Assessing the Interpretation and Handling of Clouds by
              1D Radiative Transfer Algorithms.
                   Part I: Solar Radiation
 

                      Howard W. Barker
  Cloud Physics Research Division, AES, Downsview, ON, Canada

             Graeme L. Stephens, Philip Partain
  Department of Atmospheric Science, CSU, Ft. Collins, CO, USA

                      Gerald L. Potter
    PCMDI, Lawrence Livermore Nat'l. Lab, Livermore, CA, USA
 
 

1. Introduction and objectives

All radiative transfer algorithms used in large-scale atmospheric
models (LSAMs) assume that clouds are homogeneous at unresolved
scales. This is disconcerting for over the past two decades,
several studies have demonstrated that when fluctuations in cloud
structure for scales less than about 50 km are neglected,
estimates of domain-averaged fluxes and heating rates can be in
serious systematic error. Granted, while most LSAMs attempt to
address the radiative effects of fractional cloudiness, it is not
at all clear, however, whether any of them do an adequate job;
particularly with respect to solar radiation. Clearly, there is a
pressing need to develop parametrizations for characterizing
unresolved cloud geometry as well as 1D radiative transfer
algorithms that account for the effects of unresolved clouds and
cloud overlap.
 
The motivation for this study is at least three fold. First, a
radiative transfer algorithm will be presented that will provide
benchmark calculations for broadband, domain-averaged fluxes.
This is a pseudo-line-by-line, 3D Monte Carlo photon transport
algorithm. Second, we want to test and develop parametrizations
for descriptions of subgrid-scale cloud geometry as well as
radiative transfer algorithms that are suitable for use in LSAMs
and able to handle such descriptions. Third, we are interested
in the range of flux estimates from 1D models when they operate
on identical descriptions of cloud. We believe the later point
has ramifications for the interpretation of LSAM intercomparison
studies; especially those aimed at assessing cloud-radiation
feedbacks.

Unlike the original ICRCCM study, the study proposed here
addresses issues of: how 1D codes interpret 1D descriptions of
cloud properties (such as subgrid-scale horizontal variability
and vertical overlap); and the accuracy of the radiative transfer
models themselves in the event that their interpretation of
unresolved clouds is perfect.

Note that while the focus is on 1D codes used in current LSAMs,
it is hoped that a wider group of 1D modellers will participate.
 
 

2. Methodology

Two- and three-dimensional distributions of cloud water and water
vapour as simulated by cloud-resolving models (CRMs) will be used
to define the test atmospheres. Benchmark domain-averaged
boundary fluxes and heating rate profiles will be computed for
these fields by a pseudo-line-by-line 3D Monte Carlo photon
transport algorithm. This algorithm will be participating in an
intercomparison of 3D radiative transfer codes being coordinated
by R.F. Cahalan (NASA-GSFC). Results from that study will help
establish the minimum range of estimates anticipated for the 1D
codes.

Then, from the 2D and 3D fields, degenerate 1D fields will be
created which will serve as input to the column models. These
fields are simply columns of information required by
participating 1D codes (i.e., cloud fraction, mean cloud water
paths, higher moments of cloud water (if applicable), mean water
vapour, temperature etc...).