The accuracy of Arctic sea ice surface temperatures T-s derived from a
dvanced very high resolution radiometer (AVHRR) thermal channels is ev
aluated in the cold seasons by comparing them with surface air tempera
tures T-air from drifting buoys and ice stations. We use three differe
nt estimates of satellite surface temperatures, a direct estimate from
AVHRR channel 4 with only correction for the snow surface emissivity
but not for the atmosphere, a single-channel regression of T-s with T-
air, and Key and Haefliger's (1992) polar multichannel algorithm. We f
ind no measurable bias in any of these estimates and few differences i
n their statistics: The similar performance of all three methods indic
ates that an atmospheric water vapor correction is not important for t
he dry winter atmosphere in the central Arctic, given the other source
s of error that remain in both the satellite and the comparison data.
The errors are not reduced by regression with both thermal channels an
d the satellite scan angle. A record of drifting station data shows wi
nter air temperature to be 1.4 degrees C warmer than the snow surface
temperature. ''Correcting'' air temperatures to skin temperature by su
btracting this amount implies that satellite T-s estimates are biased
warm with respect to skin temperature by about this amount. A case stu
dy with low-flying aircraft data suggests that ice crystal precipitati
on can cause satellite estimates of T-s to be several degrees warmer t
han radiometric measurements taken close to the surface, presumably be
low the ice crystal precipitation layer. An analysis in which errors a
re assumed to exist in all measurements, not just the satellite measur
ements, gives a standard deviation in the satellite estimates of 0.9 d
egrees C, about half the standard deviation of 1.7 degrees C estimated
by assigning all the variation between T-s and T-air to errors in T-s
.