Artificial change point detection in temperature series in the USHCN version 2

The original USHCN datasets contained temperature adjustments for artificial change points that were identified using digitized station histories from the metadata archives. Historical metadata, however, are known to be incomplete. Consequently, in the USHCN version 2, an evaluation of both documented and undocumented change points in serial monthly temperature values was conducted. The distinction between documented and undocumented change point tests is discussed as well as an alternative to the traditional use of composite reference series to identify relative change points. Examples of temperature time series analysis using adjusted and unadjusted values are presented.