150 likes | 288 Views
Examination of HIRS OLR from RBPGS and ATOVS Systems. Hai-Tien Lee March 13, 2005. Background . HIRS OLR retrieval is performed in two systems - RBPGS (Operational) and ATOVS (Experimental, A4)
E N D
Examination of HIRS OLR fromRBPGS and ATOVS Systems Hai-Tien Lee March 13, 2005
Background • HIRS OLR retrieval is performed in two systems - RBPGS (Operational) and ATOVS (Experimental, A4) • For consolidation purposes, we need to first confirm whether these two systems produce identical retrievals. • Known issues - limb darkening correction methods are different
Method • Collocate all-sky HIRS OLR retrievals on FOV level. Collocation criteria: the lat/lon are to within 0.1° and the time stamps are to within 30 seconds. • Calculate OLR difference • Show OLR difference as functions of local zenith angle or satellite view angle
Data • One full day of orbital data of March 8, 2005 from two systems were usedRBPGS: RBOBS data fileATOVS: R1T data file • Number of retrievals: 895,227 (RBOBS) and 610,340 (R1T). • Number of collocations: 564,642
Result • The HIRS OLR (only in the collocated subset) plotted as functions of satellite view angle for both retrieval systems ATOVS RBPGS
Result (cont.) • Following plots show the OLR differences as functions of satellite view angle (SVA) stratified by various conditions: all, day, night, and 12 ERBE Scene IDs. (ERBE Scene ID is determined in the RBPGS for daytime only)
Day Night
Conclusions • Large differences (range from about -10 to 25 Wm-2) of HIRS OLR were found between the RBPGS and ATOVS retrievals. • These differences seem to originate from the applications of different limb darkening methods. • This comparison cannot show that the limb darkening correction is the ONLY difference between the two systems. • This comparison needs to be repeated after the ATOVS system is updated.
Other Observations/Issues • RBPGS system did not perform HIRS OLR retrieval in some cases (see distribution below) but ATOVS did. Reason unknown.
Other Observations/Issues (cont.) • There are duplicated retrievals in R1T. (why?) • “Thinning in polar areas” in R1T is not desired for OLR retrieval purpose • The collocated retrievals are limited to within about 88°NS. Why? • "Overcast" seems to prevail in the RBPGS Scene ID determination (plot below shows the distribution of daytime retrievals that have scene id 12)
Limin’s explanations • "Overcast seems to prevail in the ERBPGS Scene ID determination..." The problem is resulted from a incorrect conversion from CLAVR-x cloud mask into cloud cover for each target, which I have discussed with you and Istvan. We fixed the bug and made changes on the ported codes on AIX, but not on CEMSCS yet. We need to do some tests first, after then it will be implemented into CEMSCS. • "ERBPGS system did not perform HIRS OLR retrieval in some cases but ATOVS did." The HIRS OLR is collocated with AVHRR OLR, and there are some thresholds for generating both AVHRR and HIRS OLR, which might be part of reason that you see HIRS OLR is missed in RBPGS. • “The RBPGS has more retrieval points than the R1T” There are four dummy points at each HIRS scan line in RBPGS due to the collocation need (with AVHRR GAC).