Traceability of CM-11 pyranometer calibrations at KNMI

KNMI has used Kipp&Zonen CM-11 pyranometers for operational radiation measurements since 1989. The pyranometers are calibrated regularly following WMO recommendations in the KNMI calibration laboratory against a KNMI reference CM-11 pyranometer. Operational calibrations show no degradation of sensors, and no drift in mean sensitivity, during the more than 25 years they have been in use. Variations of the calibration sensitivity are found to feature a standard deviation of 0.6%.

The KNMI reference pyranometers were calibrated in an unconditioned manner (i.e. neither ventilated, nor heated) on an irregular basis at PMOD-WRC in Davos, but roughly every two years. The sensitivities found in Davos with outdoor calibrations against a ventilated reference CM-22 showed sensitivities consistently lower than the nominal value from 2002 onwards, up to 2%. The uncertainty of these calibrations is typically 1.3%. A bias error thus seemed to be present in these calibrations. Two KNMI reference pyranometers were also calibrated at the calibration site of the Deutsche Wetter Dienst. These calibrations showed a sensitivity of the reference pyranometers close to the nominal value. The bias in the sensitivity during the calibration at PMOD-WRC is attributed to zero-offset A, which can be reduced by ventilating the KNMI reference pyranometers during calibration at PMOD-WRC, to be compliant with the specified uncertainty.

Also the laboratory calibration procedure at KNMI has been investigated by Kipp&Zonen, and apart from minor issues, is considered to be accurate. Differences up to 0.5% has been identified, which are well within the uncertainty of the calibration procedure. An adaptation of the waiting time for the measurement of the dark current from 2 minutes to 1 minute is advised.

Other advises following from this study are that it is not necessary to correct operational global radiation measurements of the KNMI network, as the bias error had not been forwarded from the KNMI reference pyranometers to the KNMI network sensors. Furthermore, calibrations using method (b) from the CIMO guide cancel zero-offset A, and are thus preferred over method (a), which are currently used at PMOD-WRC. It should be investigated if such measurements can be done routinely for the calibration of KNMI reference pyranometers.