Skip to content. | Skip to navigation

Personal tools

airs_rdr_test.text

TAI Time = 305468971.359517 ... 305469330.687260 seconds past start of 1993.
This granule spans a range of 359.33 seconds or 6.0 minutes.
Nominal AIRS granule duration is 6.0 minutes.

Read channel properties file:
.
L2.chan_prop.2002.08.30.v9.5.1.anc

Checking channel attributes that are constant for granule...
These channels should not be used because ExcludedChans indicates
known radiometric problems. ExcludedChans echos information in
field AB_State of the Channel Properties File:
238 276 291 324 342 350 353 374 381 391
396 402 412 428 431 432 440 457 458 461
463 474 487 494 495 500 512 525 531 535
537 561 582 586 589 598 603 622 681 688
733 736 747 749 753 755 762 783 800 802
803 806 826 828 834 837 849 859 867 868
875 876 878 883 884 885 887 889 892 894
899 901 906 907 912 915 919 922 923 934
937 943 947 954 956 957 962 965 974 975
981 990 1021 1066 1095 1109 1125 1126 1148 1173
1193 1203 1208 1210 1216 1240 1242 1246 1280 1351
1355 1399 1672 1691 1712 1713 1729 1730 1731 1732
1735 1736 1759 1781 1786 1788 1791 1792 1922 1985
2010 2011 2012 2013 2014 2015 2016 2017 2018 2151
2153 2243 2255 2256 2257 2258 2259 2260 2261 2262
2263 2264 2265 2357 2359 2375

Since we have access to the channel properties file, we'll also
directly check AB_State from there. This should be identical to
ExcludedChans, but it sometimes happens that we have a better
channel properties file available locally than was used when
the level-1B data was produced. In this case 0 additional
channels are flagged:


These channels should not be used because CalChanSummary indicates
a transient problem with high noise:
682 1579
But note that the high noise flag only means that the granule's
internal measurement of noise is much higher than the static
value given in the Channel Properties File. Other channels
may have higher noise levels, but are not flagged because they
are conforming to their expected behavior.

Since we have access to the channel properties file, we'll also
directly check the noise level from there. Let us say that for
our application, we wish to exclude any channel with an expected
noise level greater than 0.7 Kelvins. These 22 otherwise good
channels exceed this level:
288 405 442 447 491 571 583 590 596 597
604 636 664 678 742 832 842 949 970 1093
1397 1515

These channels should not be used because CalChanSummary indicates
a problem was encountered in calculating calibration coefficients:
(None!)


Each AIRS field-of-view also has a "state", indicating
whether the instrument was in science mode when the data
was taken and whether the data was successfully transmitted.
We'll only use fields-of-view with "state" = 0/"PROCESS"

Of 12150 fields-of-view in this granule:
12150 fields of view are in state 0=Process
0 fields of view are in state 1=Special Calibration sequence
0 fields of view are in state 2=Erroneous
0 fields of view are in state 3=Missing data
0 fields of view have some other value for state (should never happen)


We also wish to exclude from our analysis any channels that have
experienced a sudden discontinuity in signal level, or "pop".
But for these events we only need to exclude a given channel
for the specific scan during which the event took place. In
this granule there were 3 pops recorded. 3 of these were in
channels we are already not using because of static or granule-
wide problems discussed above, leaving 0 important "pops".
The locations are:
Scan 19 channel 2153 (known bad channel)
Scan 22 channel 458 (known bad channel)
Scan 23 channel 458 (known bad channel)


If you are sensitive to the shape of the spectral response
function because you are using a radiative transfer algorithm,
such as the AIRS RTA, that requires a well-charactersized shape,
then you should avoid using any channels that have the "SRF Shape"
comment in the Channel Properties File.

There are 0 channels with "SRF Shape" in the channel properties
file but 0 of them are channels that we have already eliminated from
for one reason or another, leaving 0 uniquely marked by "SRF Shape"

If you are sensitive to channel coregistration (Cij) then you can
filter for that one of two ways:
- Permanently avoid the most misaligned channels, based on the
boresight centroids in the channel properties file.
- The field "Cij" in the channel properties file gives a measure
of the spatial coregistration of a channel with reference channel.
Requiring spatial coregistration to be 96% or greater,
eliminates 775 of 2378 channels (32.6%). But 75
of these channels are already flagged for another reason,
leaving 700 uniquely flagged for Cij.
- The fields "Centroid X" and "Centroid Y" in the channel
properties file give the x- and y-offset in millidegrees from
the boresight.
Requiring the overall deviation to be less than 75 millidegrees
eliminates 77 of 2378 channels (3.2%). But 41
of these channels are already flagged for another reason,
leaving 36 uniquely flagged for centroid deviation.
- Dynamically exclude scenes where the radiance difference
between two channels with nearly the same frequency but
boresight centroids indicates that misalignment is an issue.
- SceneInhomogeneous gives one indication of such scenes.
412 of 12150 scenes in this granule trigger both bits
723 of 12150 scenes in this granule trigger just the longwave window bit
2550 of 12150 scenes in this granule trigger just the shortwave window bit
- Rdiff_lwindow and Rdiff_swindow give the actual radiance
differences for two such pairs -- one in the longwave
window spectral region and one in the shortwave window
region.
In this granule Rdiff_lwindow ranges [-11.958, 11.752] with a mean -0.756 +/- 1.697
In this granule Rdiff_swindow ranges [-0.105, 0.114] with a mean 0.005 +/- 0.012
- Or you can craft your own check using the actual radiances of
these channel pairs.
Rdiff_lwindow uses channels:
617 @ 853.95 cm**-1
606 @ 851.61 cm**-1
For the first field-of-view in the file these have
radiances of 100.50 and 99.25 milliWatts/m**2/cm**-1/steradian
respectively, with noise levels of 0.39 and 0.54 respectively

Rdiff_swindow uses channels:
2280 @ 2561.13 cm**-1
2252 @ 2560.76 cm**-1
For the first field-of-view in the file these have
radiances of 0.7305 and 0.7275 milliWatts/m**2/cm**-1/steradian
respectively, with noise levels of 0.0016 and 0.0024 respectively


When using radiances you also have to be aware that individual
readings will be flagged with a value of -9999.0 if no valid
radiance could be calculated. In this granule 704700 radiances (0.02%) are
-9999.0, but 704700 of these are in channels or fields-of-view that
are eliminated for reasons discussed above, leaving 0 marked only
by having radiance = -9999.0

In addition, remember that AIRS radiances can be zero or negative.
This typically happens when the signal level is less than or
on the order of the noise level. So it is more likely for
cold scenes and noisy channels. In this granule 464 radiances (0.00%) are
nonpositive but not -9999.0, but 464 of these are in channels or fields-of-view that
are eliminated for reasons discussed above, leaving 0 "real"
cases of radiance <= 0.0 but != -9999.0

Document Actions
NASA Logo - nasa.gov
NASA Privacy Policy and Important Notices
Last updated: Sep 09, 2009 02:26 PM ET
Top