It all depends on who's listening...

The main reason for making underwater acoustic assessments is to protect the wildlife in our oceans. Whether your motivation for this comes from the the EU habitats directive, your local national rules or and interest in protecting the environment does not change the way you should evaluate the noise pollution. 

Different habitats in the sea have different species living in them, and different species hear differently. This seemingly simple information can make a big difference when assessing our impact on the environment.
For this post I'll use the harbour porpoise (P. phocoena) and the harbour seal (P. vitulina) as example species, to show how much of an impact my own personal (and virtual) submarine has. 
The seal has relatively better hearing at lower frequencies (<64 kHz), while the porpoise excels at higher frequencies. The red line in the above plot is the difference in dB between the hearing thresholds of the two species (really it's a ratio because of the log-scale). 
Figure 1: The audiograms of a porpoise and a seal, and the difference (ratio) between the audiogram them expressed in dB. 
dB values on primary (blue) vertical axis are RMS values relative to 1┬ÁPa.
(Data from Nedwell et al. 2004 "Fish and Marine Mammal Audiograms: A Summary of Available Information.

So I modelled a (noisy!) submarine (with the prop noise from a bulk cargo ship) sailing through the Irish Sea. I doubt any submarine would make this much noise (they don't), but submarine sonograms are hard to come by anyhow.

Detection range for a harbour seal.
A harbour seal would be able to hear my submarine no matter where in the Irish Sea it was. The harbour seal's relatively low frequency hearing is the cause of this. Bear in mind that this is a model, and no other noise is present to mask the sound coming from my virtual submarine. Noise from wave motion, wind and traffic would lower the detection range considerably.
Detection range for a harbour porpoise.
The harbour porpoise on the other hand might not even detect it going past. From looking at the audiograms (Figure 1), the porpoise looks to have the most sensitive hearing (it has the lowest thresholds), but because it is specialised for high frequency hearing and has relatively bad hearing at low frequencies, with regards to the sound in this example, it is practically deaf.

With this post I wanted to show how important it can be to have knowledge about the animals, for the sake of protecting them, but also for the sake of not over-implementing protective measures where they are indeed not needed.