Scubajack,
I've written a Feature piece about channel mixing techniques, but it's still in the queue. In many cases, you can use the channel mixer for this, which is easier to use. But Apply Image is more powerful, in that you can select different blending modes. Lighten is often the best for situations like this shark.
One way to dump the red channel (the one I used) is to click on that channel on the channel palette, Select All, then Clear. The channel should go completely blank.
To bring in a copy of the green channel, keep the red channel selected on the palette menu, but click on the left column next to RGB. This way, you can see the compiled image, but Apply Image (and any other operations) will only affect the red channel.
When you start Apply Image, the red channel should already be identified as the target of the operation. Just switch the source to the green channel. Now you can play with different opacity levels and blending modes, while viewing the results on the screen. When you get what you like, click OK. That's it.
L*a*b is an alternative color space, with a much larger gamut than RGB. It's also used as the base color space for a lot of Photoshop operations. If you convert your color space from RGB to CMYK, it actually goes through L*a*b on the way.
L*a*b may seem difficult, but it's well worth learning about. I talk a bit about it in the feature piece. If you take an image, and go to Image, Mode, you'll see Lab as one of the options. Click on it, and now the color channels of your image will be the Lab channels: L (luminosity), "a" and "b". The L channel contains all the luminosity data - is essentially a black and white image. The "a" and "b" channels have all the color information - "a" values represent a contimuum from red to green, while "b" values represent blue to yellow.
You probably already know about curves, which is an alternative to Levels, but more powerful. It does everything Levels does, but much more. Curves can be used in RGB space, but are particularly powerful in L*a*b space.
If you modify the L curve, you can change the contrast, brightness and/or darkness levels of the image without affecting color at all, which can be difficult using RGB channels. Steepening the L curve increases contrast, and moving the ends in the opposite direction does the opposite.
Modifying the "a" and "b" channels is often the best way to correct a color cast, and can also be used to modify saturation levels selectively at different parts of the curve. Steepening the "a" curve will increase saturation of both reds and greens, moving both ends in the opposite direction will desaturate reds and greens. Shifting all or part of the "a" curve up will increase red values (and remove a green cast), shifting it down does the opposite. Similar for the B curve with yellows and blues.
Make the Curves window as big as possible, since a small change in these curves can make a big difference in the image. I usually find it easier to make color and contrast corrections in L*a*b color space.
Hope this helped.