Estimation of wave spectra with SWIM on CFOSAT

Estimation of wave spectra with SWIM on CFOSAT – illustration on a real case C. Tison(1), C. Manent(2), T. Amiot(1), V. Enjolras(3), D. Hauser(2), L. Rey(3), P. Castillan(1) CNES, « Altimetry and Radar » department, France (2) UVSQ, CNRS, LATMOS-IPSL, France (3) Thalès Alenia Space, France celine. tison@cnes. fr

Overview of the presentation ■ SWIM instrument and measures ■ Performance budget w Simu. SWIM – an end-to-end simulator w A real sea state condition w Results IGARSS’ 11 – July, 2011 2

The CFOSAT mission China France Oceanography SATellite ■ Status of the program: w Conception and Development phase w Launch planned end of 2014 ■ SWIM w Measurement of the oceanic wave properties w Real-aperture radar with 6 beams (Ku band) ■ SCAT w Measurement of wind sea surface w Real-aperture radar (bi polar, Ku band) ■ Ku. ROS w Airborne sensor developed by LATMOS w Validation of SWIM and SCAT More on the CFOSAT mission tomorrow – session WE 4. T 10 Altimetry I IGARSS’ 11 – July, 2011 3

SWIM instrument (1/2) SWIM: Surface Wave Investigation and Monitoring Ku-band radar (scatterometer) 6 beams IGARSS’ 11 – July, 2011 4

SWIM instrument (2/2) Swell § Nadir signal SWH and wind speed - Accuracy SWH: max(10% of SWH, 50 cm) -Accuracy wind speed: 2 m/s Wind sea § 10°, 8° and 6° beams wave spectrum - spatial sampling of 70 x 70 Km² - Detectable wavelength : λ ~ [70 - 500] m - Azimuth accuracy: 15° - Energy accuracy: 15% § All beams backscattering coefficient profiles: - Absolute accuracy < +/- 1 d. B - Relative accuracy (between beams) < +/- 0. 1 d. B IGARSS’ 11 – July, 2011 5

Estimation of wave spectra Received power Directional wave spectrum F(kx, ky) Wave topography: ξ(x, y) Signal modulation Modulation of the backscattering coefficient Modulation spectrum Pm Wave slopes Link slope/signal modulation IGARSS’ 11 – July, 2011 6

Simulations End-to-end simulation tool: Simu. SWIM ■ Simulations from the sea surface to the estimated signal w Input = sea state conditions w Output = wave spectrum computation of backscattered intensity and processing similar to the future ground segment use SWIM parameters 0° 2° 4° 6° 8° 10° Max integration time (ms) 51. 8 31. 5 22. 5 30. 1 37. 4 Bc (MHz) 320 320 320 PRF (Hz) 2125 fixed 6407 -6739 variable 6378 -6707 variable 6340 -6667 variable Nimp (fixed) 110 60 60 144 192 237 SNR (d. B) 24. 3 11. 9 8. 7 6. 8 4. 8 2. 3 IGARSS’ 11 – July, 2011

Simulations End-to-end simulation tool: Simu. SWIM Surface computation Nimp pulses per cycles Input spectrum (models, measurements) At a given azimuth direction: - Computation of the Nimp backscattered Backscattered signal pulses (knowing SWIM 2 options: geometry and properties) 1. Computation of the Nimp pulses (with geometrical migrations and noises for each) 2. Computation of one pulse and additions of noise (thermal+speckle) to create the Nimp pulses with central migrations - Addition of the Nimp signals IGARSS’ 11 – July, 2011 8

Simulations End-to-end simulation tool: Simu. SWIM Surface computation Input spectrum (models, measurements) Backscattered signal (knowing SWIM geometry and properties) Estimated modulation spectrum Quality criteria IGARSS’ 11 – July, 2011 9

■ Case of November, 2002 storm in Atlantic ocean Lead to the sinking of the Prestige (oil tanker) ■ Very different conditions during the day w 00: low wind sea + dominant swell w 06: 00: very young wind sea (high wind) + dominant swell w 08: 00: mature wind sea + dominant swell w 15: 00: crossed wind seas (old + young) Wind sea rotated by about 120° IGARSS’ 11 – July, 2011 © BSAM/Douanes françaises A real sea state condition: “Prestige case” Spain Galician coast

A real sea state condition: “Prestige case” IGARSS’ 11 – July, 2011

A real sea state condition: “Prestige case” ■ Available data w MFWAM output with ALADIN winds (Météo France models of wind and waves) 2 D polar azimuth/frequency height spectrum converted into 2 D cartesian wavenumber by bilinear interpolation w Subset of results: 00, 06, 08, 10, 15 UTC (different wind and waves cases) ■ Simulation conditions w Incidence angle: 10° w Nimp = 237 pulses per cycle ( averaging for noise reduction) w Rotation speed = 5. 7 rpm ( 49 cycles / 360°) IGARSS’ 11 – July, 2011

Simulation results Reference: 2 D spectrum from WAM model èSame detection of swell and wind sea partitions èSmall underestimation IGARSS’ 11 – July, 2011 06: 00 UTC CFOSAT/SWIM estimation (simulations from Simu. SWIM)

1 D modulation spectra 6 h UTC Swell Φ=135° (SE-NW look direction) Sea wind Φ = 235° (NE-SW look direction) Φ = angle between satellite track (assumed S-N) and radar look direction IGARSS’ 11 – July, 2011

00: 00 UTC 06: 00 UTC 08: 00 UTC 10: 00 UTC 15: 00 UTC Prestige SOS 14: 00 UTC Reference: 2 D spectrum from WAM model CFOSAT/SWIM estimation (simulations from Simu. SWIM) Hs: 5. 1 m U: 11. 7 m/s IGARSS’ 11 – July, 2011 Hs: 5. 8 m U: 22. 2 m/s Hs: 6. 1 m U: 8. 8 m/s Hs: 6. 5 m U: 21. 0 m/s Hs: 6. 5 m U: 17. 3 m/s

Performance quality Estimation errors on wave direction (Φ), wavelength (λ) and energy (E): SWELL WIND SEA Φ λ E 00 h 3° 0% 4% - - - 06 h 0° 0% 13% 11° 27% 19% 08 h 0° 1% 7% 11° 18% 6% 10 h 0° 8% 8% 11° 6% 7% 15 h 3° 12% 6% 7° 19% 5% <15° <10 -20% <15% Requirements: IGARSS’ 11 – July, 2011

Conclusions ■ Simulations of SWIM wave products w End-to-end simulations w Software with realistic sensor conditions w Accurate results with a large variety of sea state conditions ■ Next steps w Keep-on the definition of the inversion algorithms w Optimize inversion up to wave spectrum estimation of the transfer function (α) IGARSS’ 11 – July, 2011