A set of tools is being developed to detect changes in relationship between one time series (e.g. a climate record) and one or more other time series. Built on some tenets of double-mass analysis, the tools are intended to help identify subtle changes that might be otherwise overwhelmed by the large cumulative values associated with long-term records.

Several design considerations were implemented in their development. First, the tools must be able to work on datasets of any time scale and observation interval. Second, they must function properly regardless of the variables being compared. In addition they should be able to work on retrospective data and on operational incoming data flows where the future values in the stream are not available.

Many traditional methods for examining and viewing long-term climate records often mask seemingly minor station moves, instrument replacements, and sensor drift. The tools have been applied to a number of station pairs in the western United States. Among the findings were evidence of increased urban "heat island" effects, marked sensitivity to elevation, and undocumented station moves, the latter often seemingly innocuous.

This set of tools can be used in a real-time sense to detect inhomogeneities in the climate record. A particular value to the U.S. Climate Reference Network is foreseen, in both inter-station and intra-station application (the latter among redundant sensors, and across elements). Automation of these tools is under investigation, with the ultimate goal of unattended examination and reporting of points of suspicion/interest in the data record, and suggestions, perhaps probabilistic for resolution.