I tried searching for this answer, but all of the pages just use the term 'superblock' without defining what it is.
To help protect against loss/corruption, filesystems (and sometimes RAID arrays, depending on configuration) will store copies of its critical configuration information (partition size, block size, journal location, type, etc.) in several different locations throughout the physical disk. This way if the main copy of the fs configuration is lost, it can restore from one of the superblocks. The locations where this metadata is stored are called superblocks.
It's a part of the partition reserved for the software/firmware that manages the drives, Linux MD driver uses few last sectors for example.
The contests of them are usually quite simple: the array identifier (usually a UUID), number of drives in the array, array layout (whatever it's a RAID1, RAID5, etc., length of stripe), some do contain additional information: precise location of other devices in the array, or even UUIDs of specific devices forming the array.
Linux MD driver uses this space also for internal write intent log.
mdadm -Q --examine /dev/hda1
will show you most contents of the MD RAID superblock.