Why do all three models include a multiplication factor of 400, and how does it affect the data?
All three models convert DN values to reflectance, which is a ratio of calculated radiance at the sensor / total irradiance at that wavelength and has values ranging between 0.0 and 1.0. Multiplying by 400 stretches this range of values across the potential range of 1-255 available in 8-bit data. The stretched range of values are rounded and output as unsigned 8-bit data in order to create images with smaller file sizes.
