CrIS2_Polarization_Correction
Tue, Jun 4, 20191 Purpose
Use SNOs to quantify effect of proposed NOAA polarization correction on CrIS.2 SDR product.
2 Data & Code
2.1 SDR data files:
SDR data granules for CrIS.2 were pulled by Steven Buczkowski to:
/asl/data/UW_CrIS_PL/{h5_SDR_J01_FSR_PLoff,h5_SDR_J01_FSR_PLon}
These files are converted to mat format files to be used by my SNO maker,
using: convert_noaa_array.m
and can be found in:
/home/chepplew/data/cris/noaa/sdr_j01_hr/{2018,2019}/
with filenames including 'PLon' or 'PLoff' for those with/without the correction applied (respectively).
The conversion routine is:
/home/chepplew/projects/sno/makeSNO/convert_noaa_array.m
with an associated batch control script.
2.2 SNO data files:
Two sets of SNOs are produced: AIRS:CrIS.2 and IASI.1:CrIS.2 and are available at
/home/chepplew/data/sno/airs_cris2/{NOAA_pon,NOAA_pooff}/HR/{2018,2019}/
and at:
/home/chepplew/data/sno/iasi_cris2/{NOAA_pon,NOAA_poff}/HR/{2018,2019}/
Using the SNO makers in /home/chepplew/projects/sno/makeSNO/
2.3 Analysis
The SNO data are analysed using the methods available in the routines in:
/home/chepplew/gitLib/asl_sno/source/
and summary statistics data files can be found in the parent directories of the SNO data listed above.
3 Results
3.1 AIRS:CrIS.2 SNOs
The two months of SNOs (Dec 2018 and Jan 2019) supply about 240,000 SNO pairs, that are located mostly at high latitudes but do extend over warm tropical regions.
Plots are avalable and shown in the slack:SNO channel.
Although there is some less scatter of the bias between the different CrIS FOVs, the differences are at the level of uncertainty. There is some reduction in the bias compared to AIRS, in some channels more negative. It is NOT possible to say if this represents a reduction in CrIS error relative to the unknown truth.
3.2 IASI.1:CrIS.2 SNOs
The two months (as above) supply about 14,700 SNO pairs, that are exclusively located between about 70 to 75 degrees of latiutude in the north and south hemispheres. There is no statitistically signficant difference between correction applied and not for the individual CrIS FOVs, but a small reduction in the bias compared to IASI. It is NOT possible to say if this represents a reduction in CrIS error relative to the unknown truth.
4 Conclusion.
Uisng the SNO data available it is possible to compare the effect of including the polarization correction to the CrIS SDR data. THe changes are very small and in most cases comparable to the uncertainty. There does appear to be a consistent change in the bias and some improvment in the inter-FOV bias variability. At least the polarization correction does not make the bias worse.
As noted, the effect would be expected to be smaller at nadir than at larger incident angles on the instrument optics. It is presumed that the atmospheric emission is is non-polarized (notwithstanding reflected surface and short wave scattered radiation contribution tothe signal at the sensor).