Well, this is sort of embarrassing.
After a lot of research, it turns out this is just lightning. It turns out, this is the first time I’ve detected lightning since I changed the mechanical filter on the Boom from 1s (1Hz) to 20s. With the 1s filter each lightning strike is quite distinct. With the 20s filter it appears there may be low frequency infrasound generated from both pre- and post-strike charge movement. i.e. the infrasound begins to build before the sound of the lightning strike arrives (indeed before the strike occurs!) and continues afterwards. Afterwards could be the result of echoes, etc but before is hard to explain any other way.
Lightning strike times and positions were obtained from LightningMaps.org. The closest strike was at 03:04:17.558UTC. Lightning strike times are very accurate but positions have a significant error circle plotted around them. So the plotted position from the map in LightningMaps.org is used for the first pass calcs in the report and then the true distance to the strike calculated from the sound travel time.
The sound travel time was calculated at 29.442s, which at 320m/s corresponds to 9.42 kms.
The cluster of lightning strikes on the Newnes Plateau ~100kms away are also at the right time to coincide with the peaks later - at +250s and +430s. Again the low frequencies are not attenuated much so the signals all appear linked at low frequencies.
Must remember the Boom with a 20s filter can have some very different results to the standard filter!
Al.