Yes, that was what I was talking about when I made that graph. It was just for demonstration, not a representation of anybody's particular machine.
Just to re-hash what I wrote in the other thread:
The wind power curve has one shape, the generator power curve has a different shape. They're never perfectly matched up but you can get them close enough. So I plotted the graph above to show that one of the effects of the mis-match is that TSR is low at low wind speed, and it can get high for strong wind speeds.
Even with fancy controls on the resistance load, you still want a good furling system, otherwise the blades will run away in a storm.
And at the bottom end, you can see the power coming from the wind dropping way off, but the load from the generator is much higher. That means the generator needs a few hundred watts to be turned at 50 RPM, but the wind power is only 10's of watts at 5 klicks. You can't lift a 10 ton truck with a 1 ton jack.
So all that confirms what you seem to get, based on your comment about switching in additional load. That works well to put an angle into the generator power curve. You can make the generator power low when the wind is low, and higher when the wind is high. If I was wiring up for resistance heating, I would have some beefy capacitors in line like Joe suggested, and tweak the furling to keep the top speed under control, then choose a low resistance so that the turbine isn't ever running fast. Not until confident that it behaves itself under all conditions, and that the caps aren't going to blow up, would I then consider changing resistors or adding relays to switch loads on or off. Just so that I could get a feel for how much is enough to add in, rather than making guesses, and accidentally letting it run away. Remember, sizing the blades can be a bit of guesswork, too.
For the water heater, I can think of a few things that could go wrong with welding on another port, and it could be better to just get another (smaller) water heater in series with the first. Two stages are much better than a single stage. Like I said above, trying a very conservative single-stage to get a grip on the turbine's normal behaviour before complicating the controls would be the way I'd go. Maybe you're comfortable with power circuits already, maybe not, it's just not safe to assume. I don't even trust myself with this stuff.
To decide when to switch additional load on/off, you have two methods. First, the voltage rises as the power rises, and second, so does the AC frequency. You can use either one to signal the switch, or both for redundancy if you want. A circuit to measure voltage can be tricked by spikes, and a circuit that measures frequency can be tricked by harmonics. There are ways to deal with each of these but I believe measuring voltage is easiest (Joe can correct me if I'm wrong, there). Then yes, just flip a relay when the threshold is crossed and suddenly 3 more resistors are on the circuit.