We downscaled CMIP6 data using LOCA version 2 over two domains: 1) most of North America from Central Mexico through Southern Canada, and 2) California. The differences between the North American and California CMIP6 downscaled data sets are:
- The North American data set is at a spatial resolution of 6 km; the California data set is at a spatial resolution of 3 km. Both have daily temporal resolution.
- The California data set uses a hybrid downscaling scheme, which the North American data does not. Briefly, in the California data set, coarse-resolution GCM output for days at the end of the century are downscaled using a library of patterns taken from a set of 4 WRF dynamically downscaled simulations at the end of the century. The purpose of doing this is so that the future weather patterns in the downscaled data can reflect any systematic changes in weather patterns that WRF projects to occur in coming decades.
- The North American data set consists of 27 models, which is all the models available from the CMIP6 data set that had the required data when we started the LOCA2 downscaling project. The California data set consists of a subset of 15 models that performed well over the California domain, as evaluated in Krantz et al. (2022).
- The North American data set uses Pierce et al. (2021) as the precipitation training data set. I refer to this as “unsplit Livneh”; see here for more details. Essentially, unsplit Livneh is a standard kind of gridded product that uses a form of inverse distance weighting to interpolate data between meteorological station locations. The California data set takes a different approach. Data between stations is interpolated using results from WRF driven by the ERA5 reanalysis. We believe this provides a more physically realistic way of interpolating between station observations than is possible using inverse distance weighting.
- The North American data set provides daily Tmin, Tmax, and precipitation. The California data set provides daily Tmin, Tmax, precipitation, specific humidity, 10-meter U and V winds, 10-meter wind speed (which is downscaled separately from the U and V components), daily relative humidity maximum and minimum, and daily surface downward solar shortwave radiation.
- The California data set provides hourly temperature data at selected station locations developed using the method of Pierce and Cayan (2019).