Potential methods to identify inhomogeneities in historical dew point temperatures

Recent efforts to digitize US historical climate data sets as part of the Climate Data Modernization Program (CDMP) have begun to make high-resolution hourly records available. However historical metadata associated with the earliest records are sparse as are sufficient neigboring stations to form conventional reference series for dew point. Therefore alternative methods to statistically determine breakpoints in dew point temperatures are explored. A number of potential tests are assessed, using data from 1951 - 2007, to determine which method provides the best break point detection. The tests utilize coincident temperatures and dew point temperature observations both annually and under specific weather conditions, such as clear calm nights and nights with fog, as their basis. Likewise we explore the possibility of using daily minimum temperature from neighboring Cooperative Network sites as a surrogate for dew point reference series. Four statistical break point detection techniques are also used to assess which is the most accurate. After the best method for detecting inhomogeneities in dew point temperature series is determined this method will be applied to the data that dates back to the 1800s to adjust dew point temperatures and create a homogeneous dew point temperature data set for the contiguous United States.