Beyond the Echo - Helicopter rotor signature on SAR imagery

in our #beautiful_sar_images internal channel by Katherine Peterson and Qiaoping Zhang - big thanks to them.  A few days ago (5th November) we collected this neat image of Khmeimim Air Base in Syria. The colours in the CSI (colour sub-aperture image) really highlight the different human-made components of the base.

What was particularly interesting was that the western runway had two MI-26 Hind helicopters and it was really easy to count the 5 rotor blades and see the characteristic double engine nacelles above and behind the cockpit (Image 2).

Image 1. ICEYE SAR satellite image of Khmeimim Air Base, Syria, 5th November 2024 (14:43 local time)

Image 2. An MI-26 Hind helicopter on the runway, 5th November (14:43 local time). Zoom-in of the Image 1 inset.

The ICEYE satellite is in a one-day repeating ground track meaning that every day we get to see the same location from exactly the same imaging geometry - something that is really useful when trying to detect change with RADAR imagery (or even just to better understand what you are looking at).

So, two days later, when we looked at the same location, we were surprised to see a very different set of signatures.

Image 3. ICEYE SAR satellite image of Khmeimim Air Base, Syria, 7th November (14:43 local time)

Image 4. Two MI-26 Hind helicopters, 7th November 2024 (14:43 local time). Zoom-in of the Image 3 inset.

The obvious question is 'why do the helicopters look so different?' 

Not just a picture

Fortunately we have another tool at our disposal. Because we collect for 25 seconds we usually also make a video with reduced azimuth resolution (25cm) as the satellite moves past. You can see the results below (Image 5).

In this animation I've made an inset of the helicopter signatures. You can see that they vary from frame to frame, as the satellite passes by. In the first frame there is a large smear spreading out from the location where the helicopters are. It seems to spread down the runway, around the mid-way point the smear separates into two rows of dots centred on the location of each helicopter and then fades away towards the end of the video. The bright signature in the early frames dominates the final SAR image - which can be thought of as a composite of all the individual frames in the video.

But what really piqued my interest was around the middle frames where we see the signature (Image 6).

Image 5. SAR video of Khmeimim Air Base, Syria, from 7th November. Helicopter signatures inset.

Image 6. Zoom-in of the helicopter signatures at frame 8 in the video sequence.

Some geeky background information

Inside the SAR Image formation processor is a neat bit of signal processing called a Fast Fourier Transform. It's very clever, but tricky to understand (my favourite explanation is the one by 3Blue1Brown). Its main task in a nutshell is to look for repeating signatures from pulse to pulse in the RADAR data- specifically signatures that have a known Doppler frequency variation caused by the satellite moving over the scene. This is how a synthetic aperture is formed. When the SAR processor finds a regularly repeating signature it gathers all the energy into one location in the image. The location depends on the frequency detected, but importantly the displacement is always in the cross-range direction (perpendicular to the direction we observe from - I'll come back to why this is important).

Explanation

In the above image for each helicopter we see a horizontal row of dots with a brighter object in the middle. This is a characteristic signature of something that is flashing or vibrating. In this example the radar pulses are passing through the rotor blades of the helicopter and looking at its main body - which is quite bright to the RADAR. Every now and again - perhaps a few pulses in every thousand, the helicopter blade gets in the way. When this happens the body of the helicopter gets slightly dimmer.

Because the processor is really sensitive at picking out periodic changes it sees the periodic dimming as a frequency. Its not quite the signature expected to make a synthetic aperture, so it gets a bit confused and rather than putting the energy reflected from the helicopter in its true location, it splits it up and shares the energy equally into two diffuse balls of energy - one for a positive frequency, one for a negative frequency but each shifted in the instantaneous cross range direction. Higher frequencies move the signatures further from the source and many repeating flashes eg from multiple rotor blades, give rise to additional diffuse balls.

An interesting thing occurs in a SAR video - the direction that we are looking from, changes from frame to frame. This means that the direction of 'cross-range' where the flashing signatures are moved to, also changes from frame to frame. If you look back at the animation of the SAR video, you can now see that the row of dots appears to rotate around the location of the helicopter as the video plays. This is because each frame has a different cross range direction.

So what about the rest of the video

At the beginning of the video, by coincidence, the cross range direction is aligned quite closely to the length of the runway. In this orientation the helicopters look quite bright - they are imaged from the side, and also the rotor signatures are displaced on top of each other making quite a mess.

At the end of the video the signatures are barely visible. Now we have an understanding of the nature of the signature we can speculate that this is probably because the helicopters are not very 'RADAR bright' from this orientation - or it could be that we just don't see anything bright through the rotating propeller blades.

Summary

Ok so what? It's pretty obvious that helicopters parked at airfields will operate now and again. So this is just an intellectual curiosity. But I think it's kinda neat how much additional information there is in a SAR image waiting to be mined. It feels like we haven't really started yet and we have a lot of ideas of things we'd like to try. (And we also consider ideas sent to us on social media or via the university networks).

In this example, its pretty obvious that the two MI-26 Hind helicopters, sitting on the runway on the 5th November are fully operational and probably about to take off on the 7th November at 2:45 in their afternoon. These are clearly Army pilots as everyone knows that Navy pilots have their tea-break at 3 o'clock.

Thanks

I learned about Vibrometry and 'Fan-omenology' from some smart guys in Michigan, LA, and D.C. at the turn of the century. Hopefully they see this as proof that I listened rather than plagiarism 😉 (I did come up with the name though).