I saw some models (ACCES1-0, GFDL-CM3, CCSM4) of CMIP5 for the specific humidity variables have undefined values at the level of 1000hPa on the continents. Why does this happen? Thank you.
Most likely the undefined values are because the surface pressure is below 1000 hPa. In the conversion from model levels (sigma or sigma-hybrid) to standard pressure levels, a choice must be made as to whether to extrapolate to below the surface pressure or not. These models likely chose not to extrapolate.
Do areas of undefined values correspond to mountainous areas? The higher terrain would consistently be located higher than the 1000 hPa surface (i.e. lower pressure). That could be the simple explanation. (oups .. my answer came just after Brian's ... sorry for the repeat).