AIRS Cloud Cleared radiances versus clear FOV radiances

This web page last updated: 20 June 2008

Introduction

We have processed the first week of 2004/05 AIRS Cloud Cleared (CC) data with ECMWF profiles to compute obs-calc BT statistics for nighttime ocean observations. The data has been binned by cloud amount "TotCld_4_CCfinal" and latitude. CC radiances for each AIRS channel were quality filtered using Joel Susskind's "method1" with a 0.9 K error limit, and "method2" with the 3.5 noise amplification limit. This data is compared to our latitude binned monthly mean "uniform_clear" (UC) data set for the same month.

Latitude 0 to 10 degrees

Here a comparison for the 0 to 10 degrees latitude bin for various cloud fractions. The mean CC is blue and the UC is red. The two curves are generally similar except in the window regions where for cold bias is CC is noticeably larger. The green curve is the mean with +- the standard deviation added, while the black curve is mean with +- the error estimate.

clouds 0-10%
clouds 10-25%
clouds 25-50%
method2, 0-10%

Latitude -50 to -40 degrees

Here a similar comparison for the -50 to -40 degrees latitude bin. The cold bias is noticably worse and exceeds the error estimates.

clouds 0-10%
clouds 10-25%
clouds 25-50%
method2 0-10%

Passing CC Radiance Channel fraction

The following plots show the relative fraction of channels passing the quality filter if any channel passes. This is a mean over all observations with at least one passing channel. The different colors are latitude bins: -50 to -40 (blue), -40 to -30 (green), -30 to -20 (red), -20 to -10 (cyan), -10 to 0 (magenta), 0 to 10 (yellow), 10 to 20 (black), 20 to 30 (blue), 30 to 40 (green), and 40 to 50 (red). Note that window channels at high latitudes are less likely to pass than at tropical latitudes, and that the fraction passing decreases as cloud fraction increases.

clouds 0-10%
clouds 10-25%
clouds 25-50%
method2, 0-10%

Standard Deviation Comparison

The following plots show the standard deviation of the CC (blue) and UC (red) data. The standard deviation of the CC grows as cloud fraction increases. (Note: the UC calc was done with the v4 fast model rather than the V5 fast model used for the CC calc; this has little effect on the comparison except in the 650 & 2300 cm^-1 stratospheric regions where the CC/V5 benifits from some tuning not present in the UC/V4).

clouds 0-10%
clouds 10-25%
clouds 25-50%
method2, 0-10%