Statistical downscaling methodology was applied to global atmosphere reanalysis, using a 10 year dynamical modeled output as training data, to extend a record of wind and humidity back through the satellite era beginning 1979 or 1980. To extend the record of weather to cover a longer period and thus a greater number of significant fire weather events, statistical downscaling was implemented using the Localized Constructed Analogs (LOCA) statistical downscaling technique. LOCA is computationally efficient and has been designed to better simulate extreme events and spatial weather structure than previous statistical downscaling. The output hourly 3 km spatial resolution data covering the California and Nevada region was downscaled from each of the two global atmospheric reanalyses: a) the ERA5 Reanalysis covering January 1979 through December 2019; b) the MERRA2 Reanalysis covering January 1980 through December 2018. These two reanalyses, ERA5 and MERRA2, are acknowledged to provide high quality, dynamically-consistent representations of historical global weather. The time resolution was extended from daily to hourly time samples, given the 10 years of hourly, 2km WRF dynamical model historical simulation from 2004-2013 supplied by co-investigator Tim Brown and colleagues at Desert Research Institute, called the “DRI-WRF” dataset. The LOCA dataset was generated over California and Nevada for each hour of 1979-2019, the satellite era that comprises the ERA5 Reanalysis. In the LOCA downscaling process, cross validation tests were employed to evaluate LOCA performance in simulating winds and humidity, especially in cases of extreme high winds and low humidity that are particularly important in this fire weather application. The LOCA downscaled winds and humidity were supplied to co-PIs of this project for their investigation of daily weather influences on wildfire in California.