Radiometer Uncertainty Sources
Combined uncertainties depend on the product of the sensitivity functions (partial derivatives of the response with respect to the measurement equation variables) and error source magnitudes, et. The largest contributions to shade-unshade uncertainty are from the evg and evd which must include estimates of the thermal offset (10-70 jiV) described below, and data acquisition measurement uncertainty (typically <10 |iV). For example, for an all-black sensor pyranometer with a responsiv- ity of 7.0 mV per 1000Wm-2, a 70-pV offset corresponds to an irradiance error of -10Wm-2.
For the shade-unshade technique, with fixed direct beam error eB « 4.0Wm-2, zenith angle error ez « 0.06°, and diffuse irradiance error eD « 2.0Wm-2 (black - and-white sensor), the uncertainty in RS for a pyranometer is a function of: zenith angle, uncertainty in the pyranometer voltages, and magnitude of the beam; it ranges from about 1.0% at small zenith angles to >10% for zenith angles greater than 85°. The component-summation technique normally has lower total uncertainties, since the uncertainty in the direct beam is essentially the same as in the shade - unshade calibration, the uncertainty contribution from the diffuse measurement is rather low, and only one voltage measurement is involved for the instruments under test, as opposed to two voltage measurements in the shade-unshade technique (Myers et al. 2004).