Simulation study for the Stratospheric Inferred Wind (SIW) sub-millimeter limb sounder

Transcription

1 Simulation study for the Stratospheric Inferred Wind (SIW) sub-millimeter limb sounder Philippe Baron1, Donal Murtagh2 (PI), Patrick Eriksson2, Kristell Pérot2 and Satoshi Ochiai1 (1) National Institute of Information and Communications Technology (NICT), Japan (2) Chalmers University of Technology, Sweden 1

2 Context Global data on middle-atmosphere dynamics are needed (20 80 km) Analysis and re-analysis are constrained with temperature only up to ~50 km and observations have poor vertical resolution (nadir sounders) Limb sounders can provide temperature but they are few (Odin, SABER, MLS, ACE/FTS) and they are near their end. No current satellite system provides winds in the middle-atmosphere No replacements are planned for current limb sounders and there is a risk for a time gap in monitoring the stratosphere composition. SIW is one of the three instruments under study for the next Swedish micro-satellite Innosat (2020). Its objective is to provide wind, temperature and various molecules between km. 2

3 Wind observation with SMM limb sounder Wind from SMM limb sournders: MLS: km1 (O2 line at 118 GHz) SMILES: km2 (O3 and HCl lines at 625 GHz) SMILES radiometer was cooled to 4-K to get high SNR. This requires a large plaform and the mission is life limited to less than 5 years, SMILES sensitivity can also be obtained without 4K cooling: - larger bandwidth (1 GHz 8 GHz) - Strong O3 lines cluster3 O3 line distribution [1] Wu D., et al.: Mesospheric Doppler wind measurements from Aura Microwave Limb Sounder (MLS), Adv. in Space Res., 42, , 2008 [2] Baron P., et al.: Observation of horizontal winds in the middle-atmosphere between 30S and 55N during the northern winter , ACP 13(13), , 2013, doi: /acp [3] Baron P., et al, 2013.: Definition of an uncooled submillimeter/terahertz limb sounder for measuring middle atmospheric winds, Proceedings of ESA Living Planet Symposium. 3

4 Example of JEM/SMILES results Zonal-mean zonal wind 35 km 60 km 4

5 Wind measurement principles Background Limb ~2000 km High resolution spectra (JEM/SMILES, Ochiai et al. 2015) ~7 km/s FOV=~5 km LOS wind υ is derived from the Doppler shift frequency: df υ = 5 m/s df = 0.01 MHz for λ = 0.5 mm Small Doppler shift compared to linewidths =>Small wind signature Using 2 perpendicular line-of-sight allows us to reconstruct the horizontal wind vector. 5

6 SIW characteristics Double side band radiometer 625 GHz GHz Bandwidth 8 GHz / resolution 1 MHz Passively cooled Tsys(dsb) K 2 antennas with 30 cm diameter Azimuth wrt orbit: 45 and 135 Effective vertical field of view: 5.1 km Scan from 10 to 90 km (32 s, velocity 0.05 /s) Integration time: 0.5 s (1 spectrum / 1.2 km) Sun-Synchronous polar orbit (97 ) near terminator: ~60S-80N (~150 scans / orbit) Antenna beams oriented toward night Murtagh, D., 2016: SIW Stratospheric Inferred Winds. Proposal in response to a Scientific Swedish Satellite based on the InnoSat Platform 6

7 Observation assumptions in this study Calibration: hot load: 1 sec cold sky: 2 sec CAL CAL 90 km Atmosphere: 0.5 s/spectrum 10 km Antenna 1 Antenna 2 32 s 32 s 3s Antenna 1 CAL Antenna 2 CAL 7

8 Simulation Forward model Retrieval simulation (2 iterations, no regularization) Vertical resolution: 4 km Tropics, Night time Retrievals (red circles) Truth (green): MLS (blue dashed line) + perturbation 8

9 Retrieval errors: Radiometer thermal noise Single scan error (1 sigma) Tsys = (1200 * 1.2) K Night time Day time Line thickness show the latitudinal variability between 80S-80N (Winter conditions) => Most of the observations are performed in night-time 9

10 Systematic errors (bias) Spectroscopic parameters for H2O and HCl lines and 50 strongest O3 lines 1% error on the line strength 2% on air broadening parameter 10 khz on transition frequency (5 m/s) 10

11 Conclusion SIW is one of the three missions selected for the 2020 Swedish INNOSAT satellite. Currently in phase A/B study. Final decision in Autumn First SMM satellite mission designed for wind measurements: km, precision <10m/s, vertical resolution 5 km (<5 m/s between km) Temperature, O3, H2O, HCl can be measured between km with a vertical resolution of 3 km as well as various chemical species ( 18O3asym, HNO3, N2O, ClO, HO2, NO) with a resolution of 5 km. Spectroscopic parameters of O3 lines are main systematic errors for wind and H2O. Realistic constraints on the knowledge of other instrument parameters have been defined (not shown) 11

12 Additional slides 12

13 Retrieval error analysis Retrieved state: Measurement errors (thermal noise, calibration) x = x0 + (D + δd) (y + δy y +δy ) Forward model errors: (spectrosocpic parameters, lower/upper bands ratio, Pointing...) With y0 = F(x0) and D = (df/dx)-1 13

14 Measurement errors δy = Thermal noise on atmospheric spectra εatm(0.5 s) and calibration data εcold(2 s), εhot(1 s) Equivalent to a random noise with STD: Systematic error induced by calibration (hot load emission, antenna spill over, receiver non-linearity) Row radiometer output (ADU) : C = g (1 - α<c>) (Tsys + η y + (1- η)tso ) Receiver non-linearity is characterized by α * 1.2 Use of simple top/bottom calibration data average. Retrieval errors due to thermal noise Blue line: Proper noise modeling Red line: Simplified noise model 14

15 Atmosphere O3, H2O, HCl, temperature from AURA/MLS Winter (DJF, 2009/2010), 2 MLS local times Zonal mean in latitude bins of 10 Other molecules: single profile from a photochemical model 15

16 Other molecules 16

17 Instrumental / calibration biases Lower and upper bands ratio Color lines: ratio ±1% in sub-bands of 1.2 GHz Blak lines: ratio ±1% over the full bandwidth Radiance calibration Dashed lines: Hot load temperature ±1% Green lines: non linearity target, α=10-6 Blue lines: non linearity, α=10-5 Future work: the error may be reduced by retrieving the ratio since strong O3 lines are distributed over the whole bandwidth in both sub-bands. 17

18 Other uncertainties Pointing offset: A global offset of the scan will be retrieved. The remaining uncertainties will be variables and small compared to the thermal noise. Satellite velocity: A knowledge on the azimuth and nadir angles better than 0.3 mrad is needed (bias< 1m/s). This can be obtained with star trackers. Local oscillator drift and any other wind retrieval biases: They will be estimated using daily zonal average of Equatorial meridional winds and midlatitudes analysis data. Bias between km was reduced to ~ 2 m/s for SMILES. Spectrometer: Auto-correlator spectrometer will be used. They should not bring significant errors. Radiometer gain variabilities: The drift and random variabilities between 2 calibrations have not been taken into accounts.we need information on the radiometer stability and calibration algorithm that can fit the drifts. 18