Advanced characterization of aerosol properties through the combination of active/ passive ground-based remote sensing (and in situ measurements) Test of new approaches to retrieve aerosol properties from Photometer- LiDAR joint measurements Q. Hu, P. Goloub, O. Dubovik, I. Vesselovskii, T. Podvin, A. Lopatin B. Torres, T. Lapyonok, Frascati 2-3 Dec. 207
Progress in Phase 2 (207) - Instrument upgrad (Triple polar. / Near Range) - GARRLIC/GRASP algorithm update - Validation against Daytime Raman - Uncertainty on aerosols properties Scheduled activities for Phase 3 (208 and 209 ) 2
LiDAR upgrade and Data Quality LILAS is a High Performance and Transportable Multi-wavelength Raman LiDAR Three emitting wavelengths: 355nm, 532nm, 064nm Raman channels: 387nm (VR), 408nm, 530nm (RR) Three depolarized channels: 355nm, 532nm, 064nm (since march 207). As EARLINET/ACTRIS station LILAS LIDAR follows EARLINET/ACTRIS QC/QA and calibration procedures (annual request to LICAL TNA) Example : Stratospheric «Smoke «layer (Aug/Sep. 207) Good example to have insight in the data quality! GRL paper in preparation
Average Aerosol profiles (20:00-22:00, 3 Aug, 207) LiDAR stand alone retrievals (3β+2α) Backscatter [km - sr - ] -0.0030-0.005 0.0000 0.005 0.0030 20 AOD (532 nm) = 0.0 That s a lot!! Height [km] 9 8 β 064, Klett β 532, Raman β 355, Raman β 532, Klett β 355, Klett 7 α 532, Klett α 355, Klett 6 0.00 0.03 0.06 0.09 0.2 0.5 0.8 Extinction [km - ]
Spectral Volume and Particle Lin. Depolarisation Extinction and Backscatter Angström Exponent 20 EAE, BAE -4-3 -2-0 2 VLDR 355 9 VLDR 532 VLDR 064 PLDR 532 BAE 355-532 EAE 355-532 Height [km] 8 PLDR 355 Part. Lin. Depolarisation 355-532-064 nm Lille-LOA-LILAS 27 8 5 % 7 Leipzig, TROPOS 24 8 4.5% 6 0 0 20 30 40 50 60 PLDR, VLDR
Add-on Near-Field Range Capability (207- January 208) Decrease the blind zone range to 50 m Additionnal small telescope + additionnal detection channels (national fundings) Increase consistency between column (photometer) and profile (LiDAR) Better connect to in situ optical measurements (add groundbased data to cover 0-20 km range) for future integration in the retrieval- 6
GARRLIC/GRASP Algorithm update / improvments Many! Major one : inclusion of multispectral VDR in GARRLIC retrieval (sensitivity study made) 7
Aerosol Retrieval Raman & GARRLiC retrieval Profiles Night (LiDAR stand-alone) LS //, LS Calibration VLDR PLDR BAC Raman LS Raman retrieval LR EXT Elastic LS WVMR EXT: Extinction Coefficient BAC: Backscattering coefficient LR: LiDAR ratio PLDR/VLDR: Part./Vol. depol. ratio WVMR: Water vapor mixing ratio Day (Photometer+LIDAR) TOD/AOD +ALM GRASP/ GARRLiC SD AVC CRI SF SD: Size distribution SF: Spherical fraction AVC: Aerosol Vertical Concentration (profile CRI: Complex refractive index 8
What do we retrieve with GARRLIC + Uncertainty Reminder? AC f,c (h) + Uncertainty Vertical properties? (h) = i=f,c X ai AC i (h) Size Distribution (SD) (column) Aerosol Vertical Concentration (AVC) Examples: s(h) = f AC f (h)+ c AC c (h) Complex Refractive Indices (CRI) P ii ( ) =P ii,f ( ) f s + P ii,c ( ) c s,i + Uncertainties + Uncertainty ~2 Sphere Fractions (SF) Estimate (propagation) in progress (first results shown)
Primary Data (Cloud-Screened Level ) Sun/sky photometer (CIMEL) measurements & Uncertainty (via AE): ü Spectral Total/Aerosol Optical Depth (TOD/AOD) 0.0 (Abs.) ü Spectral angular sky radiances (ALM) 3 % ü Spectral angular sky polarization (POL) 0.005 (Abs.) LiDAR measurements & their uncertainties : ü Elastic backscattering LiDAR Signal (LS), 355, 532, 064 ü Raman backscattering LS: 387, 408, 530 nm. ü Perpendicular & parallel polarized LS: 355, 532,064 nm, VDR 0% 0% 5-0% Conditions : (i) colocation between photometer and LiDAR (ii) LiDAR range (200m-20 km) (link with AOD and in situ) (iii) Cloud free profile => Apply LiDAR Stand-alone and GARRLIC retrievals and compare 0
Evaluation of GARRLIC retrievals : not so simple!! Several Methods : - M- Field campaign (Tsekeri et al., 207; aerosol concentration, absorption with inherent difficulty due to difference between in situ and remote sensing) - M2- Comparison between Raman technique (3β+2α) and GARRLIC results Difficulty : for comparison (M2) - D- Raman technique => nightime - D2- GARRLIC technique => daytime Solutions : - S- improve instrument to have better Raman signal during day (Rot. Raman) - S2- include all information in GARRLIC (like depolarisation) We considered both solutions We selected a complex case with Dust and Biomass burning smoke particles (AOD(440nm) = 0.60, day time period average 8h30-9h40)
Aerosol Retrieval Biomass Burning and Dust (SHADOW campaign, in Senegal, 205-206) LILAS transportable Mie-Raman multi-wavelength LiDAR LiDAR Elastic and Depolarization Signals Hu et al., in preparation, LOA 2
Aerosol Retrieval Biomass Burning and Dust (SHADOW campaign, in Senegal, 205-206) LILAS transportable Mie-Raman multi-wavelength LiDAR GARRLIC Retrievals 3
Aerosol Retrieval Biomass Burning and Dust (SHADOW campaign, in Senegal, 205-206) LILAS transportable Mie-Raman multi-wavelength LiDAR Comparison LiDAR alone (Raman) versus GARRLIC retrievals + Impact of depolarisation 5 GARRLIC- no depolarisation & GARRLIC with 2δ-depolarisation Raman (08:30-09:44)-GARRLiC-no depol 5 Raman (08:30-09:44)-GARRLiC-depol 4 4 Height [km] 3 2 Height [km] 3 2 α 532 -GARRLiC α 355 -GARRLiC α 532 -Raman α 355 -Raman 0 0.0 0. 0.2 0.3 0.4 0.5 Extinction [km - ] Hu et al., in preparation, LOA 0 0.0 0. 0.2 0.3 0.4 0.5 Extinction [km - ] Coïncident Day time Raman and Day time GARRLIC Very good results (improvement of extinction with GARRLIC) 4
Aerosol Retrievals and uncertainty Extinction : GARRLIC with 2δ-depolarisation + error bars Still some bugs! Raman and GARRLIC technique uncertainty on extinction are comparable Near range missing 5
What about extinction-to-backscatter ratio (LiDAR ratio)??? GARRLIC- no depolarisation 5 5 GARRLIC with 2δ-depolarisation 4 4 Smoke Smoke Height [ km] 3 2 Dust 3 2 Dust LR 355 LR 532 Raman 0 0 30 60 90 20 50 LR [sr] 0 0 30 60 90 20 50 LR [sr] LR 355 LR 532 GARRLIC Inclusion of depolarisation improves LiDAR ratio also (value and spectral behavior)
Wavelength [nm] How well measurements are fitted? Optical Depth (photometer).4 Smoke + Dust Case Residual.2 error = 0.66235% TOD.0 0.8 0.6 Fit Meas 0.4 0.2 0.0 400 500 600 700 800 900 000 00
Radiance (photometer) How well measurements are fitted? Smoke + Dust Case (3 < scattering angle < 30 ) Uncertainty on radiance = 3 % Residual R 440nm 0. Total error = 4.63% meas fit 0 20 40 60 80 00 20 40 R 675nm 0. R 870nm R 020nm 0 0. 0 0. 0 20 40 60 80 00 20 40 0 20 40 60 80 00 20 40 0 20 40 60 80 00 20 40 Scattering angle [ o ]
LILAS profiles 6 5 How well measurements are fitted? error = 3.% Meas Fit Smoke + Dust Case Spectral RCS Signals and Depolarization 6 5 error = 9% 6 5 6 error = 4.9% error = 0% 5 6 5 Residuals error = 3.96% 4 4 4 4 4 Height [km] 3 2 3 2 3 2 3 2 3 2 0 0.0000 0.0002 0.0004 0.0006 0.0008 LS 355nm 0 0 2 4 6 8 0 VLDR 355nm 0 0.0000 0.0002 0.0004 LS 532nm 0 0 5 0 5 20 VLDR 532nm 0 0.0000 0.0002 0.0004 0.0006 LS 064nm
Too good? Only one case? => A second case, a pure dust case provides similar good results for GARRLIC 20
Conclusions / Perspectives Instrument LILAS contributing station to EARLINET/ACTRIS LILAS follows QC/QA and calibration procedures LILAS is high performance system (stratospheric aerosol can be detected and characterized) It will be improved again: - Three depolarisation channel (done) - Near Field range (in progress) - Automation of measurements and calibration (in progress) (Supported by national projects) 2
Conclusions Retrievals The LiDAR measured VLDR can be fitted by GRASP/ GARRLiC two mode retrieval with a certain residual. The inclusion of VLDR into GRASP/GARRLiC improves the retrieval of aerosol optical and microphysical properties, e.g. Angstrom Exponent, extinction, LiDAR ratio. Reliable Aerosol Profile Day Time! Modelling of the uncertainty (uncertainty is also a product 22 of the inversion, associated to each inversion)
Future plan (phase 3) 208 (Ph.D Defense, mi-december 208) 2 x publications as st author scheduled + Thesis report to write ( publication published, as co-author). Attempt to integrate ground-based data in the retrieval (more relevant with additionnal Near Field capability) before end of thesis or during post-doc 209 and after, searching for a post-doc support. Repeat the SHADOW campaign in the framework of EARTHCARE CAL/VAL: => include Dakar IRD observatory for Cal/Val in the Tropics (frequent cloud free situation, rich in aerosol and varibility in their properties, plus cirrus and very good technical staff there). See french proposal ACTRIS-FR to ESA EARTHCARE Cal/Val Call.. Field Campaign.5 month in China (Kashi),«Belt and Road Initiative», 209.. New Multi-pixel inversion challenge for Day, for Day and Night sun/photometer and LiDAR.. Contribution to DIVA (Phase 2, ) 23
Field Campaign at M bour station welcome! Abundant aerosol content Aerosol type variability (marine, dust, smoke, local pollution) Seasonal features Permanent AERONET obs. (22 yrs), Routine LiDAR obs. (2 yrs), Field Campaigns (, 203, 205, 206,..) M bour IRD Station of geophysics and oceanography (IMAGO), with qualified technical staff Multiple instruments are deployed during the campaign. 2-week measurements are collected in Mar, Apr, Dec 205 and Jan 206. Contribute to different studies: aerosol, wind, cloud 24
Thank you. 25
Aerosol Retrievals and uncertainty Raman technique (home made) versus SCC (Single Calculus Chain= official reference sofwtare for Level 0-> level EARLINET) 5 Raman (08:30-09:44)-GARRLiC-depol Height [km] 4 3 2 Reference 0 0.0 0. 0.2 0.3 0.4 0.5 Extinction [km - ] α 532 -GARRLiC α 355 -GARRLiC α 532 -Raman α 355 -Raman Should I show this slide?? Possible difference in smoothing method between Home made Raman retrieval and SCC?