The Further reading link brings you to The PERSIANN System Fine Resolution Precipitation Estimates Using PERSIANN-Cloud Classification System which shows that the data is merged with signals coming from both geostationary and polar/near polar satellites. My guess is that it's just banding in one of these datasets (probably from that second group of satelites).
I tried to look at raw image data from these satellites (to illustrate my assumption), but I bump into sites that are affected by the current US government shutdown, resulting in e.g. incorrect and expired CSS certificates.
If you can dig up these raw images in a later stage, you'll probably be able to map the width of those bands to the aperture of the satellite.