How "point" are point targets?

Submitted by tomas on Wed, 04/22/2009 - 07:55

Last week I ran into an existential doubt.

We are operating an airborne L band SAR. For calibration we are using corner reflectors. Due to the the radar frequency (L band) this are 4 m big as we are also planing to use them with spaceborne sensors. Due to our PRF and our plane speed we have an azimuth pixel spacing of 0.4 m so our target is much bigger than it.

When we look at the interpolated impulse response function we see an unusual wide main lobe with some secondary lobes structure on it.

Are these CR to big to be considered as Point Targets? Can the widening be a consequence of this?

‹ Help with Freeman-Durden decomposition theory reading the ASF leader file ›

#1 Re: How "point" are point targets?

Submitted by fmeyer on Tue, 04/28/2009 - 10:02.

Thomas,

Here are my thoughts on this:
The main parameter that makes a corner reflector act as a point target is its unique shape. The shape causes all incoming energy to be reflected back to the source for a wide range of incidence angles. In the case of SAR, this means that a corner reflector will reflect the full azimuth bandwidth of your system as long as the beamwidth of the antenna is less than the beamwidth of the corner reflector (which is more than 10 degrees and should be uncritical). The size of the reflector merely modulates the brightness of your return but does not change its shape. So, a corner reflector should still act as a point target even if the resolution of your system is higher than the reflector's size.

Here my suggestions about what might have happened with your data:
1. The signal energy reflected by your corner reflector may have been too strong, causing your system to saturate. Saturation will cause your point target response to be clipped and will affect the azimuth phase history (azimuth chirp). This could cause increased sidelobes and loss of resolution

2. your azimuth focusing might not have done a good job. You could check this by analyzing the phase signature in the spectrum of the point target response. If focused correctly, your spectral point target phase should be flat. An incorrect FM rate would result in a residual quadratic phase signature in the spectrum.

I suspect it is more cause #1 than cause #2, as #2 would affect other areas of the image as well (maybe not not visibly though). If saturation is the cause, I would suggest to reduce the receiver gain in your system. For a better analysis of your problem I would need to see some plots of your point target responses.

Please let me know if this answers your question.

#2 Re: How "point" are point targets?

Submitted by tomas on Fri, 05/08/2009 - 12:13.

We may be having some kind of problem in azimuth compression. Measuring the phase gradient for our corner we get:

range phase gradient (degrees/sample): 2.08473
azimuth phase gradient (degrees/sample): 58.12412

58 degrees seems too much.

We may be having some trouble with the doppler centroid estimation as it is not constant along track due to the plane's attitude, as it's typical for an airborne SAR. The better we focus the image (updating doppler along track) the more observe the peak doubling.

I'm attaching the interpolated IRF and a small subset of the corner area eight times multilooked in pauli decomposition.

[attachment=0]corner.zip[/attachment]

#3 Re: How "point" are point targets?

Submitted by fmeyer on Tue, 05/26/2009 - 13:09.

Well,

I would also look into the FM rate of your azimuth chirp and see if a variation of the FM rate fixes the problem. Dispersion of the signal in time is often caused by a FM rate mismatch (of course a strongly varying Doppler can cause similar problems). Basically, any non-linear phase signatures that is still visible in the spectrum of your focused point target needs to be removed to optimize focusing. A wrongly applied FM rate would leave quadratic phase structures in the focused image, causing dispersion and defocusing.

So I would suggest to check both Doppler Centroid and FM rate. The focusing in range looks good, by the way.

Franz

#4 One of the ideas behind a

Submitted by jnicoll on Tue, 07/07/2009 - 07:26.

One of the ideas behind a corner reflector is that if you calculate the distances, then the triple bounce out of the reflector all are the same distance, identical to the center point of the reflector. The Ideal Reflector does this, and I think it does it out to infinity. Imperfections in the reflector  could mar this, but I suspect that edge effects are messing with the corner reflector response. For the resolution you are talking about, you are probably seeing the edges of the reflector as a separate target.
The only other thing I can think of is processing imperfections, such as a poorly executed sidelobe suppression.
Oh, and maybe this only works for parallel rays, and with airborne and a resolution cell smaller than your reflector, there is something happening with non-parallel waves changing the travel distance, and therefore smearing the target?
 
It is curious.