UCLA Vector Radiative Transfer Models for Application to Satellite Data Assimilation K. N. Liou, S. C. Ou, Y. Takano and Q. Yue Department of Atmospheric and Oceanic Sciences and Institute of Radiation and Remote Sensing University of California, Los Angeles  Vector delta-four-stream method coupled with the adding principle (D4S/A)  Comparison of radiance derivatives from analytical and finite difference methods  Analysis and interpretation of AIRS thin cirrus data  Comparison of the AIRS inferred cloud properties with lidar/radar measurements

Scientific Objectives  Formulate and compute analytical radiance derivative terms based on D4S/A and compare results with those from analytical (D/A, D2S/A) and finite-difference methods. Examine their sensitivity to cloud optical depth, particle size, cloud top temperature, and water vapor absorption.  Construct a fast thin cirrus radiation model by combining the operational OPTRAN model, developed for the speedy calculation of transmittances in clear atmospheres, and a thin cirrus cloud parameterization, using a number of observed ice crystal size and shape distributions.  Apply the parameterization model to the AIRS/Aqua data over and near the ARM-TWP Manus Island site to interpret thin cirrus thermal IR window spectra. Infer the cirrus optical depth and ice crystal size and shape from the observed AIRS spectra using a  2 -minimization program. Compare AIRS-inferred cloud parameters with those from collocated ground-based radar/lidar measurements.

UCLA (Vector) Radiative Transfer Flow Diagram For Application to Satellite Data Assimilation

Evaluation of Radiance Derivative with Respect to   For thermal infrared and microwave radiation, the radiance derivative can be expressed from the D4S/adding method as follows: where the boldface parameters are obtained from the D4S method using the source function technique.  This expression can be used to obtain the radiance derivative with respect to cloud water content, which is a prognostic variable in the NCEP’s GFSA model.

Sensitivity of Radiance Derivatives With Respect to Cloud Optical Depth

Microwave T B,I and T B,Q from Doubling, D4S/A and D2SA Methods as a Function of the Viewing Zenith Angle

Microwave ∂T B,I / ∂  and ∂T B,Q / ∂  from Doubling, D4S and D2S Methods for Clear Sky

Microwave ∂T B,I / ∂  and ∂T B,Q / ∂  from Doubling, D4S and D2S Methods for Cirrus Cloud with  = 1

OPTRAN Input Files: Atmosphere profile Surface condition Surface optical property Viewing geometry Clear radiance I 0 IR database of single scattering Properties of individual non- spherical ice particles Ice crystal SD and HD models Bulk radiative properties of ice cloud (D e,, , Q ext,IR,g) Cloud top temperature T c Cloud IR optical depth Cloud emissivity Cloudy radiance A Fast Thin Cirrus Radiative Transfer Model

Single-Scattering Properties for Tropical Ice Crystal Size Distribution

Sensitivity of Thermal Infrared BT Spectra to Optical Depth and Ice Crystal Habit and Size  BT spectra are sensitive to cirrus optical depth and particle size.  Sensitivity to mean effective size increases as optical depth increases.  Different HDs are used, indicating the sensitivity to the ice crystal habit.

Test Case : , 1553 UTC (Manus Island) Selected region over Manus Island: Lon: 1.0 ° ~3.0°S Lat: 147°~149°E 15 cloudy pixels were selected. Only MMCR was operating on this date. No lidar data was available. Near the AIRS overpass, a thin layer of cirrus clouds existed between 9 and 12 km. Time (UTC) MMCR Reflectivity AIRS Overpass Height (km) MODIS

Comparison of Computed and Observed AIRS Thin Cirrus BT Spectra Computed and observed AIRS thin cirrus BT spectra agree closely. RMS residuals are within ±0.5 K for window channels in the 750–1000 cm -1 region. Differences in the 1050–1130 cm -1 region are larger but still within 1 K for most of the wavenumbers. The largest residuals occur in the ozone band.

Comparison of Cirrus Parameters Inferred from AIRS and MMCR Retrievals The AIRS-inferred optical depths are larger than the MMCR-retrieved values. The AIRS-inferred ice crystal sizes are smaller than the MMCR-retrieved values MMCR generally misses small particles. These differences could be due missing small particles from MMCR (Comstock et al., 2002), thus underestimating extinction coefficients. For pixel 14, MMCR completely missed the thin cirrus layer, so τ~0 and D e ~0. (=½ D e ) AIRS overpass Pixels τ MMCR ~0 AIRS D e (μm) MMCR ~0 AIRS

Summary and Future Work  We have formulated an analytical method to evaluate the infrared and microwave radiance derivatives with respect to optical depth on the basis of adding principle, and investigated the sensitivity of thermal IR and microwave radiance derivatives to variation in cloud optical properties. Formulate Radiance Derivatives with respect to other cloud parameters  Computed AIRS thin cirrus BT spectra based on AIRS-inferred thin cirrus cloud parameters agree with observed BT spectra closely. Updated Ice Crystal HD and SD Models  Comparison of the AIRS-inferred thin cirrus cloud parameters and the MMCR retrieved values over the ARM-TWP site are consistent. Simultaneous MPL and MMCR Case Study