while there's nothing inherently wrong with using a water pump as a dummy load, I have to say that it'll be hard to get any meaningful information from such a test.
. . . not that you couldn't get useful information, just that it would likely be more work and more expense than its worth. (I pay the bills by building/repairing instrumentation packages for physics experiments. . . . )
after all, the goal of all the testing is to see (and tweak) how well the turbine matches the gen.
using relatively straightforward math (nothing that excel or a pocket calculator can't handle-- do a search. . .) you can make a fair guess at the maximum power your turbine could supply just based on its swept area and some SWAGs at wind-speed (the hosts of this forum sell an anemometer kit); compare that to what the gen makes on the 'bench.' Unless the two numbers are radically different, there's nothing to worry about. . . and if the gen works out to much higher power than the turbine can supply, you can save a certain amount of frustration and work by knowing that before you sink lots of money and man-hours into a undersized turbine. ditto the contrapositive case- you can save heartache by knowing that the gen is too small for your proposed turbine before you put it up and end up with stir-fry. <G>
I feel it's better to rig a prony brake (if there's lots of uncertainty) or, once you're convinced that the match is reasonable, (+/-25%? use a number that makes you happy) just go ahead and risk the electronics by using your gen as the test load.
you can minimize your risk profile by using a resistor pack in place of a battery bank and charge controller, but you won't get the same data as you would with the full system connected. On the other hand, glow coils or wirewound resistors will take a frightful pounding and come up smiling. . . however, again, you'll need some basic idea of how much power/volts/current you're likely to see in order to build up a 'realistic' resistor pack.
if you're getting the feeling that this 'testing' business is often an iterative process, you've been paying attention. Welcome to the engineering profession!
For some ideas do a search for 'load tester.' Jerry built one recently. (a little overkill for testing a wind genny . . .) if you size the resistor bank to the genny specs, you can use it for a dump load later, and you won't be out any extra money.
just my 20 millidollars.
#insert<std_disclaimer.h>
your mileage may vary.
-Dan