To support high spatial- and temporal-resolution land surface modeling, this dataset provides 3-hourly time step historic weather forcing at 1-km spatial resolution for Puerto Rico and surrounding islands. The latest Daymet V4 data provides gridded historic daily weather observation at 1-km spatial resolution from 1950 to present. Using sub-daily temporal information from two meteorological reanalysis datasets (GSWP3 and NARR), Daymet was further temporally downscaled to 3-hourly time steps and provided in the format required for land surface model simulations. The process of temporal downscaling preserves the relative magnitude in each sub-daily time step from GSWP3 and NARR while maintaining the total and average values from Daymet at each day. These result in two blended datasets: 1950-2014 Daymet-GSWP3 and 1979-2019 Daymet-NARR. Available variables include surface air temperature, precipitation, specific humidity, shortwave and longwave radiation, wind speed, and pressure. These data can be used as a high-resolution meteorological forcing dataset to support high-resolution land surface modeling where accurate meteorological forcing datasets built from historic observations and/or reanalysis datasets are desirable.