1. Optical Properties of Mineral Dust Aerosol in the Thermal Infrared IRS 2016 Claas H. Köhler Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 1

2. Measure atmospheric radiance in the spectral range 800–1200 cm -1 (8–12 µm) during SAMUM-2 Set-up a simulation environment (PIRATES) capable to compute thermal infrared (TIR) atmospheric radiation in the presence of aerosols Identify a microphysical aerosol model suited to reproduce the measured radiation at bottom/top of the atmosphere (BOA/TOA) with special attention to refractive index and particle shape Goals Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 2 C. H. Koehler, Radiative Effect of Mixed Mineral Dust and Biomass Burning Aerosol in the Thermal Infrared, 2014, http://elib.dlr.de

3. The SAharan Mineral dUst ExperiMent (SAMUM-2) Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 3

4. The SAharan Mineral dUst ExperiMent (SAMUM-2) Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 4 LIDAR (IfT, LMU, DLR) In-situ size distributions (DLR, TU Darmstadt) Sample analysis (TU Darmstadt) Radiosonde measurements (IfT) Radiation Measurements (Uni Leipzig, LMU, DLR)

5. Investigate influence of a thin mixed layer on TIR remote sensing applications Example: IASI sea surface temperature (SST) product for 6 Feb 2008 TOA Radiative Effect of Mixed Smoke / Dust Aerosol Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 5 www.ghrsst.org

6. Vertical Aerosol Distribution on 6 Feb 2008 Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 6 Courtesy of S. Groß

7. Measurements Entering the Simulation Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 7 11:33 UTC

8. Comparison of Simulation and IASI Measurements Simulation w/o aerosols SST: 295 K Deviations in ozone band (1000 – 1080 cm -1 ) due to profile mismatch Good agreement, so aerosols can be ignored ? Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 8

9. Comparison of Simulation and IASI Measurements Simulation with aerosols Aerosol absorbs terrestrial radiation Simulation underestimates upwelling radiation Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 9

10. Comparison of Simulation and IASI Measurements SST: 296 K (1 K increase) Better fit between simulation and measurement O‘Caroll et al. (2012) report bias of IASI SST compared to in-situ measurements Aerosol has to be taken into account for accuracy better 1 K Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 10

11. Investigate influence of a low pure dust layer on BOA radiance Investigate influence of particle shape and refractive index BOA Radiative Effect of Mineral Dust Aerosol Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 11

12. Vertical Aerosol Distribution on 29 January 2008 Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 12 Courtesy of S. Groß

13. Measurements Entering the Simulation Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 13 22:07 UTC

14. The Spectral Signature of Mineral Dust FTIR measurements from 2024 UTC to 2105 UTC Mineral composition from samples collected on 25 Jan (similar source regions) Internal mixture of spherical particles Agreement not bad, but outside uncertainties Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 14

15. The Spectral Signature of Mineral Dust Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 15

16. The Spectral Signature of Mineral Dust Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 16

17. The Spectral Signature of Mineral Dust Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 17

18. No internal dust model matches measured signature over the entire TIR window Internal Mixtures of Spherical Particles Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 18

19. No external dust model matches measured signature over the entire TIR window External Mixture of Spherical Particles Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 19

20. Oblate spheroids with large aspect ratios (1:5) as suggested by Kleiber et al. (2009) based on laboratory studies T-Matrix does not converge for entire size distribution Use spheroids for 0.01 < x < 4, and spheres otherwise Sensitivity studies suggest, that the replacement of spheroids with spheres does not significantly alter the results (estimated error < 0.2 mW / (m 2 sr cm -1 )) Influence of Non-Spherical Particles Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 20 C. H. Koehler, Radiative Effect of Mixed Mineral Dust and Biomass Burning Aerosol in the Thermal Infrared, 2014, http://elib.dlr.de

21. External Mixture of Oblate Spheroids (AR 1:5) Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 21

22. Much better fit than spherical particles, although simulated aspect ratios do not match laboratory analysis Remaining deviations around 1100 cm -1 due to sulfates, quartz, orthoclase and illite Possible explanations Illite might require larger aspect ratios (1:18), which cannot be simulated for the given size distribution (T-Matrix method diverges) Sulfates might be modelled inappropriately by ammonium sulfate and gypsum since sea salt aging might result in peak shifts for other sulfates Sodium sulfates have needle like shape with aspect ratio 1:10, but were modelled as spheres (or spheroids with aspect ratio 6:1) Spheroids might be inadequate as well, e.g. due to missing surface roughness Influence of Non-Spherical Particles Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 22

23. Simulate dust signature for pure components (pure quartz, pure kaolinite …) with spheroidal model particles used before Size distribution from in-situ measurements Decompose measured spectrum into linear combination of pure signatures with least-squares fit (neglects interactions between individual components) Simulate aerosol signature of mixture assuming composition obtained from fit for verification purposes Optimization of Mineral Composition and Particle Number Density Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 23

24. External Mixture of Oblate Spheroids (AR 1:5) Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 24

25. Fitted microphysics does not match measured (laboratory) composition: Fitted microphysics similar to results obtained by Boer (2010) for airborne FTIR (ARIES) measurements close to Sal in September 2000 Dust model neither suited for retrieval of dust composition nor concentration Optimization of Mineral Composition and Particle Number Density Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 25 MaterialFitted number concentration [%] Measured number concentration [%] Illite52<10 Kaolinite2230 Montmorillonite11<2 Sea Salt15< 10 Other050

26. Measured spectral signature of mineral dust/biomass burning aerosol mix in the TIR window (800 – 1200 cm -1 ) including uncertainties Confirmed a distinct spectral signature at BOA and TOA and estimated the impact on remote sensing applications (e.g. SST retrieval) Oblate spheroidal model particles are much better suited than spherical particles to model mineral dust aerosols, unless optical depth is small Model based on spheroids not accurate enough to retrieve dust load/composition Further laboratory studies are desirable Summary Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 26 C. H. Koehler, Radiative Effect of Mixed Mineral Dust and Biomass Burning Aerosol in the Thermal Infrared, 2014, http://elib.dlr.de

27. Comparison of Measured and Simulated Spectra Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 27

28. Until very recently, it was generally accepted, that dust particles can be treated as spherical particles in the TIR Very few dedicated field studies addressing TIR radiative closure Laboratory studies indicate, that particle non-sphericity has to be taken into account for synthetic aerosols Deviations from Mie theory are visible in field measurements, too, but incomplete aerosol information inhibits rigorous investigation State of the Art Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 28 Image: K. Kandler

29. Programmer’s Interface to Radiative Transfer Algorithms (PIRATES) Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 29

30. Atmosphere: Tropical day (Remedios et al., 2007) with modified ozone profile CO 2 profile scaled to 380 ppmv Profiles of temperature, pressure and rel. Humidity from IfT radiosondes Well known humidity bias of Vaisala RS92-SGP sondes corrected by algorithm suggested by Wang et al. (2012) or Cady-Pereira et al. (2008) Flat sea surface (Fresnel model) with temperature based on IASI SST product Altitude and vertical extent of aerosol layers from LIDAR measurements Aerosol size distributions from airborne and ground based in-situ measurements Aerosol composition based on laboratory analysis of in-situ samples Refractive indices of mineral components from various literature sources Refractive indices of several components not found in literature, replaced by ‘analogous’ material (e.g. ammonium sulfate instead of sodium sulfate) Radiative Transfer Simulations for SAMUM-2 Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 30

31. The D&P Model 102 FTIR Commercially manufactured by Design and Prototypes (D&P) Intended for field measurements of land surface and rocks Highly portable Designed to withstand environmental conditions in the field Spectral range 650 – 3000 cm -1 Spectral resolution: ≈ 4 cm -1 Turns out to be suitable for atmospheric measurements, too! Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 31 C. H. Koehler, Radiative Effect of Mixed Mineral Dust and Biomass Burning Aerosol in the Thermal Infrared, 2014, http://elib.dlr.de

32. Validation against MAERI Simultaneous, co-located measurements (May 2008, Miami) Similar integration times MAERI data convolved to coarser D&P resolution Generally good agreement High frequency noise due to wavenumber uncertainty of D&P FTIR Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 32

33. Comparison of Different Calibration Methods Convolved to 10 cm -1 to reduce noise Factory calibration not usable for atmospheric application Drift correction and ZPD fit yield considerable improvement in the region 1000 - 1100 cm -1 D&P FTIR suitable for atmospheric measurements Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 33

34. Required to interpret measured spectra Programmer’s Interface to Radiative Transfer Algorithms (PIRATES) Built on existing algorithms Concurrent implementation of Line-by-line gas absorption using efficient multi-grid technique Continuum gas absorption Parameterized cross-sections Mie-engine T-Matrix engine Discrete ordinate solver DISORT Database for non-spherical particles compatible to scatdb (Rother et al.) Database for index of refraction for aerosol components (DIRAC) Radiative Transfer Simulations Claas H. Köhler | IRS 2016 | 2016-04-10DLR.de Chart 34