Free of any atmospheric contamination , the Hubble Space Telescope provides the best available spectrophotometry from the far-UV to the near-IR for stars as faint as V \sim 16 .
The HST CALSPEC standard star network is based on three standard candles : the hot , pure hydrogen white dwarf ( WD ) stars G191B2B , GD153 , and GD71 , which have Hubeny NLTE model flux calculations that require the atomic physics for only one atom .
These model flux distributions are normalized to the absolute flux for Vega of 3.46 \times 10 ^ { -9 } erg cm ^ { -2 } s ^ { -1 } Å ^ { -1 } at 5556 Å using precise Landolt V band photometry and the V bandpass function corrected for atmospheric transmission by M. Cohen .
The three primary WD standards provide absolute flux calibrations for FOS , STIS , and NICMOS spectrophotometry from these instruments on the HST .
About 32 stellar spectral energy distributions ( SEDs ) have been constructed with a primary pedigree from the STIS data , which extends from 1150 Å for the hot stars to a long wavelength limit of 1 \mu m. NICMOS grism spectrophotometry provides an extension to 1.9 \mu m in the IR for 17 of the HST standards and longward to 2.5 \mu m for a few of the brighter stars .
Included among these HST standards are Vega , the Sloan standard BD+17 ^ { \circ } 4708 , three bright solar analog candidates , three cool stars of type M or later , and five hot WDs .
In addition , four K giants and four main sequence A-stars have NICMOS spectrophotometry from 0.8–2.5 \mu m. The WD fluxes are compared to their modeled SEDs and demonstrate an internal precision of 1–2 % , while the A-stars agree with the Cohen IR fluxes to \sim 2 % .
Three solar analog candidate stars differ from the solar spectrum by up to 10 % in the region of heavy line blanketing from 3000–4000 Å and show differences in shape of \sim 5 % in the IR around 1.8 \mu m .