An extensive variable-weather comparison of pyrheliometers
Authors
Joseph J. Michalsky — NOAA/Earth System Research Laboratory/Global Monitoring Division & CU-Boulder/Cooperative Institute for Research in Environmental Sciences
Tom Stoffel — National Renewable Energy Laboratory
Ellsworth G. Dutton — NOAA/OAR/ESRL
Ibrahim Reda — National Renewable Energy Laboratory
Peter Gotseff — National Renewable Energy Laboratory
Stephen M. Wilcox — National Renewable Energy Laboratory
Daryl Myers — National Renewable Energy Laboratory
Afshin Michael Andreas — National Renewable Energy Laboratory
Donald Nelson — National Oceanic and Atmospheric Administration
Description
Thirty-three commercially available pyrheliometers were compared over more than nine months between November 2008 and September 2009 at the NREL Solar Radiation Research Laboratory near Golden, Colorado. Included among the 33 instruments were four all-weather absolute cavity radiometers, which proved to be as stable as the open cavities that were used to calibrate all pyrheliometers at several points during the study. The average of the four all-weather cavities was chosen as the standard irradiance to which the other 29 pyrheliometers were compared. The 95% confidence in the cavity standard was ± 1.2 W/m2. In addition to the four all-weather cavities, there were seven sets of three pyrheliometers of the same make and model plus an additional eight prototypes in the study. These test instruments include those most widely used by the international community along with these new production models. Instruments were cleaned every work day. Analysis was performed by a non-participant in the experiment who had no knowledge of the identification of the instruments except for the cavity radiometers; the analyst also knew which three pyrheliometers formed a set. If the manufacturer provided temperature corrections, they were applied. The early analysis suggests four groupings of pyrheliometers; windowed cavity radiometers are the most accurate; followed by pryheliometers that have 95% uncertainties around ± 5 W/m2; followed by pyrheliometers with uncertainties around ± 10 - 15 W/m2; and then two prototypes that were clear outliers. The results will be illustrated; the identity of the instruments is pending completion of the analysis. The comparison was organized under the auspices of the Baseline Surface Radiation Network with instruments donated from many sources. NREL staff members were responsible for most of the observational activity.
