A case study on the impact of homogenizing monthly temperature series along coastal California

Modeling and observational studies have shown that warming trends in summer daytime maximum temperatures in interior California since 1970 have led to increased onshore flow along the coast, leading to cooling trends in basins that allow for the penetration of the marine layer. Unfortunately, since the mid-1980s most of the U.S. Cooperative Observer Network stations that are available to quantify these changes transitioned to new type of thermometer known as the Maximum Minimum Temperature System (MMTS). The newer MMTS instrument records maximum temperatures that are, in general, systematically lower than temperatures recorded by the previous instrumentation (liquid-in-glass thermometers), which complicates analysis of the observed temperature fields. Homogenization (bias-correction) of the monthly temperature data is carried out to remove the impacts of the instrument and other changes. However, the real coastal cooling trends are highly localized and are of the same sign as the changes expected from the transition to the MMTS. Consequently, it is unclear how successful the homogenization process has been at separating the artifacts caused by instrument and other changes in observation practice from real mesoscale climate changes. Here we thus report on a detailed case study quantifying the adequacy of the homogenization efforts used to produce the United States Historical Climatology Network (USHCN) version 1 and version 2 monthly temperature datasets and also present a new estimate of the pattern of observed temperature changes along coastal California.