ODYSSEA In Arctic

Emmanuelle Autret, Jean-François PiolléIFREMER/CERSAT, 29280 Plouzané, FranceEmmanuelle.Autret@ifremer.frjfpiolle@ifremer.fr

Outline

- Method

- Input validation in Arctic

- L3S, L4 validation

- Conclusion

Method (1)

GHRSST L2P products

« re-mapped » L3 products

« collated monosenso L3C products » (/day /sensor)

Intercalibrated collated products (/day /sensor)

« multisensor supercollated L3S products » (/day)

Pre- processing steps

Multi-sensor analysed field

Observation quality control, correction (sses), filtering

Selection based on proximity to analysis time and best quality

Correction of large scale bias field for each sensor (daily reference:0.25 gridded field)

Merging of composite files

Optimal interpolation

Method (2)

Bias correction principles:

A daily 0.25° resolution L3 «reference field » is built every day as follows:

1) Each source is remapped on the 0.25° grid (mean):

2) A median of selected sensor data is kept at each grid point

3) An OI is applied to get a complete reference field

4) The error (correction) is determined as the difference between each sensor and the reference smoothed over 2°.

Ingested Data

AATSR Validation (Arctic)

North of 67 NNighttime drifter measurements only

NOAA-19 Validation (Arctic)

North of 67 NNighttime drifter measurements only

L3S & analysis Validation (Arctic)

North of 67 NNighttime drifter measurements only

Analysis ValidationNorth of 67 NNighttime drifter measurements only

Conclusion

- ODYSSEA uses a multisensor reference

- ODYSSEA showed no bias in 2011

- Since then, ODYSSEA shows a seasonal bias: null in Autumn and significantly negative (-0.5K) in February.

- bias correction is not efficient in the Arctic

- all main sensors show a negative bias