Researchers from Carnegie Mellon University have recently presented a paper detailing how they combined noisy signals from multiple passes of low earth orbit (LEO) satellites NOAA 15, NOAA 18 and NOAA 19 in order to create a higher quality image. For a receiver they used a low cost RTL-SDR Blog V3 mounted indoors with a whip antenna.

In a normal setup, weather satellite images from NOAA LEO weather satellites can be received with an RTL-SDR, computing device and an appropriate outdoor mounted antenna that has a good view of the sky. If the antenna is not suited for satellite reception, and/or is mounted indoors, at best only poor quality very noisy images can be received.  

The researchers demonstrate that it is possible to combine noisy images received over time, and from different satellites in order to generate a higher quality image. The challenge is that the different satellites and different receiving times will all produce different images, because the satellites will be at a different location in the sky each pass. They note that simply transforming the images in the image domain would not work very well for highly noisy images, so instead they have devised a method to transform the images in the RF domain. The RF signals are then coherently combined before being demodulated into an image.

The results show that 10 noisy satellite images from the indoor system are comparable to one from a comparison outdoor system. However, they note some limitations in that the system assumes unchanging cloud cover during passes. In the future they hope to extend the system to cover other modulation schemes used by other low earth orbit satellites in order to increase the number of usable satellites.