Discussion of the Manastash Ridge Radar Project

Manastash Ridge Radar Documentation:
Hardware Manual | Software Manual

Tutorials:
Basic Radar Information
Ionospheric Physics
S-Parameters

by Melissa Meyer

We are working on a bistatic passive radar, with the primary purpose of studying Earth's ionosphere (although the radar is applicable in many other areas that radars tend to be useful for as well).

Our system is a passive radar, which means that we don't run our own transmitter; we "listen" to commercial FM radio broadcasts with two (or more) antennas. There are many reasons for running a passive radar, especially in our specific case. Our antennas are in separate locations, about 150 km apart - thus the term bistatic. One is in Seattle (in our lab in the EE building, to be exact - it's a wire hanging out our window), and the other is to our southeast, behind the Cascades mountain range (most notably, Mt. Rainier). The remote antenna (there are actually two of them at the moment; this will become important for the discussion of interferometry) is located at the Manastash Ridge Observatory (MRO); thus, the name of our radar is the Manastash Ridge Radar (MRR).

Here's how it works. When a target scatters radio signal back down to us, it is modulating the illuminating signal: the original signal's amplitude ("loudness") and frequency are changed according to the properties of the target (specifically, the target's radar cross section (a measure of a target's ability to reflect radar signal) affects the amplitude, and its speed affects the frequency (called "Doppler shift")). The receiver at UW picks up our reference signal (from the FM radio station). The measurement of the reference signal needs to be as clean and accurate as possible. One of my lab mates is working on methods of removing noise from the reference signal we receive.

Then, after it has been reflected/backscattered off of various targets, the signal is picked up in the remote receiver (at MRO). Ideally, we would receive only that signal which had been scattered from interesting targets... however, a significant amount of the power we receive in the remote antenna comes from scattering of the signal off of "uninteresting" targets (everything is interesting in its own right :-) -- this is called clutter.

Finally, the scattered signal that we received at the remote antenna is sent back to UW over the internet for signal processing. Because of the humongous amounts of data radars generate (for us, currently about 4 MB per antenna for 10 seconds of data), this data transfer step is a challenge, and one of the largest disadvantages of bistatic or multistatic radars.

Now that we have the scattered signal and the reference signal in one place, there is a large amount of signal processing to be done on the data in order to generate useful results. First, the reference and scattered signals are multiplied together to generate the detected signal.

Next, since our eventual goal is to see the power spectrum - the power of the signal of interest as a function of frequency - we need to obtain the autocorrelation of the detected signal (the correlation of the detected signal with itself).

As it happens, the power spectral density (PSD - power content at all frequencies) of a signal is the Fourier transform of its autocorrelation function. So, we calculate the autocorrelation of our detected signal, then Fourier transform it to obtain the PSD.

To add one more layer of complexity....
Because the radio station signal takes time to travel up into the ionosphere to be scattered by a target, a scattered signal will contain the reference signal at a certain delay, depending on how far away the target is. So, say a target that we see is at range "r". Then the scattered signal we receive -- call it y(t) -- will be the signal of interest, designated s(t), times the appropriately delayed reference signal, x(t-r).

Now we may calculate detected signals for many different ranges, by multiplying our scattered signal with many copies of the reference signal - each with a different delay. We calculate the autocorrelation and PSD for each of these detected signals (each representing scattered signal from a different range), and then we may create a plot of range vs. Doppler shift (frequency) for the entire dataset, which generally represents the average of a certain amount of time; in our case, about 10 seconds. This may be thought of as taking a picture of the sky with a camera, and controlling the shutter speed of the camera to make different exposure times.

As you can imagine, there are numerous other signal processing things going on in the system, and numerous non-ideal behaviors to model and correct for, etc. As they say, "Otherwise I wouldn't have a job!" :-)

For example, interferometry is one really cool way to get additional information out of radar data.

One interesting thing that we have found is that rock music works better than talk radio for our purposes. This is related to two concepts called range-aliasing and Doppler-aliasing.

So what are we trying to study here?
The primary purpose of our radar is to study the ionosphere, particularly E-region irregularities caused mostly by plasma interactions with earth's electromagnetic fields and the solar wind. Several terms in that sentence warrant further description...
Earth's ionosphere is the highest layer of the atmosphere - above the troposphere and stratosphere and everything - above the height at which airplanes fly. It is called the ionosphere because many of the atmospheric particles (O2, for example) have been ionized (had electrons chipped off) by the high-energy radiation from the sun. (One way our atmosphere protects us... it absorbs a lot of the energy in the sunlight before it gets down to us.) These positively-charged ions and electrons, of course, are constantly trying to recombine to form neutral molecules, so once the Earth rotates so that they're on the night side, they do recombine, and the lower part of the ionosphere "disappears" (ie, becomes neutral until daylight again). The lower ionosphere is much denser than the upper ionosphere, and also a larger percentage of (the lower part) is composed of neutral particles. The lower part has been designated the E-region, and the upper part the F-region.

The ionosphere is composed of plasma, which is ionized gas, or charged particles. You could think of wind blowing in the ionosphere as an electric current, since the wind is charged particles moving. Additionally, since the plasma is charged, its behavior is governed in part by Earth's magnetic field (a dipole field, with poles at the north and south poles). Charged particles tend to travel along magnetic field lines (in a corkscrew trajectory due to the Lorentz force), and this motion is called field-aligned current. In short, these currents are what cause the aurora, or northern lights: the fast-moving charged particles collide energetically with ions and neutral particles in the atmosphere, causing the particles to emit light as they decay to lower energy levels. The different colored light is caused from the different chemical makeup of the particles (for example, nitrogen causes the exquisite pinkish-red color, while oxygen produces a green... or is it the other way around?)

More about the aurora and space weather

The phenomena we're trying to study with our radar are related and very similar to the visual aurora. These "irregularities" have been called the radar aurora because they often occur with the visual aurora. They occur in the E-region of the ionosphere (thus, the name 'E-region irregularities'), and are caused by "streaming instabilities" and density gradients in the plasma: Sometimes sonic booms occur up there because a stream of electrons is moving much faster than the surrounding media (composed of the much heavier, slower-moving, and more prone-to-collision ions and neutral particles). Large pressure waves / density gradients occur when the electron stream moves faster than the speed of sound of the surrounding media. VHF radiation (the frequency of our radar) scatters very strongly from those waves, and can be easily coherently detected (which is what we're doing).

I would like to put in some graphics to help the discussions here, but that is probably something I won't be able to commit the time to for a while.

Disclaimer....
There are, of course, bunches and bunches of better references for learning about all of the topics here...