Aircraft and Ground-based Measurements of Spectral Solar Radiation K. Sebastian Schmidt, P. Pilewskie, O. Coddington, B. Kindel University of Colorado, Laboratory for Atmospheric and Space Physics, Boulder, CO Focus: Cloud-aerosol spectral radiative effects SORCE Science Meeting, Annapolis, September 19, 2012 Aircraft and Ground-based Measurements of Spectral Solar Radiation …fill in the gaps in understanding the radiative effects of clouds and aerosols (no satellite instrument looks at Earth like SORCE at the Sun!) …offer a different perspective of clouds and aerosols (“within” and “below”) Aircraft and Ground-based Measurements of Spectral Solar Radiation …fill in the gaps in understanding the radiative effects of clouds and aerosols (no satellite instrument looks at Earth like SORCE at the Sun!) …offer a different perspective of clouds and aerosols (“within” and “below”) Summary: • Spectral cloud-aerosol property retrieval methods are augmenting current techniques • Examine issues in cloud-aerosol ‘radiative budget’ by zooming in (out) spectrally • Caution: 3D effects mess with the spectrum – does information content analysis help? Why spectral observations of clouds? …Because they carry information… (1) …about the adequacy of model assumptions (e.g., inhomogeneites; crystal shape; thermodynamic phase; aerosols; ...) through spectral consistency (1) ...that can be exploited with ‘cloud spectroscopy’ • McBride et al., ACP [2011] use slopes of normalized transmittance to retrieve optical thickness and effective radius from radiance Caution • Kokhanovsky et al. [2011]: derive vertical profile of effective radius King and Vaughan [2012]: assess uncertainty requirements (1-2%) • Ehrlich et al., ACP [2009]: retrieve properties of mixed-phase clouds • Spectral approaches now under development for aerosol-cloud scenes • Thermodynamic phase and ice crystal habit information 3D (2) ...about the radiance irradiance conversion (deriving forcing terms from space) • CERES-derived ADMs are broadband, but 3D effects introduces spectral biases • Combined aerosol-cloud forcing cannot be determined with current remote sensing approach • Satellite-derived surface forcing problematic under ‘messy’ cloud conditions Why spectral observations of clouds? …Because they carry information… or instrument issues (1) …about the adequacy of model assumptions (e.g., inhomogeneites; crystal shape; thermodynamic phase; aerosols; ...) through spectral consistency {R860 nm ; R1600 nm} { ; reff } {h ; } gas profiles {R350 nm – R3800 nm} Kindel et al., JGR, 2010 Why spectral observations of clouds? …Because they carry information… (1) the adequacy of model assumptions (e.g., inhomogeneites; crystal shape; (F…about net (top) – Fnet (bottom)) / F (top) * 100% Regularly observed: thermodynamic phase; aerosols; ...) through spectral consistency slope is only reproduced by 3D RTM Non-zero apparent absorption (1) ...that can be exploited with ‘cloud spectroscopy’with spectral slope. • McBride et al., ACP [2011] use slopes of normalized transmittance Mechanism: net horizontal to retrieve optical thickness and effective radius from radiance Caution photon transport and molecular • Kokhanovsky et al. [2011]: derive vertical profile of effective radius scattering King and Vaughan [2012]: assess uncertainty requirements (1-2%) Can be reproducedclouds by 3D • Ehrlich et al., ACP [2009]: retrieve properties ofonly mixed-phase RTM (Schmidt et al., 2010). • Spectral approaches now under development for aerosol-cloud scenes • Thermodynamic phase and ice crystal habit information 3D (2) ...about the radiance irradiance conversion (deriving forcing terms from space) • CERES-derived ADMs are broadband, but 3D effects introduces spectral biases • Combined aerosol-cloud forcing cannot be determined with current remote sensing approach • Satellite-derived surface forcing problematic under ‘messy’ cloud conditions Schmidt et al., JGR, 2010; Kindel et al., JGR, 2010; Schmidt and Pilewskie, Light Scattering Reviews, 2011 Why spectral observations of clouds? …Because they carry information… (1) …about the adequacy of model assumptions (e.g., inhomogeneites; crystal shape; thermodynamic phase; aerosols; ...) through spectral consistency (1) ...that can be exploited with ‘cloud spectroscopy’ ƒ (cld)+ƒ (aer) λ λ • McBride et al., ACP [2011] use slopes of normalized transmittance ≠ to retrieve optical thickness and effective radius from radiance ƒλ (cld+aer) Caution • Kokhanovsky et al. [2011]: derive vertical profile of effective radius King and Vaughan [2012]: assess uncertainty requirements (1-2%) Schmidt et al. (2009) • Ehrlich et al., ACP [2009]: retrieve properties of mixed-phase clouds • Spectral approaches now under development for aerosol-cloud scenes • Thermodynamic phase and ice crystal habit information 3D (2) ...about the radiance irradiance conversion (deriving forcing terms from space) • CERES-derived ADMs are broadband, but 3D effects introduces spectral biases • Combined aerosol-cloud forcing cannot be determined with current remote sensing approach • Satellite-derived surface forcing problematic under ‘messy’ cloud conditions Schmidt et al., 2009; Schmidt and Pilewskie, Light Scattering Reviews, 2011 Why spectral observations of clouds? …Because they carry information… (1) …about the adequacy of model assumptions (e.g., inhomogeneites; crystal shape; Caution (2) ...that can be exploited with ‘cloud spectroscopy’ • McBride et al., ACP [2011] use slopes of normalized transmittance to retrieve optical thickness and effective radius from radiance • Kokhanovsky et al. [2011]: derive vertical profile of effective radius King and Vaughan [2012]: assess uncertainty requirements (1-2%) • Ehrlich et al., ACP [2009]: retrieve properties of mixed-phase clouds • Spectral approaches now under development for aerosol-cloud scenes • Thermodynamic phase and ice crystal habit information 3D (3) ...about the radiance irradiance conversion (deriving forcing terms from space) • CERES-derived ADMs are broadband, but 3D effects introduces spectral biases • Combined aerosol-cloud forcing cannot be determined with current remote sensing approach • Satellite-derived surface forcing problematic under ‘messy’ cloud conditions McBride et al., JGR, 2012 Why spectral observations of clouds? R T Why spectral observations of clouds/aerosols? R T T R Why spectral observations of clouds/aerosols? Realistic water clouds (LES, G. Feingold) cloud top height variability within box? reflectance Caution Y [km] 3D spectrum from 1D calculations for cloud optical thickness and effective radius retrieved at two MODIS wavelengths cloud edge effect? spectrum from 3D calculations,0.5 km box (0.1 km resolution) effective radius variability within box retrieval wavelengths X [km] MCARaTS (Monte Carlo Atmospheric Radiative Transfer Simulator), Iwabuchi 2006 wavelength [nm] reflectance difference (residual) Spectral consistency / spectral residual zero at retrieval wavelengths spectrum from 3D calculations Caution – 3D zero at all wavelengths would be expected for a homogeneous cloud spectrum from 1D calculations for cloud optical thickness and effective radius retrieved at two MODIS wavelengths Spectral 3D effects may be complex! wavelength [nm] Multiple mechanisms possible; Rayleigh bluing near clouds (Marshak/Wen) is one of them. Spectral Residual (albedo difference) Alebdo The spectral residual carries information 1) Observe reflectance (albedo) spectrum from ice cloud (droxtals) 2) Retrieve optical thickness and effective radius, (wrongly) assuming liquid water 3) Based on retrieval, calculate full spectrum and subtract from observation (1) 4) Non-zero residual at non-retrieval wavelengths indicates phase was wrong! retrieval wavelengths retrieval wavelengths wavelength [nm] wavelength [nm] Spectral Residual (albedo difference ice cloud – water cloud) The spectral residual carries information 1) Observe reflectance (albedo) spectrum from ice cloud (droxtals) 2) Retrieve optical thickness and effective radius, (wrongly) assuming liquid water 3) Based on retrieval, calculate full spectrum and subtract from observation (1) 4) Non-zero residual at non-retrieval wavelengths indicates phase was wrong! retrieval wavelengths wavelength [nm] The spectral residual carries information Residual Spectral [%][%] residual Spectral 1)1)Observe Observereflectance reflectance(albedo) (albedo)spectrum spectrumfrom fromice icecloud cloud(droxtals) (droxtals) 2)2)Retrieve Retrieveoptical opticalthickness thicknessand andeffective effectiveradius, radius,(wrongly) (wrongly)assuming assumingcolumns liquid water 3)3)Based Basedon onretrieval, retrieval,calculate calculatefull fullspectrum spectrumand andsubtract subtractfrom fromobservation observation(1) (1) 4)4)Non-zero Non-zeroresidual residualatatnon-retrieval non-retrievalwavelengths wavelengthsindicates indicateshabit phase was waswrong. wrong! retrieval wavelengths If we did not need to worry about spectral 3D effects, we could use subtle differences in the residual spectrum to detect, e.g., thermodynamic phase or even verify whether the correct habit was used in forward calculations. wavelength [nm] Information Content Use Bayes’ theorem to sequentially update state vector x (set of cloud parameters we are interested in) with information from spectral channels [Coddington et al., 2012]. Gain in information from prior to posterior state vector is seen as narrowing of the probability density function of cloud-aerosol parameters x, and the entropy S with respect to x is calculated at each step: S(x) p(x)log p(x) 2 The gain in information at each H S prior S posterior consecutive step is measured as Question: To what extent can 3D effects cause spectral contrasts that lead to retrieval biases? Answer: Implement known 3D effects as additional model error. probability ice Probability liquid Assume ice cloud with τ=10 and effective radius 20 micron. Use a measurement uncertainty of 3%, model error 2%. wavelength [nm] 2.25 μm channel Aircraft and Ground-based Measurements of Spectral Solar Radiation …fill in the gaps in understanding the radiative effects of clouds and aerosols (no satellite instrument looks at Earth like SORCE at the Sun!) …offer a different perspective of clouds and aerosols (“within” and “below”) Summary: • Spectral cloud-aerosol property retrieval methods are augmenting current techniques • Examine issues in cloud-aerosol ‘radiative budget’ by zooming in (out) spectrally • Caution: 3D effects mess with the spectrum – does information content analysis help?