1. SST – GSICS Connections
2012 GSICS Users’ Workshop 4 September 2012, Sopot, Poland
SST – GSICS Connections
Sasha Ignatov1, Xingming Liang1,2, Prasanjit Dash1,2
1NOAA/NESDIS and 2CSU/CIRA
4 September 2012
SST and GSICS

2. Conducts SST retrievals, leads community JPSS and GOES-R SST efforts
NESDIS STAR SST Team
Conducts SST retrievals, leads community JPSS and GOES-R SST efforts
Part of the international Group for High Resolution SST (GHRSST;
SST Quality Monitor, SQUAM
Works with GSICS to link high-quality SSTs to clear-sky ocean radiances
Monitoring IR Clear-sky Radiances over Oceans for SST (MICROS;
Specific for nesdis
4 September 2012
SST and GSICS

3. SST – GSICS links SST is highly accurate/precise product used for
Climate monitoring; Operational weather & ocean forecasting; Ecosystem assessment; Tourism; Fisheries; Forcing/Validating Ocean models; Military/Defense applications; etc
SST in thermal IR is retrieved from clear-sky ocean radiances
POES – AVHRR, MODIS, ATSR, VIIRS
GEO – GOES, MSG/SEVIRI, MTSAT, GOES-R/ABI
Two types of SST retrievals
Regression: Need radiance stability of individual bands/sensors
Physical (RTM-based): Need absolute accuracy
GSICS-SST Links
GSICS – Provide highly accurate and consistent radiances to SST
SST – Derive accurate SST; Provide feedback to GSICS
Specific for nesdis
4 September 2012
SST and GSICS

4. Nighttime Heritage NOAA minus Reynolds SST in SQUAM
SSTs from different platforms are largely consistent, except N16, and to a lesser extent, N18 from 2011-on
Median, Satellite SST - Reynolds
In what follows, MICROS for the new NOAA retrieval system, ACSPO, is used from 2008-on to link SST with sensor radiances anomalies
Specific for nesdis
4 September 2012
SST and GSICS

5. Nighttime ACSPO Regression – Reynolds SST in MICROS
Note that SST anomalies in ACSPO are different from heritage
This is because of different definition of day and night, and different cloud mask
N16 shows larger deviations from family, but more regular than in heritage system
This viewgraph shows direct differencing but, “double differences” can be used to minimize the effect of unstable reference
4 September 2012
SST and GSICS

6. SST Double Differences (“DD”) in MICROS
Regression – Reynolds Biases
Regression (ACSPO) = at night, MCSST (linear combination of 3.7, 11, and 12µm bands)
Reynolds = Global daily 0.25° analysis
Double Differences (“DD”) for SST
Reynolds used as a “Transfer Standard” and cancels out, leaving “Reg1-Reg2”
DDs cancel out/minimize systematic errors/instabilities in (Sat-Ref) arising from
Errors/Instabilities in Reynolds SST
Updates to SST or cloud masking ACSPO algorithms (simultaneously made for the two platforms being compared)
Day and Night Data are monitored in MICROS. However, only night data are used for cross-platform consistency checks here, to minimize effects of diurnal variability
4 September 2012
SST and GSICS

7. Nighttime DD’s SST (Ref=Metop-A GAC)
N17 (Equator Xing Time, EXT~10:30pm) consistent with Metop-A (EXT~9:30pm)
N19 (EXT~1:30am) consistent with Metop-A and N17
N18 (EXT~1:30am) less consistent; and N16 (EXT~5am) most inconsistent
4 September 2012
SST and GSICS

8. BT Double Differences (“DD”) in MICROS
Model minus Observation (“M-O”) Biases
M (Model) = Community Radiative Transfer Model (CRTM) simulated TOA Brightness Temperatures (w/ Reynolds SST, GFS profiles as input)
O (Observation) = Clear-Sky sensor (AVHRR, MODIS, VIIRS) BTs
Double Differences (“DD”) for BTs
“M” used as a “Transfer Standard” and cancels out, leaving “Obs1-Obs2”
DDs cancel out/minimize systematic errors/instabilities in BTs arising from
Errors/Instabilities in Reynolds SST & GFS
Missing aerosol
Possible systemic biases in CRTM
Updates to ACSPO algorithm; etc
Day and Night Data are monitored in MICROS. However, only night data are used for cross-platform consistency checks here, to minimize effects of diurnal variability and solar contamination
4 September 2012
SST and GSICS

9. Nighttime DD’s @3.7 µm (Ref=Metop-A GAC)
Shape of BT DDs closely reproduces the shape of SST DDs
Cross-platform biases reach several tenths of a Kelvin
They cannot be explain by differences in the Equator Xing Time
4 September 2012
SST and GSICS

10. Nighttime DD’s @11 µm (Ref=Metop-A GAC)
Systematic cross-platform biases are taken care of by Regression SST
They can be also removed by “bias correction” in “physical” retrievals
The most troublesome for SST are radiance instabilities
SST needs help from GSICS, to flatten them out from “first principles”
4 September 2012
SST and GSICS

11. Nighttime DD’s @12 µm (Ref=Metop-A GAC)
Shape of the biases is generally reproducible from band to band, suggesting that they are coming from sensor instabilities
Several tenths of a Kelvin is a large error for SST and must be removed
4 September 2012
SST and GSICS

12. Adding NPP VIIRS and Terra/Aqua MODIS in MICROS
NPP launched on 28 October 2012
Transition to Joint Polar Satellite System (JPSS)
US: VIIRS instrument on PM platforms NPP (2011), J1 (~2016) and J2 (~2019)
Europe: AVHRR instrument on AM platforms Metop-A (2006), - B (2012), and -C (2017)
Links between AVHRR and VIIRS instruments must be established
Cryoradiator doors opened on 18 January 2012
SST Team started processing global VIIRS data on 23 Jan 2012
VIIRS SSTs are monitored in SQUAM, and Clear-Sky Ocean Radiances monitored in MICROS
MODIS Radiances and SSTs were added simultaneously
Terra and Aqua SSTs were added in SQUAM, and Clear-Sky Ocean Radiances in MICROS, simultaneously with VIIRS on 23 Jan 2012
Specific for nesdis
4 September 2012
SST and GSICS

13. Nighttime DD’s @3.7 µm (Ref=Metop-A GAC)
N16: unstable and out of family
VIIRS recalibration
Aqua: out of family by 0.3K
All AVHRRs, Terra/MODIS, and NPP/VIIRS are consistent to within 0.15K
VIIRS is in AVHRR family. Note Cal Change on 7 Mar K)
Aqua: out of family by -0.3K
4 September 2012
SST and GSICS

14. Nighttime DD’s @11 µm (Ref=Metop-A GAC)
VIIRS recalibration
N16: unstable and out of family
Terra & Aqua: out of family by 0.6K
All AVHRRs and NPP/VIIRS are consistent to within 0.15K
VIIRS is in AVHRR family. Note Cal change on 7 Mar K)
Terra and Aqua/MODIS out of family by 0.6K. Reason is not sensor but suboptimal CRTM coefficients in CRTM V2.02. Working to fix.
4 September 2012
SST and GSICS

15. Nighttime DD’s @12 µm (Ref=Metop-A GAC)
N16: unstable and out of family
VIIRS recalibration
Terra & Aqua: out of family by 0.3K
All AVHRRs and NPP/VIIRS are consistent to within 0.15K
VIIRS is in AVHRR family. Note Cal change on 7 Mar K)
Terra and Aqua/MODIS out of family by 0.3K. Reason is not sensor but suboptimal CRTM coefficients in CRTM V2.02.
4 September 2012
SST and GSICS

16. Nighttime DD’s SST (Ref=Metop-A GAC)
New Reg. Coeff. used
VIIRS recalibration
Terra & Aqua SSTs are in family, due to use of empirical regression
N16: unstable and out of family
All AVHRRs, MODISs and NPP/VIIRS SSTs are consistent to within 0.15K
VIIRS Cal Change on 7 Mar 2012: SST +0.10K – Out of family
New SST coefficients implemented 3 May 2012: SST -0.15K – Back in family
4 September 2012
SST and GSICS

17. Summary
MICROS monitors all on-orbit AVHRRs, 2 MODISs on Terra and Aqua, and NPP VIIRS
Most sensors are stable and cross-platform consistent to within <±0.1K
The remaining cross-platform systematic biases are easy to take care of by regression SST algorithm, or by empirical bias correction in physical retrievals
Some sensors are unstable (AVHRRs on NOAA-16 and -18) - Those are most challenging for SST
Two MODIS instruments are out of family. This is a combination of CRTM issues in bands 31/32 (11/12 µm) and sensor issues in band 20 (3.7 µm)
Specific for nesdis
4 September 2012
SST and GSICS

18. Ongoing Work & Remaining Questions
Consistent CRTM for various sensors: Critical for MICROS
MODIS CRTM coefficients inconsistent with AVHRR/VIIRS – appears as sensor problem. In fact, it’s CRTM issue, work is underway to fix
Terra & Aqua/MODIS inconsistent in band 20 (3.7µm) by 0.3K
This appears to be real sensor calibration problem, and the largest “out-of-family” deviation so far. Working with Jack Xiong to resolve
Q1: Are MICROS results consistent with imager trending?
Against spectrometers (AIRS, IASI, CrIS)?
Using Simultaneous Nadir Overpasses?
Q2: If results are consistent, then we should work together to
Stabilize the unstable sensors (N16, N18)
Reconcile BTs form various SST sensors
Do this from “first principles” (better understand sensors, improve cal procedures) rather than “empirical bias corrections”
Specific for nesdis
4 September 2012
SST and GSICS