Temporal data retains a comprehensive history of changes made to it, together with the dates when those changes took place.
Consider a data warehouse with a fact table and multiple dimension tables where the fact table conventionally contains time-stamped immutable facts. If the fact table continuously accumulates facts over a long expanse of time, it is likely that the dimension data will need to change over that period - but if we simply update dimension data, we are effectively changing history. We need a version of the dimension data which relates to each fact on the date of that fact.
Global and central tables are temporal such that all versions of the data are retained and historic message data can be joined with the relevant versions of dimension/reference data.