Abstract

Large ensembles of air parcel trajectories driven by the (large-eddy simulation) LES-generated velocity fields from simulations of stratocumulus clouds were analyzed, focusing on statistics of air parcel in-cloud time scales, as well as their spatial variability. In the case of a drizzling stratocumulus cloud the in-cloud residence time is 2–5 times longer than the characteristic cloud eddy turnover time. About 70% of all air parcels cycle in the cloud more than 2 times and about 50% more than 3 times, thus indicating that air cycling is an essential feature of drizzling stratocumulus cloud dynamics. The extent of cycling is different in the case of nondrizzling stratocumulus cloud, where mean in-cloud time scales are on the order of eddy turnover time. Evidently air cycling in cloud depends on boundary layer stability and flow circulation; the latter is affected by cooling of evaporating drizzle and heating by solar radiation.

Results show significant inhomogeneity of in-cloud time scales, which leads to inhomogeneity in cloud microphysical parameters. The potential effects of in-cloud residence time spatial inhomogeneity on cloud microstructure are obvious and significant. Older parcels will contain larger droplets and previously processed cloud condensation nuclei (CCN). Nonadiabatic mixing between old and new parcels provides new embryos for coagulation and accelerates drizzle formation. It is hypothesized that mixing of parcels with different histories, that is, with drop size distributions at different stages of their evolution, may contribute to the drop spectrum broadening. The results also suggest a possible positive feedback mechanism between drizzle and decoupling, namely, parcels with long time trajectories will favor enhanced drizzle growth, which, in turn, will lead to stronger evaporation below cloud base followed by a stronger increase in stability of the subcloud layer and stronger decoupling; all resulting in more air parcel cycling in cloud and more drizzle, which may eventually lead to stratocumulus cloud breakup.

Abstract

It is shown that, as a result of the flow diffluence at the upper levels of the cloud, droplets of different sizes move along different trajectories. The small droplets with negligible fall velocities will have higher probability of being carried away from the cloud and completely evaporating, while the droplets with larger fall velocities are more likely to be recycled in the cloud. During the recycling process droplets of different sizes reenter the cloud at different spatial locations and mix with the droplet-free air brought from the upper levels of the cloud. This results in the decrease of the total concentration, as well as in the concentration of larger droplets, thus facilitating rain formation through enhanced condensational growth.

It is also shown that the notion of an air parcel as an entity containing various constituents (water vapor, aerosol particles, cloud droplets, etc.), all evolving under the same dynamical conditions, may be rather limited. Our results indicate that these elements may have quite different histories resulting in inhomogeneities in cloud microstructure.

Abstract

A three-dimensional nonhydrostatic anelastic numerical model of a convective cloud with an explicit description of microphysical processes has been developed. Two distribution functions are considered in the model—one for cloud condensation nuclei (19 categories from 0.0076 to 7.6 microns) and another for cloud droplets and raindrops (30 categories on a logarithmic scale from 4 to 3250 microns). The prognostic kinetic equations for these distribution functions enable the calculation of the aerosol and drop spectra starting from activation and culminating in rain formation. The warm rain microphysical processes studied include nucleation, condensation/evaporation, coalescence, breakup and sedimentation. Analysis of supersaturation evolution with time shows that it does not experience growth with the onset of coalescence and correlates well with values derived from quasi-steady assumptions. Sensitivity tests of the accuracy of the supersaturation calculations and the effect of the salt factor in the condensation calculations are also presented.

To demonstrate the model's potential the results of a numerical experiment showing the evolution of a multicellular cloud are described briefly. They reveal such interesting features as vortex formation through merger of updrafts from cells of different intensity, interaction of the cloud with a stable environment, and repeated reentrainment into the cloud of the same air masses that circulate along spiral trajectories near cloud boundaries.

Abstract

The system of trade wind cumulus clouds observed during the RICO field project was simulated by an LES model in a domain of the size of a mesoscale model grid. More than 2000 clouds were analyzed by stratifying them by their cloud-top heights. The investigation was focused on phase transition rates (TR), which in warm tropical clouds are represented by the processes of condensation/evaporation. We previously demonstrated, based on LES data, that a nearly perfect correlation (R = 0.99) exists between upward mass flux (MFP) and condensation rate (CR), and that the correlation between MFP and evaporation rate (ER) is only slightly lower (R = 0.98). The strong dependence of TR on MFP and the linear relationship between them were explained by applying condensation theory and the concept of “quasi-steady” supersaturation. The LES-derived slope of the linear TR–MFP relationship agreed with its theoretical value, with an error of less than 5%. This result implies that supersaturation in clouds, on average, varies within a few percentage points of its quasi-steady value. In our analysis we considered parameters characterizing cloud as a whole, that is, parameters integrated over the cloud volume. However, condensation theory and LES data show that the linear fit is applicable to local variables and therefore may be integrated to obtain relationships for horizontally averaged variables. Expanding the TR–MFP relationship to vertically dependent variables may provide the framework for development of subgrid-scale latent heat release parameterization.

Abstract

A case of coastal California summer season boundary layer cloud has been simulated with the U.S. Navy Coupled Ocean–Atmosphere Mesoscale Prediction System and the results analyzed in the context of consistency with conclusions derived from large eddy simulation–based (LES) studies. Results show a pronounced diurnal cycle and fair agreement with satellite-derived observations of liquid water path. When drizzle processes are included, a significant degree of mesoscale organization emerges in the form of cloud bands, accompanied by a transition from a well-mixed boundary layer topped by unbroken stratocumulus cloud into a more potentially unstable, convective boundary layer regime. The transition and the subsequent development of mesoscale variability is analogous to the drizzle-induced cloud breakup produced in large eddy simulation studies. The dynamics of the pure stratocumulus cloud are dictated by the model's subgrid parameterization, while the more convective regime exhibits appreciable vertical velocities characteristic of an ensemble of cumulus updrafts. The existence of convective updrafts is tied to a weak drizzle-induced decoupling of the cloud and subcloud layer, after which air of higher equivalent potential temperature (θ _e ) can pool at the surface. Some similarities to the propagation of deep convection are also noted.

Abstract

A new dynamical framework for the Cooperative Institute for Mesoscale Meteorological Studies large eddy simulation model (CIMMS LES) with an explicit microphysics scheme is developed. It is shown that simulation results are very sensitive to the drop spectrum remapping technique used in condensation calculations; however, the results are almost insensitive to doubling of the spectrum resolution used in the CIMMS LES model. It is also shown that the drop coagulation procedure conserves the liquid water content as long as the predominant radius of the drop size spectrum, defined as the cube root of the ratio of the drop radar reflectivity to the liquid water content, is below a threshold value of 250 μm. Finally, it is demonstrated that for typical maritime conditions this threshold radius is exceeded only in 0.1% of all cloudy points.

Realism of the model is evaluated by a direct comparison of its predictions with the aircraft observations of a stratocumulus-topped boundary layer. The first simulation is based on the U.K. Meteorological Research Flight flight 526 measurements collected over the North Sea on 22 July 1982; the second simulation corresponds to the Atlantic Stratocumulus Transition Experiment flight A209 on 12–13 June 1992. The model is able to reproduce reasonably well most of the observed boundary layer parameters, including turbulent fluxes and variances of various fields, the intensity and vertical distribution of the turbulent kinetic energy, the upward and downward radiative fluxes, and the cloud drop spectra. It is speculated that the most noticeable discrepancy, which is an underestimation of the concentration of drops smaller than 6 μm near the cloud top, may be an indicator of the need to refine theoretical formulation of small-scale turbulent mixing.

Abstract

Unbiased calculations of microphysical process rates such as autoconversion and accretion in mesoscale numerical weather prediction models require that subgrid-scale (SGS) variability over the model grid volume be taken into account. This variability can be expressed as probability distribution functions (PDFs) of microphysical variables. Using dynamically balanced large-eddy simulation (LES) model results from a case of marine trade cumulus, the authors develop PDFs of the cloud water, droplet concentration, and rainwater variables (q _c , N _c , and q _r ). Both 1D and 2D joint PDFs (JPDFs) are presented. The authors demonstrate that accounting for the JPDFs results in more accurate process rates for a regional-model grid size. Bias in autoconversion and accretion rates are presented, assuming different formulations of the JPDFs. Approximating the 2D PDF using a product of individual 1D PDFs overestimates the autoconversion rates by an order of magnitude, whereas neglecting the SGS variability altogether results in a drastic underestimate of the grid-mean autoconversion rate. PDF assumptions have a much smaller impact on accretion, largely because of the near-linear dependence of the variables in the accretion rate formula and the relatively weak correlation between q _c and q _r over the small LES grid volumes. The latter is attributed to the spatial decorrelation in the vertical between the two fields. Although the full PDFs are both height and time dependent, results suggest that fixed-in-time and fixed-in-height PDFs give an acceptable level of accuracy, especially for the crucial autoconversion calculation.

Abstract

This paper describes a microphysics parameterization based on integral moments of the full drop size distributions (DSDs) as opposed to a partial moments approach (sometimes referred to as Kessler-type parameterization) based on the moments integrated separately over the cloud and rain drop portion of the drop spectrum. This approach does not assume a prescribed form of a DSD but employs as model variables full moments that have clear physical meaning: drop concentration and surface area, water content, precipitation flux, and radar reflectivity. These variables can be directly measured and assimilated into the model forecast cycle without intermediate retrievals. The approach avoids division of DSDs into cloud and rain drops. This eliminates the problem of defining the threshold between these two categories and subdivision of the physical coagulation process into artificial processes of autoconversion, accretion, and self-collection. The development and testing of the parameterization was made using the Cooperative Institute for Mesoscale Meteorological Studies (CIMMS) large-eddy simulation (LES) explicit warm rain microphysical model. The conversion and sedimentation rates were parameterized in the form of a product of power functions using nonlinear regression analysis to determine exponents of the approximated expressions. The comparison of bulk and explicit microphysics models demonstrated reasonably good prediction of both thermodynamic and microphysical parameters of the stratocumulus-topped boundary layer (STBL). The weaknesses and problems of the numerical implementation of the full moment approach are also discussed.

Abstract

A parameterization for giant cloud condensation nuclei (GCCN), suitable for use in bulk microphysical models, has been developed that uses precise representations of the condensational growth of aerosol particles in the subcloud layer. The formulation employs an observationally based GCCN distribution function and directly observable parameters of GCCN, such as concentration and the shape of the aerosol spectra. The parameterization couples naturally to parameterizations of sea salt flux from the ocean surface. The behavior of the GCCN parameterization in a large eddy simulation (LES) framework is consistent with simulations employing explicit, size-resolving microphysical methods. The parameterization properly represents the sensitivity of cloud, drizzle, turbulence, and radiative properties to changes in GCCN concentration for polluted and clean background CCN environments.

Abstract

Calculating unbiased microphysical process rates over mesoscale model grid volumes necessitates knowledge of the subgrid-scale (SGS) distribution of variables, typically represented as probability distribution functions (PDFs) of the prognostic variables. In the 2014 Journal of the Atmospheric Sciences paper by Kogan and Mechem, they employed large-eddy simulation of Rain in Cumulus over the Ocean (RICO) trade cumulus to develop PDFs and joint PDFs of cloud water, rainwater, and droplet concentration. In this paper, the approach of Kogan and Mechem is extended to deeper, precipitating cumulus congestus clouds as represented by a simulation based on conditions from the TOGA COARE field campaign. The fidelity of various PDF approximations was assessed by evaluating errors in estimating autoconversion and accretion rates. The dependence of the PDF shape on grid-mean variables is much stronger in congestus clouds than in shallow cumulus. The PDFs obtained from the TOGA COARE simulations for the calculation of accretion rates may be applied to both shallow and congestus cumulus clouds. However, applying the TOGA COARE PDFs to calculate autoconversion rates introduces unacceptably large errors in shallow cumulus clouds, thus precluding the use of a “universal” PDF formulation for both cloud types.