A Method to Detect Inhomogeneities in Historical Dewpoint Temperature Series

Hourly dewpoint temperature data for the 1951-2006 period at 10 stations in the contiguous United States were investigated to determine if inhomogeneities in their records could be detected. At least three instrument changes are known to have occurred during this time period. The relatively sparse network of stations with dewpoint temperature data in the United States necessitated a nonconventional method to create a reference series. Utilizing nighttime occurrences of fog, clear/calm conditions, and precipitation as meteorological situations during which dewpoint temperatures and minimum temperatures are similar, three potential reference series based on daily minimum temperature were developed to test for inhomogeneities. Four stations with independent network neighbors recording hourly dewpoint data provided a direct validation of the effect of inhomogeneities on dewpoint temperatures. It was determined that fog conditions and the combined results from all three meteorologically based tests performed best when detecting documented inhomogeneities. However, a larger number of undocumented inhomogeneities, a feature common in most traditional inhomogeneity tests, were also detected that may or may not be valid.