Grove Turbidity Sensor Voltage to NTU Formula

Hi I have been unable to see from the documentation of the Grove Turbidity Senstor v1.0 what the formula should used to convert from a voltage value on the Analogue mode to a NTU Value.

The sensor data sheet of the TSW-10 Turbidity Sensor provides has a chart but no guide to calibrate the device.

Reverse engineering the curves from the chart aI was able to work our a formula for the Min Max tolerances of the two curves provided and I was able to extropolate a median curve;
y = NTU Value
x = sensor pin voltage

// Low Formula y = -154.66x^3 + 1745.1x^2 -7020.3x + 9906.8 (y = NTU x = Voltage) from Excel Calcs
// High Formula y = -154.33x^3 + 1973.8x^2 -9075.8x + 15093 (y = NTU x = Voltage) from Excel Calcs

Extrapolated
// Ave Formula y = -148.66x^3 + 1788.1x^2 -7768.8x+ 12097 (y = NTU x = Voltage) from Excel Calcs

However the voltage measured from the Grove board sensor pin using the same distilled water sample (NTU = 0) seems be erratic being between 4.23v and 4.18v. The variation appears to be corrolated to the ambient light e.g. shine a light on the sample and it distorts the sensor increasing the Voltage of the signal upto 4.4v.

Does anybody know what the ambient light this sensor was calibrated against ?
Should the sample be shrouded in complete darkness to be accurate ?