Awesome work. SAR imagery is notoriously fickle, and it is amazing to see these results from a DIYer.

Related. Others?

Homemade polarimetric synthetic aperture radar drone - https://news.ycombinator.com/item?id=43073808 - Feb 2025 (60 comments)

I had fun reading this -- the radio astronomy technique called VLBI (very long baseline interferometry) has a ton of overlap, but the jargon words are fairly different. There are plenty of differences: our telescopes are mostly on the surface of the Earth and don't move, VLBI isn't a radar so there's no waveform, etc.

The EHT black hole telescope is an example of VLBI.

I'm an embedded engineer for 15 years now and did some really spicy stuff during my career.

This hobbyist SAR project made by hforsten absolutely nuts. Wonderful.

He mastered mechanics, electronics, software and applied higher math of the highest difficulty grade. Guess we have to call this "The Holy Quadrivium".

Back to the tech:

Radar stuff is very interesting. Reverse Engineering cheap FMCW radars from China makes these very accessable to hobbyists too. Often you "only" have to care about the VCO for the frequency sweep and then read back the I/Q channels using two sync'ed ADCs. Then you can play around with algorithms to analyze the signals.

These 24GHz Car Speed CW (you see next to streets) radars can be used for classifying rain drops for example. They antenna pattern is quite sharp and they have high gain.

This is an excellent article and great work.

We need more articles uncovering the SAR mechanics, even for Satellite Data.

Slightly tangential: this is a wonderful and deep project, that requires a lot of personal time. Lately I've been wondering what social/economic/govt conditions allow for this type of deep thinking + tinkering among working people (not academia). My very rough guess is the US of 1950-60s did, and some other countries today do, but not so much the US of today because the cost of living and time pressures are higher. I'd be curious if anyone has a more detailed answer (or a rebuttal of my thesis altogether).

I recall a post on HN about a decade ago where someone made an SAR by modifying one of those toy radar-guns into an FM CW Radar and mounting it on his bike. After trying a few different position-estimation algorithms, he got the best results just by maximizing the contrast.

This is mad in all the good ways.