Wind speed does decrease nearer to the surface, water in this case, but not as much as most people imagine - as is demonstrated by this graph.
This is correct for most "good" wind destinations with laminar flow.
At lots of turbulent flow inland areas, specifically smaller lakes ringed with trees, the wind speed can be highest at the waters surface. My home lake is like that. Sometimes you get a 20knot gust only to send the kite up into 10knot winds on a jump attempt. Very frustrating for doing anything aerial. When I travel to good kite destinations, I kind of have to learn to kite all over again as I no longer need to devote brain power to "feeling" high vs low windspeed differential. At those locations (ocean, larger inland lakes), things work very much as iriejohn illustrated.
I saw kite schools in several locations (e.g Terifa in Spain) using 12m2 kites to teach students on short lines in big winds. Sometimes they just halfed the lines and sometimes they even divided it by 4, so had 12m2 kite on 6m long lines when people riding on the water had 8m2 kites.
The reason kite schools use short lines is so beginners don't get worked so badly. On short lines, when the student loses control of the kite, it crashes sooner. It has nothing to do with the power of the kite when it's being flown under control. As others have said, you can't turn a 12 m into a 9 m by shortening the lines.
I would add just one point : shorter line is more comfortable in gusty winds specially with orientation changes. Indeed the kite needs shorter time to get at its new position in the window. This is secially noticable on a foilkite which has less instant depower until the kite reaches the end of wind window. By tbe time the foilkite reaches the end of window it pulls like a truck and accelerate when you sheet-out. So no real solution except be more patient with longer lines ...