure).  The energy in the black sub-carrier is much higher than that of the white sub-carrier, as indicated by the former being more red in colour than the latter, and also by the grey/black colour of those first few lines of the received image.

Image stop

For a system using an IOC of 576; the end of the transmitted image is indicated by a 5 s duration 450 Hz stop tone, as shown in the spectral display in the figure below.

Fax stop signal

Partial screenshot of MultiPSK in HF FAX mode for illustrating the fax stop signal.
 
The "waterfall" spectral display scrolls from top to bottom, thus the components at the bottom occurred earlier in time than those above it.  The fax stop signal spectral components can be seen in the centre section of the waterfall.  It consists of components spaced at 450 Hz intervals; falling at baseband (audio) frequencies of 550, 1000, 1450, 1900, and 2350 Hz.  Above that can be seen the black sub-carrier at 1500 Hz - the transmitter is idling on this frequency, hence the white sub-carrier (at 2300 Hz) is not present.  Note the speckled grey and white band at the bottom of the received image indicating the end of the image.

Tracking

"Tracking" is the term I use for the means of ensuring that the start and end of each received line of a fax image is displayed at the edges of the image pane, and not somewhere in the middle of it; and for ensuring that the image is not skewed on the page.  The weatherfax images shown above exhibit good tracking from the viewpoint of locating the start and end of each line, but poor tracking from the viewpoint of skewness.  On the other hand, the image used to illustrate the stop signal characteristics exhibits acceptable tracking from both viewpoints.  Any decent fax receiver - whether it is implemented in hardware, software, or a combination thereof - must incorporate a means of adjusting tracking.  The picture below illustrates the effects of these adjustments.

Fax tracking

Partial screenshot of MultiPSK in HF FAX mode for illustrating tracking adjustments.  The start signal was missed when receiving this image.  (This can occur when, for example, you tune into a fax signal after the start of the image transmission, but you still want to see what is being sent.)  The black band at the top right of the image pane, starting about one eighth of a page width from the right-hand edge of the page, indicates the start and stop of each line of the received image.  Each image line starts immediately to the right of the black band, and then "wraps around" to the left-hand edge of the page and continues to the left of the black band.  Clearly, this is an undesirable artefact from the viewpoint of producing a contiguous image.  The "Shift" adjustments (located just below the 700 - 800 Hz region of the waterfall spectral display) are used to correct the position of this band.  Progressively pressing the ">" button as each line is received moves the black band across the image pane until it reaches the desired location, where it bridges the right-hand and left-hand sides of the page, thus centring the remainder of the received image on the page.

The "Slant" adjustments (located on the left-hand side below the waterfall spectral display) correct the skewing of the image on the page.  In the case shown, a slant of +6.1 is required to reduce the amount of skew to an acceptable level.  An indication of the effect of this setting can be seen by comparing the weatherfax images obtained on 14 July 2011 - which were captured before this slant was introduced - and the image used for illustrating the stop signal characteristics - which was captured after this slant was introduced.

Back to top

This page was created by Mike Dower VK2IG: 16 Jul 2011.  Material may be copied for personal or non-profit use only.