Using 40-member ensemble mean of 1920-2005 monthly data in CESM historical experiment, we remove the ensemble mean (radiative forced component) in each member,
and obtain the internal variability. Then we calculate the annual mean (defined as March to next February) based on the monthly data. 
To obtain decadal variability, we use a 10-year low-passed Lanczos filter; the high-passed component is then treated as interannual variability. 
The variable name and its unit of each file is shown as follows.
For historical experiment:
annual_lh_minusEN_0825.mat:  latent heat (W/m2)
annual_prw_minusEN_0825.mat: water vapor path (kg/m2)
annual_transmag_minusEN_0825.mat : horizontal moisture transport (kg/m·s)
Each data includes four variables: lon—longitude, lat—latitude, int—interannual variability, and dec—decadal variability.

For RCP8.5 experiment:
RCP85_annual_lh_minusEN_0825.mat:  latent heat (W/m2)
RCP85_annual_prw_minusEN_0825.mat: water vapor path (kg/m2)
RCP85_annual_transmag_minusEN_0825.mat : horizontal moisture transport (kg/m·s)

The CESM data was obtained from https://www.earthsystemgrid.org/dataset/ucar.cgd.ccsm4.cesmLE.html. Please cite Kay et al. 2015 when use this data. 

Kay J E, Deser C, Phillips A, et al. The Community Earth System Model (CESM) large ensemble project: A community resource for studying climate change in the presence of internal climate variability[J]. Bulletin of the American Meteorological Society, 2015, 96(8): 1333-1349.