UV_A and UV_B sensors
Mr. Wallasch used his known {S(lambda)} solar spectral density values and applied to them the relative spectral responses {E(lambda)} of our Delta_T sensors.
A calibration factor can be found by comparing the integral of the {S(lambda)*E(lambda)} values computed by Mr.Wallasch to the raw readings of the Delta_T sensors. This method assumes that the same weather condition exist at the same time in Diekirch and in Offenbach (same latitude; difference of altitudes between Diekirch (218m) and Offenbach (100m) considered negligeable). From the available data (1. July to 14 September 1994) we retained only those corresponding to cloudless days; cloudless means that the meteorological station of Luxembourgs airport Findel reported no clouds during a whole day; this left 767 remaining cases.
UV_A sensor:
The plot of the computed UV_A data versus the "raw" readings gives a linear fit with a slope of 2.69; hence the UV_A readings stored in the data files should be multiplied by 2.69 to give the "true" irradiance in [W/m2]. As the UV_A sensor does not extend markedly into the UV_B region, no windowing will be applied for a first approximation, so the definitive calibration multiplier is 2.69.
UV_B sensor:
The plot gives a linear fit of 0.94, close to 1. This means that the readings of the UV_B sensor correspond approximately to the "true" power received; as this sensor extends largely into the UV_A domain, a windowing function must be applied, which gives a multiplier of 0.27.
If we neglect for the moment the difference between 0.94 and 1, the definitive multiplier 0.27 will yield the best approximation to convert the stored UV_B readings into "true" UV_B irradiance in [W/m2].
It should be noted that the UV_B sensor has a slight zero offset of ca. +0.13 ; this value should be substracted from the readings before multiplying by 0.27!