Well, I'm not a whole lot further ahead than you are, Stuart, but I'll be glad to share my limited knowledge.
When a bass note is struck it will decay over time until it blends with the noise floor of the listening room. This gradual delay is called ringing or resonance. The faster the sound decays, i.e. the lower the ringing, the tighter or more well-defined the bass will sound. A long decay results in bloated, poorly-defined bass.
The waterfall measures the rate of decay in the listening room over time. The vertical axis extends from the noise floor of the listening room, typically around 40dB, to a value high enough to contain the peaks in the response measurement, normally around 90dB. The horizontal axis shows the frequency response, normally in the range of 40Hz to the upper range of the bass that concerns you, say 100Hz-200Hz. The third axis represents time, in mille-seconds.
Edit: Based on feedback from Jason, we now agree to sisplay 15-300Hz on the horizontal scale, and to show a 60dB range on the vertical scale (which requires a measurement at a higher level than 75dB).
So, looking at Michael's waterfall, we see he has selected 65-95dB on the vertical axis (which really should have a lower value of 40dB), and he is measuring from 20Hz-170Hz. On the time axis, he has measured 0-300ms.
My understanding is that the objective is to have all bass frequencies decay to below 40dB within the first 300-600ms. Unfortunately, since Michael's waterfall only goes down to 65dB, he is not showing the noise floor of 40dB. I retract my earlier statement that the waterfall looks good--we actually don't have enough data.
Let's look at some examples (note that my time scale is 0-600ms--we'll come back to that in a moment):
First, a vertical axis of 65-95dB;
And finally, 30-95dB:
The first waterfall shows that bass frequencies have fully decayed to 65dB within 100-150ms. This waterfall is meaningless, since we want to show when the decay reaches 40dB, i.e. the noise floor.
The second waterfall is useful, because it shows that most, but not all, bass frequencies have decayed below 40dB by 600ms. The exceptions are at 35Hz, 60Hz and 65Hz, all of which seem to be relatively minor.
The third waterfall is not so useful, because it shows decay to 30dB, which is 10dB lower than the noise floor of the room.
Now let's see what varying the time axis does. The graphs above had a 600ms time axis. Let's go down to 300ms:
Big difference! This shows that within 300ms, ALL of the bass frequencies are still present above 40dB. This doesn't give us much information on where to focus our improvements. Let's try 450ms now:
This is showing better information, but there are still quite a few frequencies not fully decayed by 450ms. To arrive at a meaningful waterfall, we need to continue to increase the time delay until most, or all, of the frequencies are fully decayed. If this occurs at a time delay of 600ms or lower, we are in reasonable shape. If it exceeds 600ms, then those frequencies that are not fully decayed at 600ms should be the focus of our improvement effforts.
So, what are the take-aways?
1. Always set the lower limit on the vertial axis equal to the noise floor of your listening room (use 40dB if you don't know).
2. Set the horizontal axis to the frequency range you want to examine, 20Hz-200Hz in my examples. (Note: many experts say to ignore the frequencies below 40Hz, because there are very few treatments that can fix resonances in this range.) Edit: As mentioned above, we have agreed on a 15-300Hz range for the horizontal axis.
3. Adjust the time axis, starting at 300ms, until you have a meaningful indicator of where the resonance issues are.
Of course, the interesting discussion will be, what value of the time axis represents a "good" decay time, and what values show issues that need to be resolved. In other words, if everything has decayed in a shorter time than 600ms, is that "good enough"? A second interesting discussion will be, what types of room treatments address issues at specific frequencies, e.g. what to use for a 40Hz resonance vs. a 90 Hz resonance--the answer is likely to be different.
Tip: How do I measure the noise floor of my listening room? Answer: In REW, after you have performed the microphone calibration (Step 5 in the Guide), click the red button in the SPL Meter tool. The meter will display the ambient noise level in the room. Assuming you have gone to the effort to make sure it is as quiet as possible, the meter will display your room's noise floor.
Edited by AustinJerry - 1/21/13 at 3:58pm