Figure 4. Estimated values for surface moisture 
£p are in the range 15-35 millibars, where lower 
values are darker, higher values are brighter 
Although derived values for optical depth 
generally agree with radiative transfer 
calculations (Price, 1984), notably poor values 
exist in some locations. These problem areas are 
mostly far off nadir. 
ERROR SOURCES 
Many approximations have been used in the above 
analysis. We simply list the error sources, 
without attempting to quantify them in this short 
paper: 
1. Clouds produce variability in the 
thermal infrared data that is not associated with 
surface phenomena. This effect is relatively 
severe, as shown in areas of black/white in 
images 1-3. More sophisticated cloud filtering 
is required. 
2. Surface emissivity variations have been 
neglected. There are two effects: a) 
simultaneous variability of emissivity of 
features sensed in channels 4 and 5 increases the 
range of deduced temperatures in both channels. 
This is a second order effect, b) variability 
in the response of one spectral channel, 
independent of the other, has the same effect as 
a change in atmospheric transmittance r. This 
effect should be included in the treatment of St. 
3. The linear analysis of atmospheric 
effects is approximate, and produces increasing 
errors as the optical depth increases. In fact 
the Kneizys et al. expression for k^ leads to the 
value 1.45 for R, while actual computation for 
the 5T for this data set leads to an effective 
value of R of 1.35 (Price, 1984). This 
illustrates the more exact treatment required for 
values of r which are not much smaller than 1. 
4. The crude expressions assumed for e(p) 
and T(p) do not compare closely with actual 
sounding data. In particular moisture does not 
generally fall to zero above some relatively low 
level of the atmosphere. However, these or 
equivalent expressions are needed in order to 
derive useful information from the derived values 
of t and T , . 
Figure 2. The area in Fig. 1 has been processed 
in 40x40 blocks to produce values of optical 
depth that range from .25 (darker tones) to .45 
(lighter tones). 
Figure 3. The area illustrated in fig. 1 has 
been processed in 40x40 blocks to produce this 
image of 6p/p, with the majority of the image in 
the range of values .2-.4, i.e. moisture falls to 
zero in the range 800-600 mb in areas with 
surface pressure of 1000 mb. Higher values of 
5p/p are brighter.