ERBE-ISCCP vs ISCCP
Based on the results of the previous section, it seemed likely that a
simple coefficient was inadequate. So we look at the distribution of
ERBE-ISCCP as a function of ISCCP.
This normalized distribution
of ERBE-ISCCP vs ISCCP
shows a fairly linear dependence from 0 to 190 W/m2 (actually there
is a gentle curve to it), and then it levels off at -18 W/m2 or
so. The width varies with the size of the deflection, maxing
out at 80% within +/- 10 W/m2.
ISCCP2
We can now correct ISCCP with the median of that distribution to get ISCCP2.
normalized distribution
of ERBE-ISCCP2 vs ISCCP2.
Mean ERBE-ISCCP2
The mean difference is mostly near zero away from land. It is clearly
better than ISCCP without the correction.
RMS ERBE-ISCCP2
The RMS is almost everywhere less than 10 W/m2, and certainly looks
better that ISCCP without the correction.