Establishing Long-Term Temperature Trends in California Amidst Data Set Variations

Abstract:

Close attention is being paid to California’s water resources amidst drought conditions including the Sierra Nevada snow pack depth. Warm conditions and warm winters contribute to reduced winter snow accumulations. We examine long-term trends (1920-2015) of average daily maximum (Tmax) and minimum (Tmin) temperature as estimated by different long-term records, specifically: a) UCLA’s West Coast Surface Water Monitor (SWM), b) the Parameter-Elevation Regression on Independent Slopes Model (PRISM), c) the Berkeley Earth Surface Temperature (BEST), and c) the National Climatic Data Center’s (NCDC) (VOSE) data set. We also examine climatological values for Tmax and Tmin as estimated by Livneh et al. (J Clim., 2013) and Maurer et al. (J Clim., 2002) as these are related to the SWM gridded data set. We draw on station data from the U.S. Hydroclimatic Network (HCN) and the U.S. Cooperative Observer Network (COOP) and the temperatures published by NCDC as made available via ncdc.noaa.gov/cag/time-series/us for comparison. Within each data set, Tmin has stronger uptrends than Tmax. For both Tmin and Tmax, all but one of the data sets have increasing (mostly statistically significant) trends. Minimum winter temperature trends range from 1.3-1.8 C/100 years across the state; maximum winter temperature trends range from near zero to 1.0 C/100 years. Maps of trend magnitudes at the grid cell level show a surprising lack of agreement in spatial pattern likely due to differences in how each data set was constructed. Some data sets show nearly uniform trends due to the use of spatial smoothing, while others show highly varied local trends. We evaluate differences among the data sets in the stations used, periods of record, and gridding algorithms in an attempt to account for the variations in inferred temperature trends.