On a staggeringly smaller scale I had VERY mixed results with storing data points that look like (satellite ID, timestamp, el, az, signal strength) and then doi

On a staggeringly smaller scale I had VERY mixed results with storing data points that look like (satellite ID, timestamp, el, az, signal strength) and then doing a data analysis to map out "anomalous attenuation" by el,az (kind of a bistatic radar display). "Everyone knows" since day 1 of GPS back in the 80s that tree leaves and rain storms ruin GPS reception, but rather than complaining about it and/or avoiding it, why not use it as a data source? Turns out I had trouble pulling a useful signal out of the noise. Maybe with a lot more data or a lot more RX running in parallel or ... There are of course inherent limits as discovered by Air Forces over the decades where bistatic radar works best directly overhead whereas I was interested in "storm clouds on the horizon" exactly where it didn't work. I had some optimistic imagination in that if heavy rain causes 30 dB of attenuation, then extreme data analysis might indicate something as trivial as cloud cover, or even fog, might cause 0.01 dB of attenuation and I could map out regular non-rain clouds or something. That data could not be pulled out of the noise. Very optimistically, if I placed antennas on corners of my property, I could "edge detect" even if the signal level varied randomly-ish by comparing the multiple RX and doing some trig. I live close enough to an airport that occasionally a plane should be between me and a satellite. That did not work either. Anyway, the linked distributed project is my merely a thousand times larger than my playing around experiment.