Our goals are to 1) understand fundamental aspects of radiative transfer in realistic clouds, 2) determine the size of 3D radiative effects in actual cloud fields, and 3) learn how to approximate 3D radiative transfer with simple (computationally efficient) models. We are interested in issues important for climate modeling, such as accurate representation of domain averaged albedo and heating rates, and remote sensing issues, such as the spatial variability of reflectances. This work was supported by a DOE Atmospheric Radiation Measurement (ARM) program funded project in which Warren Wiscombe was the PI.
I have developed a stochastic cloud field algorithm to generate 3D fields from cloud profiles obtained from vertically pointing radar. I am using these simulations to determine the size of the 3D radiative transfer effect in actual cumulus fields and the adequacy of existing simplified radiative transfer approaches, such as the independent pixel approximation. The long version of a talk given at EGS 2003 (PDF) briefly describes the "data generalization model" and gives radiative transfer results from three months of Nauru boundary layer cumulus fields. A paper (PDF) published in the "Clouds and Radiation" special issue of Atmospheric Research describes the stochastic algorithm and tests of its faithfulness in generating cloud fields with the input field statistics.
Robert Pincus led an effort to determine how well the three dimensional radiative transfer effect can be estimated by time-height cross sections from a "perfect" vertical cloud profiler. An LES model produced fair weather cumulus fields which were sampled at a single point over time to produce 2D fields. The Monte Carlo model was used to calculate the 3D radiative effect (defined here as the difference between 3D and IPA domain average solar fluxes at the surface) in the radar sampled fields, 2D slices crossing the radar along the wind, all 2D slices, and in the 3D fields. By comparing the 3D radiative effect in these four types of sampled clouds we are able to determine the errors due to the frozen turbulence assumption (converting time to horizontal distance), cloud sampling (the radar samples only a small portion of the cloud field), and dimensionality (radiative transfer in 2D clouds versus 3D clouds). Averaging over time, the mean errors in the 3D radiative effect due to sampling and the frozen turbulence assumption are near zero, but the dimensionality error is significant (10\% error for overhead sun). A paper (1380 kB PDF) on this work has been published (Pincus, R., C. Hannay, and K. F. Evans, 2005: The accuracy of determining three-dimensional radiative transfer effects in cumulus clouds using ground-based profiling instruments. J. Atmos. Sci., 62, 2284-2293.)
I was also interested in developing simpler stochastic cloud models based on simulating the cloud optical depth and thickness. One issue to address first is how well these two 2D fields represent the 3D radiative transfer effects in cumulus clouds. This preliminary work was presented in June 2002 at the AMS 11th Conference on Atmospheric Radiation with a poster (GIF) and a conference paper (PDF)
One application of the data generalization stochastic cloud model was a preliminary study of angular radiance closure. An angular radiance closure study statistically compares the observed and simulated angular distribution of sky radiance. Sky radiance observations from the Whole Sky Imager of boundary layer cumulus during Nauru99 were compared to radiance fields simulated from stochastic cumulus cloud fields generated from cloud structure statistics obtained from radar. This study was presented in a poster (PDF) at the 2002 ARM science team meeting in St. Petersburg, Florida.
We used MMCR millimeter-wave radar and MWR microwave radiometer measurements from Nauru (Tropical West Pacific) to learn about cloud structure. We found that existing cloud retrieval methods worked poorly for cumulus clouds at Nauru. Hence, we developed a new Bayesian algorithm to retrieve cloud liquid water content and effective radius profiles from the MMCR and MWR. This method uses a priori information about cloud physics from in situ observations or LES microphysical modeling. The inputs to the algorithm are radiosonde profiles, cloud location and radar reflectivity from MMCR data, and MWR brightness temperatures at two channels. The outputs are liquid water path, optical depth, and vertical profiles of effective radius and liquid water content, as well as error bars on all retrieved quantities. The algorithm has been developed for liquid water clouds at the ARM site on Nauru. We have tested the retrieval algorithm using cloud fields generated by a Large Eddy Simulation model, and we have run the retrieval on 3 months of data from Nauru. A paper about the algorithm has been published (McFarlane, S. A., K. F. Evans, A. S. Ackerman, 2002: A Bayesian Algorithm for the Retrieval of Liquid Water Cloud Properties from Microwave Radiometer and Millimeter Radar Data. J. Geophys. Res., 107(D16), 10.1029/2001JD001011). Download JGR PDF
Sally McFarlane performed a solar flux closure experiment in the
Tropical Western Pacific. She used data from the ARM site on the
Republic of Nauru to perform closure experiments on broadband shortwave
fluxes. The data consisted of atmospheric profiles from radiosondes,
brightness temperatures from a microwave radiometer, cloud location and
radar reflectivity from the Millimeter Wavelength Cloud Radar, and solar
fluxes from a suite of ground-based radiometers. Radiative transfer
model predictions of the direct, diffuse, and total surface flux were
compared to the observed fluxes. Statistical comparisons were made
between the modeled and observed fluxes at one minute intervals over
about 10 months in 1999 and 2000. The scientific focus is on cloud
remote sensing issues, i.e. the uncertainties of various remote sensing
methods and the representativeness of clouds at Nauru as measured by
shortwave fluxes.
This work has been completed and written up in Sally's Ph.D. thesis
and two published journal articles:
McFarlane, S.A. and K. F. Evans, 2004: Clouds and shortwave fluxes at
Nauru. Part I: Retrieved cloud properties. J. Atmos. Sci., 61, 733-744
(download PDF).
McFarlane, S.A. and K. F. Evans, 2004: Clouds and shortwave fluxes at
Nauru. Part II: shortwave flux closure. J. Atmos. Sci., 61, 2602-2615.
(download PDF).
I was asked to give a talk at the 2001 ARM Cloud Properties Working Group meeting on the observational requirements for 3D radiative transfer models.
Frank Evans and Tim Benner studied 3D solar radiative transfer effects in cloud structure from derived from MODIS Airborne Simulator visible and thermal infrared imagery obtained during TOGA/COARE and CEPEX (1993) in the tropical Pacific. We focussed on quantifying the 3D radiative transfer effects on domain averaged, broadband solar fluxes in a statistically representative sample of tropical small cumulus fields. We found that averaged over all scenes, the 3D effects in these cloud fields are quite small (1 W/m2 for daytime average). A poster presented at the 1999 AMS Radiation Conference describes the results. This work is described in a JGR paper in PDF format (Benner, T. C., and K. F. Evans, 2001: Three-dimensional Solar Radiative Transfer in Small Tropical Cumulus Fields Derived from High-Resolution Imagery. J. Geophys. Res., 106, 14975-14984.)
A project carried out by Paquita Zuidema and Frank Evans used
overcast stratocumulus 2D cloud fields derived from the NOAA/ETL 35 GHz
radar during ASTEX. This was the first time observed cloud top
variability had been considered in radiative modeling of stratocumulus.
We found that the Independent Pixel Approximation (IPA) worked quite
well for domain averaged visible albedo, but not for nadir radiances,
such as might be observed with Landsat. We characterized the spatial
variability of the simulated 1D radiance fields and discovered that
while radiative diffusion smooths the reflectance field, the geometric
effects of cloud top topography shadowing caused a roughening of the
reflectance field for low sun angles. A
compressed Postscript paper describing this work is available
(Zuidema, P. and K. F. Evans, 1998: On the Validity of the Independent
Pixel Approximation for Boundary Layer Clouds Observed during ASTEX.
J. Geophys. Res., 103, 6059-6074).
Examples of the internal 2D radiation
fields for the 9 clouds show the differences in radiation flow
between 2D and IPA radiative transfer.
An example exploring 3D radiative transfer modeling in an LES stratocumulus cloud was presented as a poster at the International Radiation Symposium (1996) in Fairbanks, Alaska. A Postscript version of the conference paper is available.
