1. AIRWAVE: An (A)ATSR based TCWV data record from ESA S. Casadio 1,2, B. R. Bojkov 2, B.M. Dinelli 3, E. Castelli 3, E. Papandrea 3, A. Burini 2,4 1 SERCO SpA 2 ESA-ESRIN 3 CNR-ISAC Bologna 4 RHEA S.a. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)1

2. Preamble Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)2 ESA need for a long term analysis of Along Track Scanning Radiometer (ATSR) stability and consistency of Level 1 products. The ESA LTDP ATSR Long Term Stability (LTDP-ALTS) project aims at: SWIR recalibration (now all channels) using “novel techniques”, Cloud Mask (new algorithm), Water Vapour (new algorithm), Analysis of South Atlantic Anomaly (SAA) using SWIR noise night-time signals, Historical (A)ATSR documentation rescue Constraint: Cloud masks and water vapour algorithms shall be “purely physical” (i.e. no empirical fitting or “soft-calibration” against independent products). The aim is to verify the stability of the instruments in terms of radiometric and spectral calibration.

3. The Along Track Scanning Radiometer Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)3 ATSR-1 (1991-1996, ERS-1) ATSR-2 (1995-2003, ERS-2) AATSR (2002-2012, ENVISAT) Pros Dual view capability (0° and 55°) Stable orbit tracks (ANX 10:00pm, 10:30pm) TIR channels calibrated on-board (BB) Relatively high spatial resolution (1x1 km 2 ) Same instrumental setup Cons Small swath (500 km across track) Only 10.8 and 12 µm BTs always available (+SWIR)

4. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)4 ATSR-1 ATSR-2 AATSR Wavelength µm MODTRAN CO2 transmissivity [NADIR] Along Track Scanning Radiometer TIR filter functions (as measured in the laboratory)

5. TCWV retrieval: assumptions Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)5 1) NADIR and FORWARD views are CO-LOCATED 2) The operational Cloud-Mask in Level 1 products is CORRECT 3) The Radiometric Calibration is ACCURATE 4) The instrumental noise is ACCORDING TO SPECIFICATIONS V2 V3

6. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)6 AIRWAVE Method (simplified) The current version of AIRWAVE works for cloud-free / sea surface scenes. Same retrieval parameters for day-time and night-time conditions. ATSR BTs are converted in radiance (inversion of Planck’s law) Single channelSingle channel SingleviewSingleview <= “pseudo-column” Wien’s law

7. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)7 AIRWAVE Method (simplified) Dual view exploitation: linear combination of NADIR and FORWARD pseudo-columns Φ to get rid of the unknown “G” The combination parameters α, β, and δ can be calculated using RTM! All contributions are well separated and can be analysed/evaluated/updated independently from each other!

8. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)8 Empirical test on “constant δ” assumption using ERA-I TCWV and ATSR pseudo-columns AIRWAVE Method (simplified)

9. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)9 AIRWAVE Method (simplified) Apart from the measured ATSR radiance, other “known” parameters are: ATSR filter functions (10.8 and 12 µm): as measured before launch H 2 O absorption cross sections: from MODTRAN simulations (6 atmospheres, 18 TCWV values, 2 viewing angles) “constant δ”: RTM simulations (CNR-ISAC), “empirically” using ERA-I CO 2 optical thickness: MODTRAN simulations (6 atmospheres, 345->385 ppmv, 2 viewing angles) Sea Surface Emissivity: spectral data (model) created by Owen Embury, Christopher Merchant, and Mark Filipiak (University of Edinburgh, Institute of Atmospheric and Environmental Science, http://datashare.is.ed.ac.uk/handle/10283/17) http://datashare.is.ed.ac.uk/handle/10283/17 … we are done! λ 1 [µm]λ 2 [µm]δαβ ATSR-110.915911.91071.6550.7-49.7 ATSR-210.930212.04851.6350.5-49.5 AATSR10.844512.03211.6253.1-52.1 Casadio et al. (2015) “Total Column Water Vapour from Along Track Scanning Radiometer Series Using Thermal Infrared Dual View Ocean Cloud Free Measurements: the Advanced Infra-Red WAter Vapour Estimator (AIRWAVE) algorithm”, Remote Sensing of Environment (in press)

10. AIRWAVE on ESA G-POD Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)10

11. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)11 TCWV field (0.25° grid, 1991-2012 average)

12. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)12 AIRWAVE TCWV zonal average anomalies (reference: 1999-2002). The following El Niño events are clearly detectable: 1991-1992 (moderate), 1997-1998 (very strong); 2003 (moderate); 2009-2010 (moderate)

13. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)13 Percentage of cloud-free ATSR pixels (ocean, monthly mean). Blue = ERA_I – ERA-I(cf) Green = ERA-I – AIRWAVE Brown = ERA-I(cf) – AIRWAVE Red = ERA-I – AIRWAVE (collocated) Same colour code for RMSE (standard deviation). Vertical lines indicate the different ATSR operation periods. CLOUD-FREE DRY BIAS = 0.2 g/cm 2 CLOUD-FREE RMSE = 0.3 g/cm 2

14. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)14 Percentage of cloud-free ATSR pixels (ocean, monthly mean). Blue = ERA_I – ERA-I(cf) Green = ERA-I – AIRWAVE Brown = ERA-I(cf) – AIRWAVE Red = ERA-I – AIRWAVE (collocated) Same colour code for RMSE (standard deviation). Vertical lines indicate the different ATSR operation periods.

15. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)15 Percentage of cloud-free ATSR pixels (ocean, monthly mean). Blue = ERA_I – ERA-I(cf) Green = ERA-I – AIRWAVE Brown = ERA-I(cf) – AIRWAVE Red = ERA-I – AIRWAVE (collocated) Same colour code for RMSE (standard deviation). Vertical lines indicate the different ATSR operation periods.

16. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)16 Left panel: AIRWAVE (AATSR) TCWV; right panel: TCWV from Weather Research and Forecasting Model (WRF, http://www.wrf-model.org/index.php). Lee waves, generated by the Mistral wind (NW) after the passage of a cold front, are clearly visible downwind Corsica and Sardinia in both datasets

17. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)17 AIRWAVE vs. Analysed RadioSoundings Archive (ARSA, LMD) TCWV (1991- 2012). Left panel: AIRWAVE vs. ARSA scatterplot; right panel: location of ARSA stations used for validation. Colours indicate the number of collocations (red=high, blue=low) Cloud-Land masking errors

18. Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)18 Malaysia MY 48615 KOTA BHARU 6.17 102.28 5 1973 2013

19. Conclusions Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)19 New algorithm for TCWV retrieval for day&night – cloud free – sea scenes using dual view TIR measurements from Along Track Scanning Radiometer series. It is based on the knowledge of: The instrument filter functions (measured) The viewing angles (known) The absorption cross sections (RTM) The sea surface emissivity (model) and No atmospheric or sea surface temperature a priori / initial guess No fitting/calibration to independent TCWV dataset Validation against radiosondes (level 2): on-going (preliminary results very promising) Comparison with SSM/I REMSS (level 2): on-going (some examples in Casadio et al. 2015, RSE) Comparison with ESA MWR TCWV (level 2): started collocation entire dataset Comparison with ENVISAT-MERIS (level 2): not started yet Comparison with SSM/I HOAPS (level 3): after data available up to 2012 Future AIRWAVE developments Refinement of retrieval parameters for sea scenes TCWV retrieval for cloud free - land scenes Easily adaptable to Copernicus Sentinel-3 SLSTR (Sea and Land Surface Temperature Radiometer)

20. Thank you for your attention! Casadio et al.GVAP Workshop, 4-5 November 2015, Madison (USA)20