Page 1:

• Organized cloud systems and large-scale dynamics
• Cloud systems and microscale processes

Go to Page 2:

• Advanced multi-purpose numerical models

Parameterization and Organized Convection

A long-time vexing challenge for regional-scale models (grid-length D ~10's km) has been for numerical weather prediction (NWP) models to accurately predict the space-time distribution of precipitation which originates from convection, and is organized on mesoscales. High-resolution global NWP models (D ~ 40 km) are now at this juncture. Similar problems occur with regard to tropical 'super-clusters' in models having D ~ 80 km, which is near that of seasonal prediction. The implications are fundamental: the very existence of organized convection is at odds with the scale-separation assumption upon which parameterization is based. Changhai Liu and Mitchell Moncrieff studied this problem with regard to summertime convection over the continental United States. As illustrated in Figure 24, sequence-a shows observed sequences of precipitation (Carbone et al. 2002); sequence-b-c-d shows numerical realizations from MM5, using three different convective parameterizations; and sequence-e shows realizations from a cloud-system-resolving model (CSRM) using D = 1 km. The observed speed of propagation is approximately realized by the CSRM, but by none of the parameterizations. This preliminary result suggests that when organized convection is treated explicitly, thereby obviating the scale-separation problem, sequences of precipitation with realistic right space-time structure are realized.

Figure 24. Sequences of precipitation from: a) radar analysis (Carbone et al. 2002); b), c) and d) MM5 using the Kain-Fritsch, Grell and Betts-Miller convective parameterizations, respectively; and e) a cloud-system-resolving model (CSRM) simulation. CSRM is a '5-member ensemble' with the same large-scale forcing specified each of the 5 days. The others are 10-day sequences.

The studies reported in the remainder of this sub-section show that organized convection is a key issue, not only in modern NWP models, but also at time and space scales pertinent to climate research.

Super-parameterization: an aquaplanet study

Wojciech Grabowski continued to investigate the interaction between equatorially-trapped disturbances and tropical convection, using a nonhydrostatic global model, and applying the cloud-resolving convection parameterization (CRCP, also known as super-parameterization). The super-parameterization represents sub-grid scales of the global model by embedding a 2D cloud-resolving model in each column of the global model. The simulations are important for the understanding of the coupling between the large-scale dynamics and deep convection in the tropics, on intraseasonal time scales. The most recent simulations explore the role of large-scale variability of the free-tropospheric moisture on the Madden-Julian Oscillation (MJO). In the control simulation, strong MJO-like coherent structure develops with the equatorial waveguide (Fig. 25). However, when large-scale fluctuations of convectively generated, free-tropospheric moisture are artificially removed, on a time scale of a few hours, MJO does not develop; if already present, it disintegrates rapidly. These results are illustrated in Fig. 26 and 27 and strongly support the significance of the moisture-convection feedback, which was previously hypothesized to explain the large-scale organization of tropical convection and its coupling with SST fluctuations.

Figure 25. Results from the control simulation CTRL. Upper panels show Hovmoller diagrams of the surface precipitation (upper left) and precipitable water (upper right) at the equator. Precipitation intensities larger than 0.2 and 5-mm hr -1 are shown using gray and black shading, respectively. Precipitable water smaller/larger than 65/75 kg m-2 is shown as white/black; gray shading is for precipitable water between 65 and 75 kg m-2. Middle two panels show vertical (middle left) and horizontal (middle right) velocities in the vertical plane at the equator at day 80. Contour interval is 2 cm s-1 (10 m s-1) for vertical (horizontal) velocities starting at 1 cm s-1 5 m s-1); solid (dashed) contours are for positive (negative) values. Lower two panels show spatial distribution of the surface precipitation (lower left) and the sum of surface sensible and latent heat fluxes (lower right) along theequator, also at day 80.

Figure 26. As Fig. 25 CTRL, but for the simulation QVRLX where large-scale free-tropospheric moisture fluctuation are removed on a time scale of a few hours. The precipitable water thresholds for the gray scale are 72 and 74 kg m-2. The vertical velocity contour interval is 0.5 cm s-1 starting at 0.25 cm s-1. The horizontal velocity contour interval is 2 m s-1 starting at 1 m s-1.

Figure 27. As Fig. 25 CTRL, but for days 50 to 60 the simulation from which R-QVRLX is restarted at day 60 and R-QVRLX for days 60 to 70. In R-QVRLX, free-tropospheric moisture fluctuation are removed as in QVRLX. The precipitable water thresholds for the gray scale are 75 and 77 kg m-2. The vertical velocity contour interval is 1 cm s-1 starting at .5 cm s-1. The horizontal velocity contour interval is 5 m s-1 starting at 2.5 m s-1. The temporal resolution for the data for days 50-60 is 6 hrs, whereas it is 2 hrs for R-QVRLX.

Unified dynamics of organized convection and MJO-like systems

Moncrieff set out to quantify properties of the MJO-like structures simulated by Grabowski's super-parameterization. A dynamical model of the large-scale circulation and the role of organized convection was formulated to address questions, such as: 1) What is the simplest possible non-linear model of the MJO?; 2) What role does organized convection play?; 3) How is the mean flow affected?; and, 4) Is there an analytic counterpart of numerical super-parameterization? Moncrieff's nonlinear model shows atmospheric super-rotation is an inevitable consequence of a modon-like coherent structure. A set of dimensional quantities representing a Rossby gyre and embedded organized convection define the MJO-like system. A canonical form of the dynamical model represents the main dynamical characteristics of Grabowski's super-parameterization, including the meridional and vertical transports of zonal momentum.

Super-parameterization in the community climate system model (CCSM)

Michael Ziemianski (postdoctoral fellow from the Institute of Meteorology and Water Management, Poland), in collaboration with Grabowski and William Collins (NCAR/CGD/MMM), began a project to include super-parameterization in the Community Climate System Model (CCSM). The project focuses on three processes which are poorly simulated in the CCSM: cloud-radiative interactions at the sub-grid scale, the warm season convection over land, and gravity wave drag over major mountain chains. The strategy is to apply the super-parameterization over a fraction of the entire CCSM grid, and to focus on each of the above processes in a separate set of model simulations. Ziemanski, Grabowski, and Collins focused on cloud-radiative interactions over the tropical western Pacific warm pool, using an approach in which only one-way coupling between the CCSM and super-parameterization was allowed (i.e., no feedback). These simulations showed that temperature and moisture profiles produced by super-parameterization differed considerably from those in the CCSM. This finding could have a significant impact on the coupled super-parameterization/CCSM simulations.

Resolved convection and the inter-tropical convergence zone

The inter-tropical convergence zone (ITCZ) is difficult to represent accurately in climate models. Liu and Moncrieff extended their 2001 research by investigating the effects of surface friction and cloud-interactive radiation on the ITCZs in an equatorial beta plane, through explicit two-dimensional numerical modeling. Surface friction reduces the convective peak in the ITCZs, but cannot alter the patterns of the spatial convective distribution. In contrast, cloud-interactive radiation enhances the convective activity in the ITCZs and produces more persistent off-equator ITCZs. The frictional impact is physically related to the reduction of surface wind speed and its spatial variability, whereas, the impact of cloud-interactive radiation is largely attributed to the differential radiative heating/cooling between the active convective region and the nearby clear region. This work contributes to the objectives of the Tropical Rainfall Measuring Mission (TRMM).

Cumulus congestus and diurnal variability in TOGA COARE

Liu and Moncrieff performed a four-month cloud-resolving numerical simulation of convective cloud systems over the western Pacific warm pool, during the Tropical Ocean Global Atmosphere Coupled Ocean-Atmosphere Response Experiment (TOGA COARE). Their primary objectives were to quantify the contributions of cumulus congestus in convective heating and moistening, examine the impact of tropospheric humidity and stratification on cumulus congestus, quantify one-dimensional plume models in representing cumulus congestus, and explore the diurnal variation behavior of tropical convection and thermodynamics through wavelet analyses. Extensive evaluation of the simulation results against observations, and a detailed analysis of the model data set, are under way.

Explicit simulation of TRMM-LBA convective systems

Liu and Moncrieff performed three-dimensional cloud-resolving simulations of two tropical continental mesoscale convective systems observed during TRMM-LBA. One system formed as a squall line, with a leading line of strong convection and a trailing region of decaying convection and weak stratiform precipitation, in a moderately-sheared environment. Another system was short-lived and quasi-stationary, in a weakly-sheared environment, and was likely initiated by the thermal forcing in the planetary boundary layer. The model successfully reproduced many observed features, such as the precipitation pattern, lifecycle, convective line orientation, and propagation behavior. Sensitivity experiments illustrated that the ice-phase microphysics played a minor role in the formation of convective bands, but they were important to realistically replicate the stratiform regions in the late evolution stage. Momentum budgets suggested that upgradient and downgradient transport occurred simultaneously in different layers along the line-normal direction, and downgradient transport dominated in the line-parallel direction, in agreement with theory. In addition to the case studies, Liu and Moncrieff are currently performing statistical studies of TRMM-LBA organized convection within the easterly (break) and westerly (monsoon) regimes, using multi-day real-time simulations.

Large-scale tropical dynamics and cloud microphysics

Cloud microphysics is one of the most uncertain aspects of weather prediction and climate research. Grabowski investigated the impact of cloud microphysics on the coupling between moist convection and large-scale tropical dynamics. Two different modeling strategies were used. The first considered the convectively coupled Kelvin waves in two-dimensional cloud-resolving simulations. This study showed that, in simulations without ice microphysics, convectively coupled waves have larger horizontal wavelength, compared to simulations with ice microphysics (Fig. 28). The simulations suggest that the impact of ice microphysics on the organization and longevity of the mesoscale convective systems is the key. This is illustrated in Fig. 29. The second set of simulations considered the impact of cloud microphysics of convective-radiative quasi-equilibrium, on a rotating constant SST aquaplanet, using super-parameterization. The latter simulations demonstrated that the impact of cloud microphysics is associated with two distinct effects: the impact on convective dynamics (where the cloud microphysics affects the partitioning between latent heating and convective transport, for a given radiative cooling tendency) and the impact on cloud-radiative interactions. The two sets of simulations illustrate complementary strategies to investigate the impact of cloud microphysics on weather and climate.

Figure 28. Hovmoller (space-time) diagrams of the surface precipitation rate for the simulation including ice physics (panel a) and the simulation with warm rain micropghysics only (panel b). The light and dark shading represents precipitation rates between 0.2 and 5 mm hr -1, and larger than 5 mm hr -1, respectively. The lines show propagation speed of -10 m s-1 (solid lines in both panels), 8 m s-1 (dashed line in panel a), and 12 m s-1 (dashed line in panel b), all relative to the earth-stationary observer.

Figure 29. A schematic diagram of the impact of ice microphysics on the coupling between deep convection and the large-scale Kelvin wave as suggested by model simulations. The panels show horizontal and vertical structure of the temperature perturbations associated with the large-scale wave (solid and dashed contours for positive and negative values, respectively), cloud outlines within the envelope of convection which forms the convectively active part of the wave, and spatial distribution of the resulting surface precipitation. The upper two panels illustrate the situation with ice microphysics, whereas the bottom two panels are for the warm rain case.

Sequences of precipitation from resolved convection

Liu and Moncrieff conducted two-dimensional cloud-resolving simulations to examine the dynamics and parameterization issue of warm-season convection (see Fig. 24 sequence a-e). The model was forced with the composite diurnal boundary conditions, surface fluxes and advective forcing derived from a 10-day simulation, using MM5. A regular diurnal pattern in the convective development was produced; convection was initiated over the Rockies during afternoon and evening, then propagated eastward at about 14 m/s and caused nocturnal rainfall over the eastern plain, consistent with radar-derived statistics. Convective organization was dominated by fast-moving systems, which possessed leading and trailing anvils, in the environment of moderate low-level shear. In contrast, two convective modes occurred in the weakly sheared case; the fast one had a structure comparable to systems in moderate shear, whereas, the slow mode featured an extensive forward-directed stratiform. Sensitivity experiments indicate that the diurnal convective behavior was only slightly affected by cloud-interactive radiation. By comparison, the convective intensity and diurnal variability were substantially weakened when either orography or large-scale advection was excluded.

Mesoscale clustering of precipitating convection

Liu and Moncrieff studied the effects of planetary rotation on the nonlinear response of an initially quiescent, uniformly stratified fluid, to steady thermal forcing. This setup idealizes convectively generated, diabatic, latent heating in a numerical model. It was found that planetary rotation traps subsidence-induced, adiabatic warming surrounding the heated region, on time scales comparable to the lifetime of mesoscale convective systems. Consequently, the environment adjacent to a convective area is thermally stabilized and dried. This rotation-induced degradation of convective instability and subsequent drying is detrimental to the persistent clumping of convection. This hypothesis is supported by cloud-resolving modeling of convective systems on f-planes maintained by constant radiative cooling and surface fluxes of heat and moisture. It is also consistent with the observation that convection tends to be more clustered and aggregated in the tropics than in the higher latitudes.

Cloud microphysics, surface processes, and radiative transfer in subtropical shallow convection

Grabowski and Gregory McFarquhar (University of Illinois, Urbana-Champaign) have been collaborating on a project aimed at determining the factors that affect cloud cover in the Indian Ocean region. Previous studies have suggested that the low cloud cover observed in this region may be associated with a semi-direct effect, whereby absorption of solar radiation by soot particles causes the dessication of cloud layers. Using a cloud-resolving model, Grabowski and McFarquhar are determining the relative role of several factors on cloud cover, and comparing their simulations with observations collected during the Indian Ocean Experiment (INDOEX) by Andrew Heymsfield and McFarquhar. Preliminary results show that on individual days, surface fluxes substantially affect cloud cover. Ongoing research will quantify the effects of aerosols on the water and energy budget of the Indian Ocean region.

Aerosol, glaciation and precipitation in cumulus clouds

John Latham, in collaboration with Vaughan Phillips (Princeton University), Tom Choularton (UMIST, UK), and Alan Blyth (University of Leeds, UK), examined the influence of aerosol concentrations on the glaciation and precipitation characteristics of cumulus clouds. The principal computational tools were the UMIST Explicit Microphysics Model and the UK Meteorological Office Cloud Resolving Model. In the shallow phase of cloud development, increasing CCN concentrations produced a significant decrease in the precipitation efficiency and in the ice crystal concentrations. In deeper clouds, the anvil ice-crystal concentrations increased with increasing CCN, but the precipitation rate was essentially unaffected.

Microphysical properties of optically thin clouds and parameterizations

Optically thin cirrus clouds cover as much as 30% of tropical and subtropical regions and may trap a considerable amount of longwave radiation emitted from the earth's surface, thereby influencing climate. Currently the microphysical and radiative properties of these clouds are poorly understood. The Cirrus Regional Study of Tropical Anvils and Cirrus Layers Florida Area Cirrus Experiment (CRYSTAL-FACE) research project provided an opportunity to study these issues by supporting in-situ measurements of the microphysical properties of subvisual cirrus on three occasions. The NCAR Video Ice Particle Sampler (VIPS) probe, a version that was developed for CRYSTAL and flown on the NASA WB57 aircraft, provided excellent size distribution and extinction information on two of these flights. Heymsfield and Carl Schmitt collected and analyzed the data, an example of which appears over a one-hour period as was shown in Figure 10. Direct measurements of the extinction made by the Cloud Integrating Nephelometer (CIN) probe, operated by Timothy Garrett (University of Utah), compare favorably to the estimates from the VIPS data. Measurements within this layer have been related to measurements of size distributions from a more conventional microphysical probe, a Cloud Aerosol and Precipitation Spectrometer (CAPS), operated by Darrel Baumgardner (Universidad Nacional Autónoma de México).

Thundercloud ice characteristics

Latham and James Dye, in collaboration with Hugh Christian and Wiebke Deierling (both NASA/MSFC), Alan Gadian (UMIST, UK), Alan Blyth, (University of Leeds, UK), and Rumjana Mitzeva (University of Sofia, Bulgaria), examined the extent to which it is possible to determine thundercloud ice characteristics from satellite observations of lightning, now routinely made on a global scale, using NASA/MSFC devices. A specific goal is to ascertain whether measurements of lightning frequency f can yield estimates of precipitating and non-precipitating ice fluxes. Computations predict that f is roughly proportional to the product of the downward flux fg of graupel through the body of the thundercloud and the upward flux fi of ice crystals into its anvil. This raises the possibility of determining, on a global basis, values of fg and/or fi from the lightning measurements. Such information could have considerable climatological and nowcasting importance.

Small-scale turbulent mixing in clouds

In collaboration with Miroslaw Andrejczuk and Szymon Malinowski (both Warsaw University, Poland), Grabowski and Piotr Smolarkiewicz completed a modeling study of decaying moist turbulence. Its importance, beyond fundamental understanding, is for applications such as radiative transfer through clouds, initiation of precipitation in warm (i.e., ice-free) clouds, and parameterization of small-scale and microscale processes in models resolving larger scales. In the moist case, kinetic energy of small-scale motions originates not only from the classical downscale energy cascade, but is also generated/enhanced internally by the phase change processes and droplet sedimentation. A series of moist simulations was performed and contrasted with corresponding dry reference runs. The results suggest that, as far as the evolution of turbulent kinetic energy and enstrophy is concerned, significant impact of moist processes is only possible at relatively low levels of the large-scale input of the kinetic energy (Fig. 30). Then, phase change processes and droplet sedimentation invigorate substantially dry turbulent mixing. However, significant anisotropy, consistent with that observed in laboratory experiments on mixing between cloudy and cloud-free air, prevails even at high large-scale input of the kinetic energy.

Figure 30. Evolution of turbulent kinetic energy (TKE) in simulations of 3D decaying turbulence. Upper, middle, and lower panels correspond to the high, moderate and low initial TKE, respectively. Solid lines show evolutions for a reference dry case, while dashed-dot and dashed lines are for moist simulations using bulk and detailed microphysics, respectively.

Implicit turbulence modeling

In collaboration with Andrzej Domaradzki (University of Southern California) and Len Margolin (Los Alamos National Laboratory), Smolarkiewicz continued his evaluation of the implicit subgrid-scale modeling property of the nonoscillatory transport algorithm (MPDATA), which is important for those applications where the complexity of natural flows makes the explicit modeling of subgrid-scale motions difficult. Numerical solutions of the Navier-Stokes equations, using nonoscillatory methods (known as MILES, VLES, etc.), have been highly successful in reproducing the dynamics of high Reynolds number turbulence, without the need to invoke explicit subgrid-scale models. Margolin and Smolarkiewicz have demonstrated the realizability of inviscid MPDATA results, using a combination of mathematical analysis and a computational study of convergence in resolution and viscosity. Domaradzki and Smolarkiewicz simulated homogeneous rotating/nonrotating turbulence in the limit of vanishing viscosity, obtaining correct spectra and kinetic energy decay rates, as well as an effective spectral eddy viscosity that evinces the same qualitative behavior as Kraichnan's classical eddy viscosity.

Turbulence and the collision rate of cloud droplets

Professor Lian-Ping Wang (University of Delaware), in collaboration with Grabowski, studied the effects of turbulence on the collision of cloud droplets when droplet inertia, gravity, and turbulence microstructure are considered. This is an important problem because the impact of cloud turbulence on microphysical processes (warm rain initiation in particular) remains ambiguous. Direct numerical simulations were used to generate the turbulent flow. The relative droplet inertia and settling velocity were specified according to the conditions typical for convective clouds. Numerical results, illustrated in Figure 31 show that droplet inertia and fluid turbulence can increase the collision kernel, relative to the gravity-only case, by as much as 70% for droplets of 10 to 50 micrometers in size, as a result of relative velocity fluctuations and preferential concentration of the droplets. For larger droplet sizes, the preferential concentration is a significant contributor. The relative velocity fluctuations play an important role for droplets of similar sizes, (i.e. when the gravity-only case results in no collisions). A leading order analysis was found to accurately predict the droplet-droplet relative velocity statistics.

Figure 31. Collision kernels for different droplet size combinations. For each plot, the size of the second droplet is varied while the size of the first droplet group is fixed at (a) 10; (b) 20; (c) 30; (d) 40; and (e) 50 microns. Results are shown for gravitational case (solid lines), as well as for two cases with different intensities of cloud turbulence..