The identification of systematic nonmeteorological changes in precipitation observations is critical to many applications, but especially so to climate change studies. At gage sites, these changes commonly involve location, elevation, exposure, or instrumentation, leading to systematic increase or decrease in precipitation amounts. Other, more subtle effects involve changes in observation time, method, consistency, etc. These kinds of changes may have a greater effect on precipitation frequency, timing, variability, and completeness than on the actual precipitation totals. They may also be more difficult to find.

Although procedures exist to identify "change-points" in individual time series, the large natural temporal variability in precipitation seriously complicate methods that rely solely on long-term characteristics from a single observing location. Comparison with other sites that are likely to share the same precipitation characteristics can provide additional information to the search for subtle changes. To facilitate this process, it is useful to formulate time-dependent measures of spatial correlation. Time series of these measures can then be subjected to change-point analyses that may identify times at which a station's relationship to its neighbors has significantly changed. We describe several such indices and assess their usefulness by applying them to daily precipitation observations measured at selected U.S. cooperative climate stations.