Weather Satellites

Updated 29/5/01 And I might update it again soon


Introduction

This page is an introduction to the reception of NOAA weather satellite images, in real time, directly from the satellites.
For many years, there have been several NOAA weather satellites in orbit. Currently, the four operational ones are NOAA 12, 14, 15 and 16. The satellites are in 100 minute polar orbits, and so can be received from anywhere on Earth at some time each day. Although the satellites contain many sophisticated instruments, most of the data transmitted is quite hard to receive. However, these satellites transmit continuous images of the the ground directly beneath them, and modulate it onto a 137 MHz carrier using FM, and an audio tone. This band is tuneable on even the cheapest of radio scanners, and is of such a wavelength that home made antennas are both easy to construct, and compact. This page describes the steps needed to receive and decode these images, concentrating on the antenna,

Stuff needed

  1. Radio receiver - must be able to tune to 137.62MHz (NOAA 14, 16) and 137.5MHz (NOAA 12,15) and demodulate FM signals. A wide band FM receiver is preferred, but narrow band receivers work. The radio I have cost me GBP5 (second hand!), and is the cheapest of the cheep. There is nothing 'pro' about the Realistic pro50 scanner, but it works.
  2. External Antenna - the built in rubber 'dummy load' antennas that come with scanners are useless for this (and indeed most) applications, and need replacing with a carefully constructed external antenna
  3. Connecting co-ax.
  4. PC with soundcard, and perhaps recording software
  5. Software for predicting passes of the satellites e.g. WinOrbit
  6. Software for decoding the audio data e.g. WxSat
There are two methods of connecting the computer, radio receiver, and antenna. The first uses a short wire between the soundcard line in jack, and the receiver. A long piece of 50 ohm RG58 co-ax connects the receiver to the external antenna. The second involves connecting the antenna to the receiver via a short piece of co-ax, and having the radio outside with the antenna, then using a long piece of co-ax to connect between the sound card and radio. The second method is preferred, since co-ax is lossy, and long lengths introduce noise into the system, increasing the noise figure. If this occurs between the antenna and receiver, then the signal to noise ratio is worsened and so reducing the overall performance (The antenna signals are weak, so the noise can swamp the signal). However, the signal at the headphone socket of the receiver could be 10dBm, so the noise has an insignificant effect on the demodulated audio signal. Of course this means the radio needs to be outside, fine for a cheep battery powered scanner and portable antenna, but not so good for a GBP2000 pound receiver bolted to the roof!

All the software I used was shareware and obtained from the www.

Antenna construction

You need
  1. 4, 5mm diameter 51cm stainless pipes (each one is slightly shorter than a 1/4 wavelength)
  2. Suitable length of 75 ohm phasing co-ax
  3. Short length of 50 ohm co-ax to connect antenna to radio (fit BNC / PL259 etc.)
  4. Wooden cross, to attach the antenna rods to (see diagram)
  5. Fixing hardware (cable ties)
  6. Wood / Plastic mast
The antenna is a turnstile antenna, made from two half wave dipoles at right angles, connected together with a pi/2 phasing piece of co-ax. It has a nearly isotropic radiation pattern.

My antenna is made from 4 car aerials which I purchased from the local scrap yard for an extortionate price. If you do this, be sure that each aerial is not only long enough (which it ought to be), but also of the same diameter as the other 3. The collapsible nature of these aerials enables the antenna to be folded away for stowage, as well as being able to vary the length of the rods, either to tune it to another band - e.g. 2m or to fine tune the performance.

Phasing co-ax length = wavelength / 4
= velocity / (137.5M * 4) = 1 / (L0 * C0 * 550M) (in m)
Where L0 and C0 are inductance and capacitance per unit length of the co-ax, related by 75 = sqrt (L0 / C0). (Get these numbers from the data sheet).
Alternatively, If you know the velocity factor (around 60-70%) then the length is given by
Length=[300*Velocity_Factor/137.5]/4 = 0.545 * Velocity_Factor (in m)
The length is about 35cm - check for you co-ax though! Add a couple of cm on to this length to allow the ends to be stripped
Bonus if you have a network analyser - put a connector on one end of a length of coax (perhaps 50cm), and carefully calibrate the analyser for a one port measurement (50R calibration is ok in this case). Connect the coax to the analyser and view the phase of the port impedance. Chop small bits of the coax off and observe the points where the phase goes from -pi/2 to +pi/2. These points occur at odd quarter wavelengths, and move up in frequency as the line gets shorter. Keep cutting until the first of these points is just below 137MHz, (ie the coax is a bit too long).
This works because an open circuit TL has impedance Z=-j.Z0.cot(2*pi*l) where l is the length of the coax measured in wavelengths. This shows that an open circuit TL always has an imaginary impedance, and that the phase changes abruptly every quarter wavelength. (This assumes that the loss in the coax is very small. A fair assumption for a short length of coax working at low frequencies.)

The wooden cross is constructed such that the whole thickness is 25mm, i.e. each wood block has a 25*25*12.5 slot in the centre, where they meet. This makes for a stronger assembly, ensures that the arms are at 90 degrees (use a T square!), and makes a neater finish. The grooves are filed down the topside of one set of arms, and the bottom side of the other two arms. This allows the two halfwave dipoles to be kept apart slightly, which eases construction. The fact that the two dipoles are offset by 20mm or so in the vertical direction does not affect performance.

Affix the 4 rods to the wooden cross using cable ties, doing them up loosely. Strip and tin the phasing co-ax by about one cm on each end. Strip and tin the 50 ohm co -ax by about 1 cm at the antenna end (fit a suitable plug to the other end). Now solder the 50 ohm co-ax to the 75 ohm co-ax, matching up the centre and braid in each case. you should now have a single long cable, with a join 50 cm from the free end. Insert the braid part of the free end under one of the topside antenna rods, and pull its cable ties very tight. Likewise fix the centre conductor under the other topside antenna rod. Turn over the cross with the rods attached, and fix on the two remaining connections in the same manner. These two are quite fiddly as the length of tinned conductor is short, and the co-ax is stiff. When all 4 rods are connected, push each one in towards the centre, such that 180 degree opposite rods are as close together as possible without touching. Ensure that the screws holding the cross together are not touching the rods, paint the wooden cross if there is any chance of this happening.
Finally, fix another long length of 25*25 hardwood to the centre cross between two of the side arms. This arm is just to fix the antenna to a mast.

The photos may help to show what to do!

Note, the polarity of the wires on the antenna rods does matter - One way round gives right-hand circular polarisation, the other left-hand. Unfortunately, I'm not sure which is which! The antenna didn't appear to perform differently either way round.

A network analyser would be rather handy at this stage, but I don't have one at home, and I don't have the antenna here in London to take into university with me. If you have one, get tuning!

Dipole Antenna

I have also obtained as good, if not better results using a 1/2 wave dipole antenna. I mounted it horizontally, with the arms perpendicular to the motion of the satellite (ie aligned NE/SW at my latitude). I also made a small ballun using a ferrite core from a TV modulator and skinny copper wire. This isolates the unbalanced co-ax cable from the balanced antenna, and improves the radiaition pattern. Of course, the transfomer introduces some insertion loss, and again I haven't yet measured this on a network analyser. One day.

Photos

Wooden cross arrangement
Pictures of my antenna - note the wooden fixing post fits inside the yellow plastic pole, which can be inserted into the tripod. Also shown in the right hand picture is my attempt at building a mast head pre-amplifier to allow long lengths of co-ax to be used between the antenna and receiver. Unfortunately, it is a 1mW 1.5 GHz oscillator at the moment. Time to go back to the spectrum analyser, and soldering iron :-(

Sample Pictures

Received in Harrogate, UK during the summer of 1998, and decoded using WxSat.


Image 1 - NOAA 14. Note the visibility of Lake Geneva.

Image 2

Image 3

Image 4- Both channels shown, both visible light (left) and IR (right). This imaged received late at night, notice the lack of contrast compared with Image 1, and the highlighting of the edges of the clouds.