There are various ways of measuring ground motion. Intensity, used in the discussion so far, is a simple classification of strength of ground motion based on observed effects (like the Beaufort wind scale). Engineers usually prefer something more physically quantitative. For many years peak ground acceleration (pga) has been used as the favourite parameter for earthquake ground motion (even though it doesn't correlate particularly well with actual damage levels). Since they need know what design values to take for other frequencies of ground motion (taking pga as 40 Hz), the usual practice in the past has been to create a design spectrum from a "typical" earthquake as the guide to what values to take for other frequencies relative to the 40 Hz hazard value (which is the one that falls out of the hazard calculations). The design spectrum is thus shifted up or down according whether the hazard in pga terms is higher or lower.
This doesn't take into account the individual character of the hazard at the site - is it dominated by small, close earthquakes or large, further away ones? At some frequency values the hazard will be over-estimated, at others it will be underestimated. An alternative option is that the hazard should be calculated separately for a range of frequencies, using the same source zone model and a regionally-valid spectral attenuation model. From this can be built up a spectrum that reflects the real levels of hazard at the site at all frequencies. This is known as a Uniform Risk Spectrum (or Uniform Hazard Spectrum, usage varies).