The atmospheric transmission's dependence on frequency is complicated. There are water vapor resonances above 100 GHz which one would want to avoid. There is frequency-dependent scattering from droplets, and there is a continuum background from water vapor dimers that rises with frequency. It is a very non-trivial situation. Having said that... the 2008 Olympics demonstration had a broadcast range of about 1 km, and people have demonstrated ranges up to several km at higher frequencies, even close to 300 GHz. So although it will never be as good as the range one can obtain at lower frequencies, there's a lot one can do.