A New Look at Calibration and Use of Eppley Precision Infrared Radiometers. Part I: Theory and Application

Abstract
The calibration and accuracy of the Eppley precision infrared radiometer (PIR) is examined both theoretically and experimentally. A rederivation of the fundamental energy balance of the PIR indicates that the calibration equation in common use in the geophysical community today contains an erroneous factor of the emissivity of the thermopile. If a realistic value (0.98) for the emissivity is used, then this leads to errors in the total flux of 5–10 W m−2. The basic precision of the instrument is found to be about 1.5% of the total IR irradiance when the thermopile voltage and both dome and case temperatures are measured. If the manufacturer’s optional battery-compensated output is used exclusively, then the uncertainties increase to about 5% of the total (20 W m−2). It is suggested that a modern radiative transfer model combined with radiosonde profiles can be used as a secondary standard to improve the absolute accuracy of PIR data from field programs. Downwelling IR fluxes calculated using the Rapid Radiative Transfer Model (RRTM), from 55 radiosondes ascents in cloud-free conditions during the Tropical Oceans Global Atmosphere Coupled Ocean–Atmosphere Response Experiment field program, gave mean agreement within 2 W m−2 of those measured with a shipborne PIR. PIR data from two sets of instrument intercomparisons were used to demonstrate ways of detecting inconsistencies in thermopile-sensitivity coefficients and dome-heating correction coefficients. These comparisons indicated that pairs of PIRs are easily corrected to yield mean differences of 1 W m−2 and rms differences of 2 W m−2. Data from a previous field program over the ocean indicate that pairs of PIRs can be used to deduce the true surface skin temperature to an accuracy of a few tenths of a kelvin.