Like all other statistical downscaling methods, LOCA uses observations as the basis for constructing the final fine spatial resolution field.
LOCA2 (CMIP6, ca. 2022):
For daily precipitation we use what I call the “unsplit Livneh” data described in Pierce et al. 2021. This does a better job preserving daily precipitation extremes than does the previous Livneh et al. data, such as was used in LOCA (CMIP5).
For daily minimum and maximum temperature, we use an extension of Livneh et al. 2015 that updates the data to 2018, prepared by Lu Su of the Department of Geography, Land Surface Hydrology Research Group at UCLA.
You can download all the LOCA2 training data here:
https://cirrus.ucsd.edu/~pierce/nonsplit_precip/
LOCA (CMIP5, ca. 2016):
For daily precipitation, maximum, and minimum temperature, we use the Livneh et al. (2015) data set over the period 1950-2005. The data can be downloaded from:
http://data.nodc.noaa.gov/cgi-bin/iso?id=gov.noaa.nodc:0129374
The paper describing the Livneh data set can be found in the Bibliography.