A Comprehensive, High-Resolution Database of Historical and Projected Climate Surfaces for Western North America

We present a comprehensive set of interpolated climate data for western North America, including monthly data for the last century (1901–2009), future projections from atmosphere–ocean general circulation models (A2, A1B, and B1 scenarios of the WCRP CMIP3 multimodel dataset), as well as decadal averages and multiple climate normals for the last century. For each of these time periods, we provide a large set of basic and derived biologically relevant climate variables, such as growing and chilling degree days, growing season length descriptors, frost-free days, extreme minimum temperatures, etc. To balance file size versus accuracy for these approximately 20,000 climate surfaces, we provide a stand-alone software solution that adds or subtracts historical data and future projections as medium-resolution anomalies (deviations) from the high resolution 1961–90 baseline normal dataset. The program further downscales the baseline data through a combination of bilinear interpolation and elevation adjustment using partial derivative functions. Observations from 3,353 weather stations were used to evaluate the climate estimates of our downscaling algorithms. We found that the algorithms substantially improved prediction accuracy of the monthly climate variables, especially for temperature variables. They eliminated up to 65% of the unexplained variance in observed monthly temperatures and reduced standard errors of climate estimates by up to 40%.