Multiangular radiances of an isolated convective cloud comparison

Multi-angular radiances of an isolated convective cloud: comparison between MISR measurement and Monte-Carlo simulation C. Cornet(1), R. Davies(2) (1) Laboratoire d’Optique Atmosphérique, Université de Lille, France (2) Department of physics , The university of Auckland, New Zealand 1. (1, 2) initially at the Jet Propulsion Laboratory, Pasadena, CA

Why studying deep convective cloud ? Deep convective cloud = high vertical extent large water content Satellite retrieval assumes an independent homogeneous cloud pixel = 1 D retrieval limitations in case of deep convective cloud: • often strong heterogeneity effects : - subpixel cloud variability - smoothing of the radiance field - brightness and shadowing effects 1 D retrieval leads to errors in the optical thickness determination • important vertical extent and large water content: - lost of energy by the cloud sides - asymptotic limit of radiances 1 D retrieval tends to underestimate the optical thickness

MISR: Multi-angle Imaging Spectro-Radiometer 9 view angles at Earth surface: 70. 5º, 60º, 45. 6º, 26. 1º forward/ aftward and 0º Multiple spectral bands at each angle: MISR/Terra: 446, 558, 672, 866 nm 400 -km swath: 9 -day global coverage 275 m - 1. 1 km sampling 7 minutes to observe each scene at all 9 angles

Deep convective cloud and MISR advantages: 9 cameras + high resolution (275 m) observation of the cloud sides and top September, 2 th, 2003 Pacific ocean: lat ~ 7°; lon ~ 180° • reconstruct cloud contour • compute radiances along the cloud to match MISR measurement

Cloud reconstruction Obtaining cloud top height and cloud position 1) Distance between a features seen with 2 cameras: 2) ∆D 2) Cloud top height calculation: CTH = ∆D / [tan ( 2 - 1)] (Xc, Yc) CTH 3) Parallax correction : Xc = Xg + CTH * tan( ) * sin ( ) Yc = Yg + CTH * tan( ) * cos ( ) D (Xg, Yg)

Cloud reconstruction Obtaining cloud top height and cloud position

Cloud reconstruction Comparison between stereographic and manual results Height ~ 8 km Thickness ~ 8 km Seiz and Davies, Rem. Sens. Env. , 2005

Cloud reconstruction Cloud envelop approximation Height ~ 8 km Thickness ~ 8 km

Mapping the radiances along the cloud Use camera direction to know radiances going out from the cloud Height ~ 8 km Thickness ~ 8 km

Mapping the radiances along the cloud

Radiances simulation Use a forward Monte-Carlo model based on : 1 ) the storage of the photon collision density 2 ) the integration along each line of sight increase the rapidity of the calculations Forward camera Geometrical configuration: sun: s= 22º s= 98 camera AN: = 0º camera AF, AA: = 22. 1º camera BF, BA: = 45. 6 camera CF, CA: = 60º camera DF, DA: = 70. 5º Forward: 160 º Aft: 340 º DF CF BF AF AN AA BA CA DA Aft camera 0º

Clear sky: MISR measurements Use MISR cloud mask for 200 x 200 km around the cloud: ~ 100 pixels s ~ 22º s ~ 278º

Clear sky: Monte-Carlo Simulation Wind speed = 2. 5 m. s-1 (Cox and Munk model, 1955) Raleigh molecular scattering: R=0. 042 Aerosol optical thickness: a=0. 09 Aerosol phase function: mixture between sea-salt and sulfate s ~ 22º s ~ 262º

Deep convective cloud : Monte-Carlo simulation - account for the cloud morphology + constant extinction coefficient (2, 5, 10 km-1) Extinction coefficient Radiances closest to the perpendicular of the cloud surface

Simulation with a constant extinction coefficient Radiances along the cloud for the 9 MISR camera

Simulation with acloud constant extinction coefficient Deep convective : Monte-Carlo simulation MISR radiances along the cloud Monte-Carlo simulation: cloud morphology + ext=10 km-1 + C 1 Phase function

Sensitivity to the phase function MODIS phase = mixed or uncertain + cloud top ≈ 8 km --> test phase function effects • IHM (Inhomogeneous Hexagonal Monocrystal; Labonnote et al. , JGR, 2001)�

Sensitivity to the phase function MISR radiances along the cloud Monte-Carlo simulation: cloud morphology + ext=5 km-1 + IHM phase function for z >= 6 km

Simulation with a variable extinction coefficient Use differences between measured and simulated An radiances to adjust horizontally the extinction coefficient.

Simulation with a variable extinction coefficient MISR radiances along the cloud Monte-Carlo simulation: cloud morphology + horizontally variable extinction � coefficient

Comparison between - radiances accounting for the cloud morphology (3 D) - radiances of an infinite homogeneous cloud (1 D) Simulation of the 1 D radiances for the 9 MISR cameras for different optical thickness

Comparison between 1 D/3 D radiances 1 D radiances + MISR measurement 1 D retrieval (4 km) 5 -10 ; (9 km) 10 -20; (11 km) 30 �

Comparison between 1 D/3 D radiances 1 D radiances + MISR measurement + 3 D simulation 1 D/3 D retrieval (4 km) 5 -10/54 ; (9 km) 10 -20/13; (11 km) 30/158 �

Sensitivity along the cloud sides ? 1 D radiances + MISR measurement + 3 D simulation Sensitivity of the radiances along the cloud sides is low

Retrieved 1 D/3 D optical thickness 1 D optical thickness 3 D optical thickness

Conclusion Thanks to MISR : - reconstruction of the contour of an isolated deep convective cloud - angular distribution of the radiances along the cloud surface Deep convective cloud: - Accounting for cloud morphology Higher optical thickness than the 1 D retrieval possible underestimation of the LWC - Low sensitivity for the side difficulty to retrieve the vertical cloud properties (for this configuration) Possible future: -� Reconstruction of a complete 3 D cloud - Use the angular distribution to develop of a method to retrieve cloud properties accounting for the morphology

Deep convective cloud : collision density cloud morphology + constant extinction (5 km-1) Sun=22º almost perpendicular to the cloud profile