From the descriptions provided about the activities on the top of the tank, most likely an electrostatic ignition of the vapours within or at the opening where someone was inserting something conductive - hand, metal, etc. or an internal conductive element lost its bonding to the tank shell. The organic liquid surface can be at very high potential just due to settling out of dispersed water that is common in pygas tanks and transport.
if the tanks were internal floating roof, may have been similar, particularly if the level was below the leg height.
Will be interesting if they publicize the actual technical and operational findings.