The study uses the National Weather Service (NWS) CPC's methodology for downscaling seasonal temperature forecasts for use in NWS Western Region climate products. The downscaling involves regression of the observed temperatures at the forecast division and the individual stations. The regression coefficients are adjusted for the difference between the probability density functions at specific stations and their corresponding climate divisions. An adjustment is also done to account for the difference in temperature trends between the stations and climate divisions. The regression relationships are then applied to the CPC probability of exceedance (POE) forecasts to obtain predictions for specific localities in the Western US.

The method was applied for 87 sites located in NWS Western Region, in partnership with the NWS CPC and CSD, with the ultimate goal to develop local climate products. Spatial analysis was conducted for each of 12 overlapping three-month seasons and included identification of locations where the variance explained by the linear regression between the stations and climate divisions was less than fifty percent. In general, the variance explained by the regression is about eighty percent for most stations and seasons. The seasons that included the month of June show variance explained by the regression as poor as nine percent for coastal stations located in northern California and Oregon. The relationship is strongest in the winter seasons.