Since the IRAS mission was primarily designed to measure point sources and DIRBE was primarily designed to measure extended emission, we believe the COBE/DIRBE calibration is valid and the DIRBE results provide a check on the large scale performance of IRAS. The COBE/DIRBE Explanatory Supplement (version 19 July 1993) presents a preliminary linear transformation between IRAS and DIRBE data and is presented in the table below. This linear transformation was derived based on carefully selected DIRBE data compared to IRAS scan data. The transformations are applicable to small scales on the order of an IRAS scan but there is no certainty as to their applicability at smaller spatial scales down to 0.7 degree. The IRAS point source calibration is not affected by these numbers.
The offset term is relevant to the total intensity IRAS product, the IRAS Zodiacal History File (ZOHF). It is not applicable to the IRAS Sky Survey Atlas (ISSA) images since the zodiacal emission was subtracted. The gain term is applicable to both the ISSA and the ZOHF and shows the IRAS measurements at 60 and 100~um were too bright relative to DIRBE at large spatial scales. There is no simple prescription for unraveling the varying spatial response of the detectors to effect a more accurate calibration for the ISSA images. The data given in the table below, serve, however, to bound the problem to being no worse than a 28% effect at 100 um, 13% at 60 um, and 12% in a 60/100 um color.
Wavelength Gain Offset (MJy/sr) 12 um 1.06 +/- 0.02 -0.48 +/- 0.43 25 um 1.01 +/- 0.02 -1.32 +/- 0.74 60 um 0.87 +/- 0.05 +0.13 +/- 0.65 100 um 0.72 +/- 0.07 -1.47 +/- 0.88[*] in the sense I(DIRBE)=Gain x I(IRAS) + Offset.