Common Principles
The interoperability of many GHRSST data sources provides optimum scientific return yet it requires a common and uniform method for the estimation of uncertainties across all data sources. Some common principles for Single Sensor Error Statistics (SSES) have been agreed at GHRSST X, refined at G-XI, G-XII and G-XIII.
SST should be the best estimate prior to the production of SSES and that is the responsibility of the SST data producers. A common scale for the quality levels is agreed ranging from quality level 2 (worst quality for this specific sensor) to 5 (best quality for this specific sensor). This is clearly defined for each SST data producer. Furthermore the derivation of quality indicators is traceable, i.e. documented and available to users.
The SSES include bias and standard deviation relative to the agreed reference source of drifting buoys and are supported by a quality level. They are uniquely defined for each L2P product according to the SSES Common Principles shown below and are documented and traceable.
- A bias and a standard deviation reflecting the local accuracy (at pixel) of the SST estimate
- Application of SSES is consistent with the product definition (skin; sub-skin)
- The reference sensor is drifting buoys
- Hierarchical references can be used, e.g. global statistics using drifting buoys
- Regional statistics using other reference sources such as radiometers, GTMBA, L4 analyses (really not recommended for IR)
- Common match-up thresholds (Centre pixel clear, +/- 2 hrs – ideally 30 mins)
- No discontinuities between quality levels so continuous fields are ensured although discontinuities may be inevitable.
- SSES must be free from diurnal variability, ideally estimated from night-time match-ups only
- L2P producers that provide SST-skin should use, as a minimum, a constant offset of 0.17 K to adjust SST-skin to SST-sub-skin for SSES production
- If sufficiently accurate wind speed data is available then L2P producers are encouraged to allow for the wind speed dependence of the skin to sub-skin adjustment.
SSES Descriptions
This page will contain descriptions of Single Senor Error Statistic (SSES) schemes used by GHRSST L2P GDS 2.0 compliant data sets:
- (A)ATSR
- Match-up Database (MDB) covers all 3 missions
- Spatial matching to the nearest pixel
- Temporal matching 2 hours
- Consistent SSES scheme applied to all 3 missions (not optimal for early ATSR-1)
- Bias and standard deviation compared to drifting buoys using a threshold for wind speed of 6 m/s.
- Global field
- Skin adjustement to drifters: 0.17 K
- METOP AVHRR (OSI-SAF)
- In situ drifter data from the Global Telecommunications System (NCEP GTS), within 5C of climatology
- Time collocation +/- 3 hours
- Spatial collocation criteria using central SST pixel over a box of 21*21 only sea water pixels
- METOP IASI (EUMETSAT)
- Collocate distance less than 12 km
- Collocate AVHRR and IASI within 1 hour
- Only night-time observations over water, SSTskin to SSTsub-skin correction adding 0.17K
- In situ drifting buoy SST within 8K of climatology
- Based on observational uncertainty estimations from 1D-VAR SST retrieval
- Quality level assigned for each observation according to Total Column Water Vapour (TCWV)
- NOAA 16-19 & MetOp-A AVHRR (NOAA STAR NESDIS)
- In situ data used are drifters and tropical moorings
- SST within 1hr and 20 km from in situ
- NOAA 11-19 AVHRR (IMOS)
- Using only drifting buoys
- Applying sub-skin to skin adjustement of -0.17 C
- Use SST with quality levels 2 to 5
- MatchUps restricted to wind speeds higher than 6 m/s during daytime and higher than 2 m/s during nighttime
- MSG SEVIRI (OSI-SAF)
- In situ drifter data from the Global Telecommunications System (NCEP GTS), within 5C of climatology
- Time collocation +/- 1 hour
- Spatial collocation criteria using central SST pixel over a box of 21*21 only sea water pixels
- SSES determined by quality level value at each pixel
- SSES are attributed after operational statistics are stabilized
Please contact the GPO for more details.