MAX-DOAS Measurements of Trace Gas and Aerosol Vertical Profiles. Udo Frieß Institute of Environmental Physics University of Heidelberg Germany

Transcription

1 MAX-DOAS Measurements of Trace Gas and Aerosol Vertical Profiles Udo Frieß Institute of Environmental Physics University of Heidelberg Germany

2 Outline MAX-DOAS: The idea Retrieval techniques MAX-DOAS instrumentation Retrieval of trace gas vertical profiles Retrieval of aerosol vertical profiles Summary

3 MAX-DOAS Instrumentation: Telescope 2D scanner (elevation and azimuth) using fused silica prisms allowing to collect light from any direction in the sky Brushless servo motors with position encoder and transmission enabling high accuracy in focusing Achromatic optics with enhanced aluminium coating No polarisation sensitivity due to fibre optics Diffusor plate for direct sun measurements Integrated calibration lamps

4 MAX-DOAS Instrumentation: Spectrometer Unit Three temperature stabilised spectrographs covering the full UV/Vis wavelength range with high spectral resolution Embedded PC allowing fully autonomous measurements and remote control over TCP/IP Indoor operation under stable conditions nm nm MAX nm

5 Multi-Axis DOAS DOAS (Differential Optical Absorption Spectroscopy) of scattered sunlight yields the integrated concentration of trace gases along the atmospheric light path Zenith Sun Stratosphere 45 Zenith-sky measurements: Sensitivity strongly weighted towards the stratosphere 20 Multi-Axis measurements: Spectrograph Troposphere Increasing light path through the troposphere with decreasing elevation angle Trace gas and aerosol vertical profiles can be retrieved using inverse methods (i.e., optimal estimation).

6 Trace Gases detectable by UV/Vis scattered light DOAS O 3 UV (log) SO 2 BrO IO OIO NO 2 I 2 NO 3 O 3 Vis ClO OClO OBrO HCHO Br 2 (CHO) 2 H 2 O HONO O 4 O Wavelength [nm]

7 NO 2 Measurements at Hohenpeißenberg, enberg, Germany Diurnal variation Comparison with in situ measurements

8 NO 2 Measurements at Hohenpeißenberg, enberg, Germany Diurnal variation Wind direction Comparison with in situ measurements

9 Retrieval of NO 2 vertical profiles Cabauw Intercomparison Campaign, June NO 2 Profiles Vis 6.0 Altitude [km] 3 2 NO 2 mixing ratio [ppb] Time [UT]

10 NO 2 surface mixing ratio In situ versus MAX-DOAS 30 Cabauw Intercomparison Campaign NO 2 surface mixing ratio [ppb] EMPA in situ MAX-DOAS Date 2009

11 Retrieval of formaldehyde vertical profiles Cabauw Intercomparison Campaign, June HCHO Profiles Altitude [km] HCHO mixing ratio [ppb] Time [UT]

12 Aerosol retrieval using O 4 absorption MAX-DOAS measurements of a trace gas with a known vertical profile: Contain information on the light path through the atmosphere Allow to gain information on atmospheric aerosols Most suitable trace gas for aerosol retrieval in the UV/Vis is the oxygen collision complex O 4 : Numerous absorption bands, easy to detect with DOAS O 4 concentration proportional to the square of the O 2 concentration Scale height of O 4 profile: ~4km O 4 absorption cross section [arb. units] Absorption cross section of O Wavelength [nm]

13 Sensitivity to observation parameters Elevation α: The atmospheric light path and thus the optical depth of O 4 generally increase with decreasing elevation Information on aerosol extinction profile Wavelength λ: The visibility (average scattering distance along line of sight) and thus the O 4 optical depth generally decreases with decreasing wavelength. Information on wavelength dependence of aerosol extinction (Angstrom coefficient) Further information on aerosol extinction profile Relative azimuth β: Scanning in different azimuth directions yields O 4 optical depth as a function of scattering angle Information on angular dependence of scattering (phase function and single scattering albedo)

14 Sensitivity to observation parameters Elevation α: The atmospheric light path and thus the optical depth of O 4 generally increase with decreasing elevation Information on aerosol extinction profile Wavelength λ: The visibility (average scattering distance along line of sight) and thus the O 4 optical depth generally decreases with decreasing wavelength. Information on wavelength dependence of aerosol extinction (Angstrom coefficient) Further information on aerosol extinction profile Relative azimuth β: Scanning in different azimuth directions yields O 4 optical depth as a function of scattering angle Information on angular dependence of scattering (phase function and single scattering albedo)

15 Sensitivity to observation parameters Elevation α: The atmospheric light path and thus the optical depth of O 4 generally increase with decreasing elevation Information on aerosol extinction profile Wavelength λ: The visibility (average scattering distance along line of sight) and thus the O 4 optical depth generally decreases with decreasing wavelength. Information on wavelength dependence of aerosol extinction (Angstrom coefficient) Further information on aerosol extinction profile Relative azimuth β: Scanning in different azimuth directions yields O 4 optical depth as a function of scattering angle Information on angular dependence of scattering (phase function and single scattering albedo)

16 Comparisons with Raman Lidar Intercomparison measurements in Cabauw, May 2008 Lidar data courtesy of A. Apitouley, RIVM Raman lidar: diurnal variation of the range corrected signal MAX-DOAS: diurnal variation of the aerosol extinction profile MAX

17 Comparisons with Raman Lidar Intercomparison measurements in Cabauw, May 2008 Lidar data courtesy of A. Apitouley, RIVM Raman lidar: diurnal variation of the range corrected signal MAX-DOAS: diurnal variation of the aerosol extinction profile MAX

18 6. Comparisons: : Sun Photometer Time (UTC) 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17: Sun Photometer DOAS Intercomparison measurements in Cabauw, May 2008 AOD at 550 nm AOD at 550 nm Sun photometer data courtesy of B. Henzing, TNO :00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 08:00 09:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 Time (UTC)

19 6. Comparisons: Raman Lidar 09:30-10:00 UTC of :30-11:00 UTC of a priori MAX-DOAS retrieved MAX-DOAS retrieved Raman lidar a priori MAX-DOAS retrieved MAX-DOAS retrieved Raman lidar Altitude (km) Altitude (km) Aerosol extinction (km -1 ) Altitude (km) :30-13:00 UTC of a priori MAX-DOAS retrieved MAX-DOAS retrieved Raman lidar Aerosol extinction (km -1 ) Intercomparison measurements in Melpitz, June Aerosol extinction (km -1 ) Lidar data courtesy of D. Althausen, IFT

20 Summary Multi-Axis DOAS measurements allow for retrieving trace gas and aerosol vertical profiles and optical properties Typical vertical resolution of m resolve the structure of the boundary layer Measurements can be performed with simple and cost effective fully automated instrumentation Inherently self- calibrating Simultaneous measurement of aerosols and numerous trace gases in the entire UV/Vis range MAX-DOAS is an important tool for satellite validation Potential for integration in world wide remote sensing networks

21 Parameters Affecting DOAS Measurements ( ) σ ( λ ) ρ ( s) + kr ( s) + km ( s) ds Lambert-Beer law: I( λ) = I0 ( λ) e Light path through the atmosphere and trace gas absorption are determined by: Viewing geometry (SZA, elevation, azimuth) Wavelength (dependency of light path and extinction on λ) Aerosol extinction Trace gas profile... DOAS measurements contain (indirect) information on the atmospheric state (e.g., trace gas an aerosol profile) Established method for the retrieval of atmospheric parameters: Optimal Estimation

22 MAX-DOAS: The Idea Spectrally resolved observations of scattered sunlight in the UV/Vis along different lines of sight Detection of various trace gases (NO 2, BrO, HCHO, ) by identifying their individual absorption features Analysis of spectra based on the Lambert- Beer law: I( λ) = I I 0 (λ), I(λ): σ(λ): ρ(s): k r (s), k m (s): ( λ e 0 ) incident and transmitted intensity absorption cross section trace gas concentration Rayleigh and Mie extinction coefficients Basic quantity measured by DOAS is the slant column density (SCD) of an absorber, i.e. the integrated concentration along the light path: S = ρ( s) ds ( σ ( λ ) ρ ( s) + k ( s) + k ( s) ) Instrument Zenith Problem: Length of light path is difficult to determine, requires radiative transfer modelling r m ds Sun Θ Stratosphere α=45 Boundary layer α=20 α=10 α=5 α=2

23 Remote sensing of atmospheric trace gases: Differential Optical Absorption Spectroscopy (DOAS) When sampling the light intensity on a discrete wavelength grid λ k (and neglecting the pressure and temperature dependence of the absorption cross section), the Lambert- Beer law can be solved numerically by minimising n ln I( λk ) ln I0( λk ) + σ i ( λk Si + cn λ k 2 χ ) k i n to determine the integrated concentrations along the light path (Slant Column Density, SCD): The Optical Density is defined as S i L ρ ( s) 0 The polynomial Σc n λ kn removes the broad-banded σ ( λ ) 4.5 A spectral structure caused by Rayleigh- and Miescattering. Thus only compounds with high frequent 0.8 σ '( λ ) = σ ( λ ) - σ b ( λ ) 0.4 absorption features can be detected. The high frequent parts of σ and τ are referred to as the differential absorption cross section and optical -0.6 B -0.8 density σ and τ i ds I( λ) τ ( λ) σ ( λ) S = ln I0( λ) σ [10-19 cm 2 ] σ ' [10-19 cm 2 ] σ b ( λ ) λ [nm]

24 Retrieval of trace gas and aerosol vertical profiles: Inverse Modelling Forward modelling Forward model F Optimal Estimation Atmospheric state [Rodgers, 1990] Simulated Measurement x (Radiative transfer model) The Maximum A Posteriori (MAP) solution xˆ is determined y = F(x) iteratively by minimising 2 χ = T 1 T 1 [ y F( xˆ) ] S [ y F( xˆ) ] + [ xˆ x ] S [ xˆ x ] ε Deviation of Inverse modelling modelled from actual measurement a a a Deviation of state vector from a priori A priori state vector x a, S a Measurement y Measurement error S ε Inverse model (based on F) Estimate for atmospheric state xˆ Error of state vector Ŝ

25 Retrieval of trace gas vertical profiles Forward modelling Atmospheric state x Forward model F (Radiative transfer model) Simulated Measurement y = F(x) The measurement vector The state vector y S( α 1 ) = M S( αm ) Inverse Trace modelling gas SCDs at different x = elevation M angles α A priori state vector x a, S a ρ( z 1 ) ρ( z n ) Trace gas vertical profile Measurement y Measurement error S ε Inverse model (based on F) Estimate for atmospheric state xˆ Error of state vector Ŝ

26 Retrieval of aerosol vertical profiles Atmospheric state x y The measurement Forward vector modelling The state vector τ O4( λ1, α1) Forward k( z1) model F M (SCIATRAN) τ O4( λm, α ) m = k( z n ) I( λ1, α1) I0( λ1) x = q1 Inverse Relative modelling intensity M properties M I( λm, αm ) I0( λm ) qr A priori state vector x a, S a O 4 optical depth Simulated Measurement Aerosol y = extinction M F(x) profile Aerosol optical - phase function - single scattering albedo - size distribution Measurement y Measurement error S ε Inverse model (based on F) Estimate for atmospheric state xˆ Error of state vector Ŝ

27 MAX-DOAS Instrumentation for long-term Measurements: Requirements Large wavelength range to cover many trace gases Sufficient spectral resolution ( nm) High mechanical stability of spectrograph unit Indoor operation, temperature stabilisation High detector sensitivity to achieve low detection limit Flexible telescope unit to observe light from any direction in the sky Direct sun- and moonlight capability Fully autonomous operation Self-calibration capabilities

28 Newly developed MAX-DOAS Instrumentation Spectrometer unit Telescope unit

29 NO 2 Measurements at Hohenpeißenberg, enberg, Germany Hohenpeißenberg

30 NO 2 Measurements at Hohenpeißenberg, enberg, Germany Measured vs. modelled NO 2 SCDs

31 Cabauw Intercomparison Campaign of Nitrogen Dioxide measuring Instruments CINDI

32 Comparison of NO 2 Profiles during CINDI

33 NO 2 profiling Averaging Kernels 4.0 Altitude [km] Altitude [km] Averaging Kernel

34 Aerosol retrieval: Comparison of measured and modelled O 4 dscd and intensity Time (UTC) of O 4 optical density measured retrieved Elevation O 4 optical depth, 360 nm O 4 optical depth, 477 nm O 4 optical depth, 577 nm intensity, 360 nm intensity, 577 nm intensity, 477 nm Intensity (a.u.) Comparison of measured and retrieved O 4 optical depths and intensities, for the in Cabauw Time (UTC) of

35 Aerosol Retrieval Results Aerosolextinction (km -1 ) :00 UTC 09:00 UTC 11:00 UTC 13:00 UTC 15:00 UTC 17:00 UTC 4 3 retrieved a priori 3 Altitude (km) Altitude (km) Altitude (km) Single retrieved aerosol extinction profiles and corresponding averaging kernels, for the in Cabauw Averaging Kernel 0

36 6. Comparisons: : Sun Photometer AOD from DOAS Aerosol optical depth in May % 30% 15% 30% AOD from Sun Photometer 1:1 15% 45% Intercomparison measurements in Cabauw, May 2008 AOD from MAX-DOAS measurements are underestimated by ~15% compared to Sun Photometer values