Search Results

You are looking at 11 - 20 of 22 items for

  • Author or Editor: Zachary J. Lebo x
  • Refine by Access: All Content x
Clear All Modify Search
Vanessa M. Przybylo
,
Kara J. Sulia
,
Zachary J. Lebo
, and
Carl G. Schmitt

Abstract

The Ice Particle and Aggregate Simulator (IPAS) is used to theoretically represent the aggregation process of ice crystals. Aggregates have a variety of formations based on initial ice particle size, shape, and falling orientation, all of which influence water phase partitioning. Aggregate dimensional properties and density changes are calculated for monomer–monomer (MON–MON), monomer–aggregate (MON–AGG), and aggregate–aggregate (AGG–AGG) collection to be used by ice-microphysical models for improvement in aggregation parameterizations. Aggregates are chosen from a database of 9 744 000 preformed combinations to be further collected (see ). AGG–AGG collection results in more extreme and a smaller range of aggregate aspect ratios than MON–AGG collection. A majority of aggregates are closer to prolate than oblate spheroids, regardless of collection type, except for quasi-horizontally oriented particles that have extreme aspect ratios to begin with. MON–AGG collection frequently results in an increase in density upon collection, whereas MON–MON and AGG–AGG collection almost always result in particle density decreases, with extreme reductions near 99% for MON–MON collection. MON–MON collection results in the greatest decreases in density but then quickly becomes unaffected by the addition of more monomers due to inherent size differences between monomers and aggregates. Finally, a holistic analysis to in situ observations of cloud particle images is presented. IPAS 2D aspect ratios surround a median value of 0.6 and closely follow that of previous studies while varying by no more than ≈12% on average from observed aggregates.

Restricted access
Vanessa M. Przybylo
,
Kara J. Sulia
,
Carl G. Schmitt
, and
Zachary J. Lebo

Abstract

A vast amount of ice crystal imagery exists from a variety of field campaign initiatives that can be utilized for cloud microphysical research. Here, nine convolutional neural networks are used to classify particles into nine regimes on over 10 million images from the Cloud Particle Imager probe, including liquid and frozen states and particles with evidence of riming. A transfer learning approach proves that the Visual Geometry Group (VGG-16) network best classifies imagery with respect to multiple performance metrics. Classification accuracies on a validation dataset reach 97% and surpass traditional automated classification. Furthermore, after initial model training and preprocessing, 10 000 images can be classified in approximately 35 s using 20 central processing unit cores and two graphics processing units, which reaches real-time classification capabilities. Statistical analysis of the classified images indicates that a large portion (57%) of the dataset is unusable, meaning the images are too blurry or represent indistinguishable small fragments. In addition, 19% of the dataset is classified as liquid drops. After removal of fragments, blurry images, and cloud drops, 38% of the remaining ice particles are largely intersecting the image border (≥10% cutoff) and therefore are considered unusable because of the inability to properly classify and dimensionalize. After this filtering, an unprecedented database of 1 560 364 images across all campaigns is available for parameter extraction and bulk statistics on specific particle types in a wide variety of storm systems, which can act to improve the current state of microphysical parameterizations.

Restricted access
Matthew R. Kumjian
,
Kevin A. Bowley
,
Paul M. Markowski
,
Kelly Lombardo
,
Zachary J. Lebo
, and
Pavlos Kollias
Full access
Matthew R. Kumjian
,
Kevin A. Bowley
,
Paul M. Markowski
,
Kelly Lombardo
,
Zachary J. Lebo
, and
Pavlos Kollias

Abstract

An engaged scholarship project called “Snowflake Selfies” was developed and implemented in an upper-level undergraduate course at The Pennsylvania State University (Penn State). During the project, students conducted research on snow using low-cost, low-tech instrumentation that may be readily implemented broadly and scaled as needed, particularly at institutions with limited resources. During intensive observing periods (IOPs), students measured snowfall accumulations, snow-to-liquid ratios, and took microscopic photographs of snow using their smartphones. These observations were placed in meteorological context using radar observations and thermodynamic soundings, helping to reinforce concepts from atmospheric thermodynamics, cloud physics, radar, and mesoscale meteorology courses. Students also prepared a term paper and presentation using their datasets/photographs to hone communication skills. Examples from IOPs are presented. The Snowflake Selfies project was well received by undergraduate students as part of the writing-intensive course at Penn State. Responses to survey questions highlight the project’s effectiveness at engaging students and increasing their enthusiasm for the semester-long project. The natural link to social media broadened engagement to the community level. Given the successes at Penn State, we encourage Snowflake Selfies or similar projects to be adapted or implemented at other institutions.

Free access
Hugh Morrison
,
Mikael Witte
,
George H. Bryan
,
Jerry Y. Harrington
, and
Zachary J. Lebo

Abstract

This study investigates droplet size distribution (DSD) characteristics from condensational growth and transport in Eulerian dynamical models with bin microphysics. A hierarchy of modeling frameworks is utilized, including parcel, one-dimensional (1D), and three-dimensional large-eddy simulation (LES). The bin DSDs from the 1D model, which includes only vertical advection and condensational growth, are nearly as broad as those from the LES and in line with observed DSD widths for stratocumulus clouds. These DSDs are much broader than those from Lagrangian microphysical calculations within a parcel framework that serve as a numerical benchmark for the 1D tests. In contrast, the bin-modeled DSDs are similar to the Lagrangian microphysical benchmark for a rising parcel in which Eulerian transport is not considered. These results indicate that numerical diffusion associated with vertical advection is a key contributor to broadening DSDs in the 1D model and LES. This DSD broadening from vertical numerical diffusion is unphysical, in contrast to the physical mixing processes that previous studies have indicated broaden DSDs in real clouds. It is proposed that artificial DSD broadening from vertical numerical diffusion compensates for underrepresented horizontal variability and mixing of different droplet populations in typical LES configurations with bin microphysics, or the neglect of other mechanisms that broaden DSDs such as growth of giant cloud condensation nuclei. These results call into question the ability of Eulerian dynamical models with bin microphysics to investigate the physical mechanisms for DSD broadening, even though they may reasonably simulate overall DSD characteristics.

Full access
Lianet Hernández Pardo
,
Hugh Morrison
,
Luiz A. T. Machado
,
Jerry Y. Harrington
, and
Zachary J. Lebo

Abstract

In this study, processes that broaden drop size distributions (DSDs) in Eulerian models with two-moment bin microphysics are analyzed. Numerous tests are performed to isolate the effects of different physical mechanisms that broaden DSDs in two- and three-dimensional Weather Research and Forecasting Model simulations of an idealized ice-free cumulus cloud. Sensitivity of these effects to modifying horizontal and vertical model grid spacings is also examined. As expected, collision–coalescence is a key process broadening the modeled DSDs. In-cloud droplet activation also contributes substantially to DSD broadening, whereas evaporation has only a minor effect and sedimentation has little effect. Cloud dilution (mixing of cloud-free and cloudy air) also broadens the DSDs considerably, whether or not it is accompanied by evaporation. This mechanism involves the reduction of droplet concentration from dilution along the cloud’s lateral edges, leading to locally high supersaturation and enhanced drop growth when this air is subsequently lifted in the updraft. DSD broadening ensues when the DSDs are mixed with those from the cloud core. Decreasing the horizontal and vertical model grid spacings from 100 to 30 m has limited impact on the DSDs. However, when these physical broadening mechanisms (in-cloud activation, collision–coalescence, dilution, etc.) are turned off, there is a reduction of DSD width by up to ~20%–50% when the vertical grid spacing is decreased from 100 to 30 m, consistent with effects of artificial broadening from vertical numerical diffusion. Nonetheless, this artificial numerical broadening appears to be relatively unimportant overall for DSD broadening when physically based broadening mechanisms in the model are included for this cumulus case.

Free access
Yishi Hu
,
Zachary J. Lebo
,
Bart Geerts
,
Yonggang Wang
, and
Yazhe Hu

Abstract

Part I of this series presented a detailed overview of post-frontal mixed-phase clouds observed during the Measurements of Aerosols, Radiation and CloUds over the Southern Ocean (MARCUS) field campaign. In Part II, we focus on a multi-day (February 23-26, 2018) case with the aim of understanding ice production as well as model sensitivity to ice process parameterizations using the Weather Research and Forecast (WRF) model. Interestingly, the control simulation with the Predicted Particle Properties (P3) microphysics scheme underestimates the ice content and overestimates the supercooled liquid water content, contrary to the bias common in global climate models. The simulations targeted at ice production processes show negligible sensitivity to cloud droplet number concentrations. Further, neither increasing ice nuclei particle (INP) concentrations to an unrealistic level nor adjusting it to MARCUS field estimations alone guarantees more ice production in the model. However, the simulated clouds are found to be highly sensitive to the implementation of immersion freezing, the thresholding of condensation/deposition freezing initiation, and the propensity of rime splintering process. By increasing immersion freezing of cloud droplets, relaxing thresholds for condensation/deposition freezing, or removing rime splintering thresholds, the model significantly improves its performance in producing ice. The relaxation of temperature threshold to observed cloud top temperature suggests an in-cloud seeder-feeder mechanism. The results of this work call for an increase in observations of INP, especially over the remote Southern Ocean and at relatively high temperatures, and measurements of ice particle size distributions to better constrain ice nucleating processes in models.

Restricted access
Timothy W. Juliano
,
Zachary J. Lebo
,
Gregory Thompson
, and
David A. Rahn

Abstract

The ability of global climate models to simulate accurately marine stratiform clouds continues to challenge the atmospheric science community. These cloud types, which account for a large uncertainty in Earth’s radiation budget, are generally difficult to characterize due to their shallowness and spatial inhomogeneity. Previous work investigating marine boundary layer (MBL) clouds off the California coast has focused on clouds that form under the typical northerly flow regime during the boreal warm season. From about June through September, however, these northerly winds may reverse and become southerly as part of a coastally trapped disturbance (CTD). As the flow surges northward, it is accompanied by a broad cloud deck. Because these events are difficult to forecast, in situ observations of CTDs are few and far between, and little is known about their cloud physical properties. A climatological perspective of 23 CTD events—spanning the years from 2004 to 2016—is presented using several data products, including model reanalyses, buoys, and satellites. For the first time, satellite retrievals suggest that CTD cloud decks may play a unique role in the radiation budget due to a combination of aerosol sources that enhance cloud droplet number concentration and reduce cloud droplet effective radius. This particular type of cloud regime should therefore be treated differently than that which is more commonly found in the summertime months over the northeast Pacific Ocean. The potential influence of a coherent wind stress cycle on sea surface temperatures and sea salt aerosol is also explored.

Full access
Timothy W. Juliano
,
Matthew M. Coggon
,
Gregory Thompson
,
David A. Rahn
,
John H. Seinfeld
,
Armin Sorooshian
, and
Zachary J. Lebo

Abstract

Modeling marine low clouds and fog in coastal environments remains an outstanding challenge due to the inherently complex ocean–land–atmosphere system. This is especially important in the context of global circulation models due to the profound radiative impact of these clouds. This study utilizes aircraft and satellite measurements, in addition to numerical simulations using the Weather Research and Forecasting (WRF) Model, to examine three well-observed coastally trapped disturbance (CTD) events from June 2006, July 2011, and July 2015. Cloud water-soluble ionic and elemental composition analyses conducted for two of the CTD cases indicate that anthropogenic aerosol sources may impact CTD cloud decks due to synoptic-scale patterns associated with CTD initiation. In general, the dynamics and thermodynamics of the CTD systems are well represented and are relatively insensitive to the choice of physics parameterizations; however, a set of WRF simulations suggests that the treatment of model physics strongly influences CTD cloud field evolution. Specifically, cloud liquid water path (LWP) is highly sensitive to the choice of the planetary boundary layer (PBL) scheme; in many instances, the PBL scheme affects cloud extent and LWP values as much as or more than the microphysics scheme. Results suggest that differences in the treatment of entrainment and vertical mixing in the Yonsei University (nonlocal) and Mellor–Yamada–Janjić (local) PBL schemes may play a significant role. The impact of using different driving models—namely, the North American Mesoscale Forecast System (NAM) 12-km analysis and the NCEP North American Regional Reanalysis (NARR) 32-km products—is also investigated.

Full access
Zachary J. Lebo
,
Ben J. Shipway
,
Jiwen Fan
,
Istvan Geresdi
,
Adrian Hill
,
Annette Miltenberger
,
Hugh Morrison
,
Phil Rosenberg
,
Adam Varble
, and
Lulin Xue
Full access