You'll only get a clean cube curve with an MPPT controller. At the max power point (which occurs at the design TSR) the RPM and voltage go up with the wind speed and the torque and current with its square. (And because the heating of the alternator goes with the square of the current and thus the FOURTH power of the wind speed, it's very important to have good furling and/or a current-limiting function in the controller if you're going to do MPPT on a mill.)
Into a resistive load you'll get something like a square curve, because the current will be proportional to the voltage due to the ohms-law . But this will be skewed because the mismatched drag torque will make the TSR, and thus the angle of attack, and thus the blades' efficiency, vary with wind speed. Efficiency drops with higher wind speed and the function is not linear. It's driven by the aerodynamics of your blades and has pathologies as various regions of your blade surfaces stall.
In a battery charging system it gets even more complex, because the battery load holds the terminal voltage essentially constant (at the sum of the battery voltage and the diode drops), causing the induced voltage (and thus the RPM) to be proportional to a constant plus a linear function of the output current. You get no current until cutin, after which the current ramps up on a steep curve looks similar to that of the resistive load case but displaced by the cutin speed.
If you look at your graph it looks like it is a rough fit for a straight line that intersects zero around 5MPH (probably your cutin speed) and bends down near the end (as your blade stalls and/or the mill furls). Seems to me that's not unreasonable for a PMA/diodes/battery system. So it wouldn't surprise me if your actual wind speed/charge current curve is similar to such a fit to this comparison-of-metered-peaks.
