wide-view channel detector. Estimated
exitance variation as a function of angle at the exit port was less than 0.1%.The ERBE design has been tested and characterized over many years; the major issues of determining and maintaining a calibration have been addressed. Many aspects are documented in the literature, allowing a relatively detailed examination of ERBE. Thus, we shall discuss the ERBE instruments, calibration and perform-ance as a solid basis for CERES.
The equivalence of the electrical power needed to produce the same temperature response as an unknown amount of incident radiant power makes these radiometers (to zero order) essentially self-calibrating. Indeed, active cavity radiometers have been used as transfer standards (Hesser and Potter 1983). However, some items need still to be addressed to achieve the desired accuracy (< 1%) for climate studies. For ERBE, the ground-based calibrations were performed in a special facility at TRW, described in detail by Falbel and Iannarelli (1981). The calibration chamber was a cylindrical vacuum tank about 2 m in diameter and about 2.5-m long. The radiometers were mounted in the center on a carousel which could be indexed to a set of locations. At one end of the tank, a 20"-diameter integrating sphere was used to provide short-wavelength calibration. At the other end of the tank was a Master Reference Black Body (MRBB) source, which was used to calibrate the long-wavelength channels. In addition, a Space Reference Source was used to radiometrically simulate cold space, since the scanner references all signals to cold space in every scan.
The MRBB was a grooved, black anodized plate, whose temperature was controlled to ~0.015C, within a temperature-stabilized structure. A gold-plated, reflective cone is used to couple the radiation into the wide-field-of-view (WFOV) radiometer: this is a compromise needed only for the wide-field channel. The MRBB proto-type spectral emittance was measured from 2 to 20 mm; the calibration is traceable to the International Precision Temperature Scale (IPTS).
To enhance spatial uniformity, the integrating sphere was as large as could be accommodated by the TRW tank. Four lamps (with diffusers) provided illumination,
which was detected and controlled by silicon photocells. The exit aperture was 5" in diameter to accommodate the 132 wide-view channel detector. Estimated
exitance variation as a function of angle at the exit port was less than 0.1%.
An albedo simulator plate was used to simulate radiation over the same solid angle the detector would cover in space. The plate was painted white, maintained at 260K, and illuminated by four 250-W quartz halogen lamps.
The space reference source was similar in construction to the MRBB, but was maintained at liquid nitrogen temperature.
The virtue of this ground-based calibration system was that it provided a close approach to the actual measurement environment in space. As noted by Crommelynck (1977), the source spatial and spectral distribution, the optical system geometrical characteristics, detector spectral characteristics, transfer functions of the electronics, and digitizers all contribute to the real calibration of an instrument. Thus, one should either measure all the relevant spatial, spectral and electronic transfer functions, or one can make the calibration facility as close as practical to the actual measurement geometry.
The on-board calibration system consists of two sources. Concentrically grooved blackbody assemblies that can be heated above instrument ambient conditions serve as the reference for the longwave and total channels. Two platinum resistance thermometers (PRTs) are imbedded in these plates for accurate temperature measurements. A tungsten lamp/optical projection system is used for the shortwave channel; three intensity levels are selectable. Finally, a solar calibration can be performed for the shortwave and total channels. The solar flux is attenuated by a "mirror attenuator mosaic" (MAM) assembly; the attenuation factor is 5.0.
The three ERBE telescopes are aligned with better than 98% field-of-view overlap and rotate through a complete scan cycle every 4 seconds, first observing space, then scanning across the earth in about 2.3 seconds, then viewing internal calibration sources (Barkstrom 1989).
On the non-scanner ERBE, the heater voltage is digitized to 13 bits, with the measurement range for all but the solar monitor being less than 40% of full scale for the "total" channels and less than 25% for the short-wave channels (Luther et al. 1986). Thus, the effective digitization for these latter channels is typically at 10-12 bits.
On the scanner instrument, the digitization is at 12 bits, and the on-board calibration systems provide coverage up through the range of actual earth measurements (Kopia 1986).
A careful analysis of the heat flows in the solar monitor is given by Lee et al. (1987), together with tables illustrating the uncertainties in the determinations of the detector parameters. The conclusion drawn in that paper is that "The measurement precision is dictated primarily by the uncertainty in the electrical heater power caused by electronic noise and the resolution of the 13-bit analog-to-digital converter." The stated
Band Uncertainty (Wm-2 sr-1) ------------------------ ---------------------- Shortwave (0.2 - 5.0m)
0.3 and +0.75 Longwave (5.0 - 50.0
precisions and accuracies are estimated to be better than +0.1% and +0.2%, respectively.
Kopia (1986) states the ERBE scanner precision (3s) and systematic error bounds, respectively, as
