Accurate representation of the marine stratocumulus regime in climate modelsHow to represent accurately the marine stratocumulus regime in climate models remains a challenge and to take on this challenge requires a better understanding of how radiation, evaporation, and turbulence interact with one another in maintaining the cloud layer. Moeng, Bjorn Stevens (University of California Los Angeles), and Peter Sullivan used an observed cloud case from the second field study of the DYnamics and Chemistry Of Marine Stratocumulus (DYCOMS-II) as a benchmark to demonstrate that large eddy simulation (LES) can reproduce the turbulence and cloud field reasonably well. They then used the LES flow to address the mixed-layer modeling framework, the use of smooth-top vs. sharp-edge coordinates cloud-top interface properties and entrainment stability. LES shows that the cloud-top interface is not well defined and therefore difficult, if not impossible, to use as a vertical coordinate for mixed-layer models. The simulations show a strong dependency of the entrainment rate and cloud-top instability criterion on the moisture properties of the cloud-top interface. The amelioration of global warming by the advertent and controlled enhancement of the albedo A and longevity L of low-level maritime cloudsJohn Latham further extended his research into a novel idea for the amelioration of global warming by the advertent and controlled enhancement of the albedo A and longevity L of low-Level maritime clouds. More detailed calculations coupled with some limited computer modeling support the quantitative validity of the proposed technique, which involves increasing the droplet concentration in such clouds, with a corresponding increase in both A and L, and thus cooling. The idea involves the dissemination at the ocean surface of small seawater droplets in sufficient quantities to act as the dominant Cloud Condensation Nuclei (CCN) on which cloud droplets form. Satellite control of the overall dissemination rate is envisaged. Collaborators include Keith Bower and Tom Choularton (both of University of Manchester Institute of Science & Technology, United Kingdom); Alan Blyth, Alan Gadian, and Michael Smith (all of University of Leeds, United Kingdom); Stephen Salter, (University of Edinburgh, United Kingdom); and Thomas Wigley (CGD). If this technique proves workable on the scales required, it could be of great societal importance. Clear-air boundary layers (top)Nocturnal boundary layerDuring the previous year, Peggy LeMone, Kyoko Ikeda (RAP), Robert Grossman (CoRA), and Mathias Rotach (Swiss Federal Institute of Technology, Switzerland) reported that just before sunrise, the potential temperature of the air at two m (Q2m) varied linearly with elevation during CASES-97. This relationship has been reported previously, but without explanation. However, the fact that the relationship between Q2m and elevation takes longer to develop over smaller domains (or areas with terrain variation on smaller scale) suggests that advection plays a role. The watershed is shallow (150 m elevation change over 40 km) and mostly crops and grasses, so the local rate of cooling should be similar on the mesoscale. Furthermore, observations show that the drainage currents on nearly calm nights do not vary significantly with elevation. Based on these observations, the researchers developed the explanation illustrated by Table 1. At zero time, the air starts moving downhill. The air flowing downhill cools at the same rate everywhere; in this case, three K for each unit distance traveled down hill. The shaded air starts at the ridge top. Thus the first ridge air to reach the valley floor cools from 300 K to 288 K. The air at ridge top cools more slowly, because it is influenced by air sinking from above. Once air from the ridge reaches the valley, the local cooling rate everywhere is the same as at ridge top, horizontal variability is at a maximum, and Q2m changes linearly with elevation.
Two-dimensional modeling of boundary-layer convectionWith the increasing use of 2D modeling to represent 3D atmospheric convection, it is important to understand the physical aspects of the resolved 2D motion and its ability to simulate 3D turbulence. Moeng, James McWilliams (University of California, Los Angeles), Richard Rotunno, Sullivan, and Jeffrey Weil (visitor, CIRES/University of Colorado) argued that the governing equations of a 2D model that represents 3D convective flows cannot be formally derived from the fundamental fluid mechanics equation, and hence a 2D model can at best represent a parameterization of 3D convection; it is not a simulation. They then showed that a 2D model can be tuned through a simple adjustment to its subgrid-scale eddy viscosity parameter to produce the proper amount of turbulent kinetic energy and other related statistics. For convection with shear, the 2D model can give a reasonable momentum flux (and other wind-related statistics) only if its numerical domain is oriented perpendicular to the mean shear. Verification studies of implicit turbulence modelingIn collaboration with Len Margolin and Andrzej Wyszogrodzki (both of the Los Alamos National Laboratory) Piotr Smolarkiewicz continued evaluation of the implicit subgrid-scale modeling (ISSM) property of the nonoscillatory transport algorithm MPDATA. ISSM greatly improves the predictability of the simulation of flows whose complexity makes the explicit modeling of subgrid-scale motions difficult. In addition to continuing their efforts to validate the ISSM property against experiment, they have also begun a verification effort to study the physical properties of ISSM in the context of known analytic (asymptotic) properties of turbulent flow and by comparison with direct numerical simulation of idealized turbulence. In particular, they have diagnosed probability distributions for fluid velocities (Fig. 33) and two-point correlations in high Reynolds number turbulence, and found that these agree qualitatively with DNS and experimental data.
Their detailed studies of the dissipation mechanisms in MPDATA simulations of decaying turbulence attracted the interest of the Fluid Mechanics community, as judged by invited talks and accompanying papers. Eddy viscosities in implicit large eddy simulations of turbulent flowsIt is frequently argued that truncation errors of monotone numerical schemes serve as an implicit subgrid-scale (SGS) model, an approach to turbulence modeling known loosely as the Monotonically Integrated Large Eddy Simulation (MILES). Despite its importance, little is known about details of numerical dissipation in MILES methods and its relevance to the actual dissipative effects of natural turbulence. Andrzej Domaradzki (University of Southern California) and Smolarkiewicz have developed a methodology for quantifying numerical dissipation as an implicit turbulence model (Phys. Fluids 2003, in press). The nonoscillatory MPDATA scheme was employed as an example of MILES. A series of simulations of decaying homogeneous turbulence was performed in the limit of vanishing viscosity. Thorough comparisons of the numerical results and theoretical predictions revealed that MPDATA dissipation reproduces qualitative features of an ideal SGS model, but that the quantitative comparability depends on the simulation parameters (e.g., the Courant number) and may vary in time. Their results indicate that transforming MILES techniques into universal turbulence models will require customized improvements of nonoscillatory methods to assure adequacy of numerical dissipation for a variety of turbulent flows and physical conditions. Direct numerical simulation of oceanic boundary-layer-current separationThe fastest major currents in the oceans flow along the western
boundaries of ocean basins. The separation of these currents
from their western boundaries as they turn into the open ocean
has proven extremely difficult to capture realistically in
numerical models of ocean circulations. The largest errors
in sea surface temperature in climate models are typically
associated with these problematic boundary currents; for instance
in the Labrador Sea, where the northwestern extension of the
Gulf Stream fails to penetrate. Building upon their development
of a high-performance nonoscillatory forward-in-time (NFT)
nonhydrostatic numerical model for a broad range of oceanic
flows, Frank Bryan (CGD), Matthew Hecht (Los Alamos National
Laboratory), and Smolarkiewicz have applied the model to simulate
directly the relevant laboratory experiments of Baines and
Hughes (J. Phys. Oce. 1996). Accurate simulation of the laboratory
boundary-current separation has been demonstrated. Cold-air outbreak convection (top)Turbulent dispersion of scalarsTurbulent dispersion of scalars can be best understood in
a Lagrangian framework. Jeffrey
Weil (visitor, University
of Colorado/CIRES) , Chin-Hoh Moeng,
Peter Sullivan, Mary Barth, and Si-Wan
Kim (Seoul
National University, Korea) applied a Lagrangian particle model,
developed by Weil, that
follows a large sample
of passive particles along the LES-generated turbulent velocity
flow and computes the probability distribution of the particles
and other dispersion statistics. They described this coupled
LES-Lagrangian modeling approach with focus on the subfilter-scale
treatment in the Lagrangian model and also showed example results
from the model for surface-layer sources in the convective
PBL, including comparisons with laboratory data and surface-layer
similarity theory. They addressed the fumigation process where
an elevated source is entrained into a growing PBL. Crosswind-integrated
concentration shows good agreement with water tank experimental
data. They found a faster onset but a slower completion of
the fumigation and lower maximum concentration with a vertically
thicker initial plume. Ground-level concentration distributions
of plumes initially located above the entrainment zone are
simply time shifted with respect to different initial heights
of inserted plumes. |
|||||||||||
