The Bane of the Digital Underwater Photographer
Your biggest frustration if you go digital will be a phenomenon called "shutter lag." Simply put, it's the delay between what you tried to shoot when you pressed the button and what the flash actually captured. A few tenths of a second can ruin the composition.
There's no real engineering reason why the shutter lag is so bad in today's digital still cameras. It's mostly a question of inadequate engineering, probably because the designers don't recognize the importance of the problem. It's exacerbated by attempts at cost-cutting.
Shutter lag varies tremendously, depending on the detailed state of the camera, the control settings, and the shooting conditions. So it's difficult to simplify it to a single number. Also, since each camera has different dependencies, it's hard to even provide a direct comparison. Instead, I'll adopt a narrative form for each camera.
The C3030
There's one big mistake you can make with the C3030 that will guarantee excessive lag, and that's having the flash in red-eye mode. Since you can't see the top LCD panel in the Tetra, you can't tell if the camera is in this mode except by pressing buttons. A simple way to check is to press the menu button twice. The flash mode icon (you want the flash "off" while shooting underwater) is displayed near the upper left corner of the color display. Just be sure it's the lightning bolt with the circle around it, not the eye-shaped symbol, or you'll get a couple seconds of delay anytime the exposure meter indicates under-exposure. This happens even when the internal flash is disabled!
There are two ways you can shoot the C3030 - raw shutter depression and half-depression. When you half depress the shutter button (it's worth practicing this), the camera locks its focus and exposure settings, and puts the LCD viewfinder in a special mode. The green LED next to the optical viewfinder glows when you do this, but unfortunately you can't see that in the Tetra housing. If you move the camera, you'll notice that the screen update has gotten jerky.
The smallest delays occur while in this "half-depressed" mode. Actual* delays from the shutter button to the flash are about 100 milliseconds (msec). This may seem small, but it is noticeable. It's also in line with what's been reported elsewhere for the C3030.
There's another effect in the C3030 that will almost certainly cause you grief. The 100 msec delay I've quoted above is from the depression of the shutter to the flash. However, usually you'll be using the color LCD as your viewfinder, and the image displayed is lagged from reality. So if you press the shutter at the instant the fish appear perfect in the viewfinder, you have the 100 msec actual delay plus the viewfinder delay.
The viewfinder delay in the C3030 depends on the f-stop you've selected for the shot. At f/2.8, it adds about 150 msec, and at f/11 it adds 250 msec. So the total delay is 250 to 350 msec. You can minimize this while shooting critical subjects by viewing the subject directly (not using the LCD viewfinder) under critical conditions. You'll find this remarkably difficult in practice.
Because of the difficulty of half-pressing the shutter, you'll find more often than not that you're simply using raw shutter depression for most of your shots. If you're using the camera in manual exposure mode, continuous autofocus, and using the LCD as viewfinder, you'll get about 1100 msec of delay, and you'll find that very annoying. It's composed of about 100 msec of viewfinder delay (since you're not half-depressing, it's not f-stop dependent), 350 msec (minimum) of autofocus delay, and 650 msec of set-up.
If you use manual focus, you can eliminate the 350 msec autofocus delay to bring the total down to about 750 msec. It's still enough to be very annoying.
If you're shooting from the hip using autofocus and raw shutter depression, things can get even worse. The autofocus can take another 100 to 200 msec to find your subject. And if you're in macro mode, the seek is longer still and can take up to 500 msec. These are worst case times, and in practice seldom occur.
Summary for the C3030 - Using half-depression of the shutter, you'll get about 300 msec of lag. Raw shutter depression will give you about a second. If you accidentally get into red-eye mode, you'll see 2.5 seconds.
The CP990
My discussion of shutter lag in the CP990 is rather limited because the loaner I received from Nikon didn't allow the shutter to half depress. So I'm forced to guess a bit on the CP990 behavior based on my CP950 and published reviews. I hope to update this if I can get my hands on a working CP990. For those readers interested in the CP990, I ask you read the above on the C3030 since I won't be duplicating explanations here.
The CP990 seems to actually understand when you turn off the internal flash. I haven't run into the "red-eye" lag problems (there's probably a way to do it, but I haven't stumbled into it).
The CP990 also clearly changes the LCD viewfinder display when the shutter is half-depressed. This is very nice. Add to this the Nikon's "focus confirmation" feature, and you'll feel you have a good handle on what the camera is thinking.
In "half-depressed" mode, actual delays from the shutter button to the flash are about 100 milliseconds (msec), similar to the C3030. The main difference is that you'll find this mode much easier to use with the Ikelite lever shutter and the superior indications in the CP990 LCD screen. The CP990 adds about 100 msec delay due to display lag, but unlike the C3030, this lag is independent of the f-stop. So overall, the CP990 is 50 to 150 msec better in lag than the C3030 in the half-depressed mode.
If you're using raw shutter depression, you'll get a fairly consistent 900 msec total delay from the CP990. This doesn't improve when in manual focus (or alternately, once the autofocus finds lock, it causes no delay). It does get worse if you're shooting from the hip and autofocus has to seek, and macro mode exacerbates this.
Summary for the CP990 - Using half-depression of the shutter, you'll get about 200 msec of lag. Raw shutter depression will give you a little less than a second. Half-depressed shutter isn't too hard to do, and focus confirmation lets you anticipate how long the excess autofocus lag will be.
*Measurement Technique - I've tried to make my measurement technique for determining shutter lag reflect actual usage conditions. I shoot an analog stopwatch, and try to capture the moment when the sweep hand hits the "5" at the bottom of the dial. I calibrate using my Fuji S-1, which I deem "perfect," and find a reflex delay of about 200 msec with a variance of about 50 msec. I subtract this 200 msec from my other measurements.