Help comparing UOBGS .dat waterfalls with SatNOGS Station 91 for RFI analysis

Satellites & Observations
1

Hi everyone,

I am currently working on an RFI-detection project using waterfall spectrograms from UOBGS and comparing them with observations from SatNOGS Station 91 – M0EYT / 2E0NOG.

The satellite I am analysing is ROBUSTA-3A (NORAD 60243) around 436.750 MHz.

I would really appreciate some guidance from the SatNOGS community on the most scientifically appropriate way to perform this comparison.

For the UOBGS observations, I have the original numerical waterfall recording files in .dat format. From these I can recover the time-frequency waterfall matrix and perform numerical processing.

For the corresponding SatNOGS Station 91 observations, however, what I currently have available is mainly:

  • observation metadata,

  • waterfall PNG images,

  • audio where available,

  • observation timing and transmitter information.

My main question is:

How would you recommend comparing the two stations’ waterfalls to investigate whether an unusual spectral feature is likely to be RFI?

For example, if an anomalous feature appears in the UOBGS waterfall, I would like to investigate whether a corresponding feature is also visible at Station 91 at the same absolute UTC time and corresponding RF frequency.

At the moment I am considering:

UOBGS .dat numerical waterfall → anomaly/RFI-candidate detection → UTC/frequency registration → comparison with the Station 91 waterfall.

However, because the Station 91 product available to me is a rendered PNG rather than the same type of numerical waterfall data, I am unsure what comparison method would be considered scientifically defensible.

In particular, I would really appreciate advice on:

  1. How best to align the two observations in UTC and RF frequency, considering that each station experiences a different Doppler trajectory.

  2. Whether comparing spectral morphology/features between a numerical .dat waterfall and a SatNOGS PNG is reasonable, and what metrics or methods would be recommended.

  3. How SatNOGS users normally distinguish a genuine RFI feature from receiver artefacts, satellite signal structure, noise, or processing artefacts in waterfalls.

  4. Whether the absence of a feature at Station 91 can provide meaningful evidence that an anomaly observed at UOBGS is local, or whether receiver/antenna/sensitivity differences make that conclusion unsafe.

  5. Whether the original numerical waterfall data used to generate the Station 91 PNGs may exist somewhere, for example as a .dat, HDF5 waterfall, or another numerical product.

If numerical Station 91 waterfall data are available, that would potentially allow a much more direct time-frequency comparison with my UOBGS .dat recordings.

I already have several same-pass ROBUSTA-3A observations from Station 91 and can provide the exact observation IDs, UTC dates and times if that would help.

Any advice on how SatNOGS operators/researchers would approach this comparison would be greatly appreciated.

Thank you!

2

Maybe you can clarify UOBGS ? I am not familiar with the abbreviation.

Station 91 is using a pretty old SatNOGS client and also hasn’t activated the artifacts (HDF5) option.

Have a look if these observation that do have artifacts enabled and see if those make your investigation easier.

Look at the bottom left corner.

image
image925×739 132 KB

Jan | PE0SAT

3

3290 - University of Bristol

Hi PE0SAT,

Thank you very much for the clarification.

UOBGS refers to the Station 3290 (University of Bristol)

That explains an important issue I was encountering with Station 91. For my UOBGS recordings I have numerical waterfall .dat files, whereas for the corresponding Station-91 observations I currently only have the rendered waterfall PNGs, metadata and audio.

I will look at the observations you linked that have HDF5 artifacts enabled. My aim is to investigate whether unusual spectral features detected in the UOBGS numerical waterfall are also present at an independent SatNOGS station at the corresponding UTC time and RF frequency, while taking station-specific Doppler into account.

Would the HDF5 artifact be the product you would recommend for this type of numerical time-frequency comparison?

Thanks again — this is very helpful.

1 Like