Events (Upcoming & Past)

Past MMM Events

Constraining Storm-Scale Ensemble Forecasts of Convective Initiation with Dense Surface Observations

Luke MadausNCAR / ASP 

Efforts to increase forecast accuracy and lead-time for convective storms are hampered by an inability to adequately capture storm-scale processes that lead to convective initiation (CI).  The potential contributions of dense networks of standard surface observations to this problem will be discussed.  A series of idealized simulations of 21 environments where CI can occur due to boundary-layer processes alone are performed to isolate storm-scale CI processes from larger meso- or synoptic-scale forcing.  Ensemble simulations of CI in these environments are examined to identify common features in standard surface fields prior to CI.  The horizontal scales and magnitudes of these features as well as correlation length scales throughout the CI process suggest surface observation densities of less than 4 km between observations could capture the observed features.  Observing system simulation experiments (OSSEs) are performed with simulations of one of these environments.  In these OSSEs, skillful storm-scale forecasts of CI are possible when simulated observations from a sufficiently dense surface observing network are assimilated.  However, skill is only possible in forecasts initiated within one hour prior to the onset of precipitation in the developing storms.  Additional aspects of convective-scale data assimilation are considered, particularly with regard to maintaining adequate ensemble spread and the structure of the assimilation increments. Finally, we briefly consider one candidate dense surface observation network---smartphone pressure observations---to assess the quality of these observations and their utility for this problem. 

 

Building:
Room Number: 
1001
Type of event:
Will this event be webcast to the public by NCAR|UCAR?: 
Calendar Timing: 
Friday, December 2, 2016 - 5:30am to 7:00am

Constraining Storm-Scale Ensemble Forecasts of Convective Initiation with Dense Surface Observations

Luke Madaus
NCAR / ASP 

Efforts to increase forecast accuracy and lead-time for convective storms are hampered by an inability to adequately capture storm-scale processes that lead to convective initiation (CI).  The potential contributions of dense networks of standard surface observations to this problem will be discussed.  A series of idealized simulations of 21 environments where CI can occur due to boundary-layer processes alone are performed to isolate storm-scale CI processes from larger meso- or synoptic-scale forcing.  Ensemble simulations of CI in these environments are examined to identify common features in standard surface fields prior to CI.  The horizontal scales and magnitudes of these features as well as correlation length scales throughout the CI process suggest surface observation densities of less than 4 km between observations could capture the observed features.  Observing system simulation experiments (OSSEs) are performed with simulations of one of these environments.  In these OSSEs, skillful storm-scale forecasts of CI are possible when simulated observations from a sufficiently dense surface observing network are assimilated.  However, skill is only possible in forecasts initiated within one hour prior to the onset of precipitation in the developing storms.  Additional aspects of convective-scale data assimilation are considered, particularly with regard to maintaining adequate ensemble spread and the structure of the assimilation increments. Finally, we briefly consider one candidate dense surface observation network---smartphone pressure observations---to assess the quality of these observations and their utility for this problem. 

 

First Name: 
Caroline
Last Name: 
Haws
Phone Extension (4 digits): 
8189
Email: 
haws@ucar.edu
Building:
Room Number: 
1001
Host lab/program/group:
Type of event:
Calendar Timing: 
Thursday, December 1, 2016 - 3:30pm to 5:00pm

Stochastic ice nucleation and its effect on the microphysical properties of mixed-phase stratiform cloud

 Fan YangAtmospheric Sciences Program, Michigan Technological University

Mixed-phase stratiform clouds can persist with steady ice precipitation for hours and even days. The origin and microphysical properties of the ice crystals are of interest. Vapor deposition growth and sedimentation of ice particles along with a uniform volume source of ice nucleation lead to a power law relation between ice water content (wi) and ice number concentration (ni) with exponent 2.5. The relation is confirmed by both a large-eddy simulation cloud model and Lagrangian ice particle tracking with cloud volume source of ice particles through a time-dependent cloud field. Initial indications of the scaling law are observed in data from the Indirect and Semi-Direct Aerosol Campaign (ISDAC). Based on the observed wi and ni from ISDAC, a lower bound of 0.006 m-3s-1 is obtained for the volume ice crystal formation rate.

Results from Lagrangian ice particle tracking method also show that more than 10% of ice particles have lifetimes longer than 1.5 h, much longer than the large eddy turnover time or the time for a crystal to fall through the depth of a nonturbulent cloud. An analysis of trajectories in a 2-D idealized field shows that there are two types of long-lifetime ice particles: quasi-steady and recycled growth. For quasi-steady growth, ice particles are suspended in the updraft velocity region for a long time. For recycled growth, ice particles are trapped in the large eddy structures, and whether ice particles grow or sublimate depends on the ice relative humidity profile within the boundary layer. Some ice particles can grow after each cycle in the trapping region, until they are too large to be trapped, and thus have long lifetimes. The relative contribution of the recycled ice particles to the cloud mean ice water content depends on both the dynamic and thermodynamic properties of the mixing layer. In particular, the total ice water content of a mixed-phase cloud in a decoupled boundary layer can be much larger than that in a fully coupled boundary layer.

In the end I’ll present our resent work about the turbulence cloud chamber at Michigan Technological University. Both observational and modeling results show the aerosol effect on cloud droplet size distribution: increasing aerosol number concentration will narrow the cloud droplet size distribution due to reduced supersaturation fluctuations.

Thursday, 17 November 2016, 3:30 PM

Refreshments 3:15 PMNCAR-Foothills Laboratory

3450 Mitchell LaneBldg. 2, Main Auditorium, Room 1022 

Building:
Room Number: 
1022
Type of event:
Will this event be webcast to the public by NCAR|UCAR?: 
Calendar Timing: 
Friday, November 18, 2016 - 5:30am to 7:00am

Stochastic ice nucleation and its effect on the microphysical properties of mixed-phase stratiform cloud

 Fan Yang
Atmospheric Sciences Program, Michigan Technological University

Mixed-phase stratiform clouds can persist with steady ice precipitation for hours and even days. The origin and microphysical properties of the ice crystals are of interest. Vapor deposition growth and sedimentation of ice particles along with a uniform volume source of ice nucleation lead to a power law relation between ice water content (wi) and ice number concentration (ni) with exponent 2.5. The relation is confirmed by both a large-eddy simulation cloud model and Lagrangian ice particle tracking with cloud volume source of ice particles through a time-dependent cloud field. Initial indications of the scaling law are observed in data from the Indirect and Semi-Direct Aerosol Campaign (ISDAC). Based on the observed wi and ni from ISDAC, a lower bound of 0.006 m-3s-1 is obtained for the volume ice crystal formation rate.

Results from Lagrangian ice particle tracking method also show that more than 10% of ice particles have lifetimes longer than 1.5 h, much longer than the large eddy turnover time or the time for a crystal to fall through the depth of a nonturbulent cloud. An analysis of trajectories in a 2-D idealized field shows that there are two types of long-lifetime ice particles: quasi-steady and recycled growth. For quasi-steady growth, ice particles are suspended in the updraft velocity region for a long time. For recycled growth, ice particles are trapped in the large eddy structures, and whether ice particles grow or sublimate depends on the ice relative humidity profile within the boundary layer. Some ice particles can grow after each cycle in the trapping region, until they are too large to be trapped, and thus have long lifetimes. The relative contribution of the recycled ice particles to the cloud mean ice water content depends on both the dynamic and thermodynamic properties of the mixing layer. In particular, the total ice water content of a mixed-phase cloud in a decoupled boundary layer can be much larger than that in a fully coupled boundary layer.

In the end I’ll present our resent work about the turbulence cloud chamber at Michigan Technological University. Both observational and modeling results show the aerosol effect on cloud droplet size distribution: increasing aerosol number concentration will narrow the cloud droplet size distribution due to reduced supersaturation fluctuations.

Thursday, 17 November 2016, 3:30 PM

Refreshments 3:15 PM
NCAR-Foothills Laboratory

3450 Mitchell Lane
Bldg. 2, Main Auditorium, Room 1022 

First Name: 
Caroline
Last Name: 
Haws
Phone Extension (4 digits): 
8189
Email: 
haws@ucar.edu
Building:
Room Number: 
1022
Host lab/program/group:
Type of event:
Calendar Timing: 
Thursday, November 17, 2016 - 3:30pm to 5:00pm

 New Applications of Advanced Data Assimilation: Improving the model, the observations and detecting extreme events

Eugenia KalnayDepartment of Atmospheric and Oceanic Science, University of MarylandCollege Park, MD

 Data assimilation has been traditionally used to obtain accurate initial conditions for forecasting by combining observations with short-range forecasts. We will present new applications to improve the models and the observations, to allow one submodel (like the atmosphere) to assimilate observations from another submodel (like the ocean), and to use the ensemble of forecasts to predict extreme events.

Thursday, 3 November 2016, 3:30 PM

Refreshments 3:15 PMNCAR-Foothills Laboratory3450 Mitchell LaneBldg. 2, Main Auditorium, Room 1022

Building:
Room Number: 
1022
Type of event:
Will this event be webcast to the public by NCAR|UCAR?: 
Calendar Timing: 
Friday, November 4, 2016 - 3:30am to 5:00am

 New Applications of Advanced Data Assimilation: Improving the model, the observations and detecting extreme events

Eugenia Kalnay
Department of Atmospheric and Oceanic Science, University of Maryland
College Park, MD

 Data assimilation has been traditionally used to obtain accurate initial conditions for forecasting by combining observations with short-range forecasts. We will present new applications to improve the models and the observations, to allow one submodel (like the atmosphere) to assimilate observations from another submodel (like the ocean), and to use the ensemble of forecasts to predict extreme events.

Thursday, 3 November 2016, 3:30 PM

Refreshments 3:15 PM
NCAR-Foothills Laboratory
3450 Mitchell Lane
Bldg. 2, Main Auditorium, Room 1022

First Name: 
Caroline
Last Name: 
Haws
Phone Extension (4 digits): 
8189
Email: 
haws@ucar.edu
Building:
Room Number: 
1022
Host lab/program/group:
Type of event:
Calendar Timing: 
Thursday, November 3, 2016 - 3:30pm to 5:00pm

Data Assimilation Program Seminar 

Extended ensemble Kalman filters for data assimilation in hierarchical state-space models

Matthias KatzfussTexas A&M UniversityCollege Station, Texas

 The ensemble Kalman filter (EnKF) is a computational technique for approximate inference on the state vector in spatio-temporal state-space models. It has been successfully used in many real-world nonlinear data-assimilation problems with very high dimensions. However, the EnKF is most appropriate for additive Gaussian state-space models with linear observation equation and no unknown parameters. Here, we consider a broader class of hierarchical or conditional state-space models, which include two additional layers: The parameter layer allows handling of unknown variables that cannot be easily included in the state vector, while the transformation layer can be used to model non-Gaussian observations. For approximate Bayesian inference in such hierarchical state-space models, we propose a general class of extended EnKFs, which approximate inference on the state vector in suitable existing Bayesian inference techniques using the EnKF and the ensemble Kalman smoother. Our extended EnKFs enable approximate, computationally feasible filtering and smoothing in many high-dimensional, nonlinear, and non-Gaussian models with unknown parameters. Focusing on ensemble-based particle and Gibbs approaches, we highlight several interesting examples, including on-line model selection, parameter inference, and assimilation of non-Gaussian observations.

 Friday, 28 October 2016, 10:00 AM

Refreshments 9:45 AMNCAR-Foothills Laboratory3450 Mitchell LaneBldg. 2, Small Seminar, Room 1001

Building:
Room Number: 
1001
Type of event:
Will this event be webcast to the public by NCAR|UCAR?: 
Calendar Timing: 
Friday, October 28, 2016 - 10:00pm to 11:00pm

Data Assimilation Program Seminar 

Extended ensemble Kalman filters for data assimilation in hierarchical state-space models

Matthias Katzfuss
Texas A&M University
College Station, Texas

 The ensemble Kalman filter (EnKF) is a computational technique for approximate inference on the state vector in spatio-temporal state-space models. It has been successfully used in many real-world nonlinear data-assimilation problems with very high dimensions. However, the EnKF is most appropriate for additive Gaussian state-space models with linear observation equation and no unknown parameters. Here, we consider a broader class of hierarchical or conditional state-space models, which include two additional layers: The parameter layer allows handling of unknown variables that cannot be easily included in the state vector, while the transformation layer can be used to model non-Gaussian observations. For approximate Bayesian inference in such hierarchical state-space models, we propose a general class of extended EnKFs, which approximate inference on the state vector in suitable existing Bayesian inference techniques using the EnKF and the ensemble Kalman smoother. Our extended EnKFs enable approximate, computationally feasible filtering and smoothing in many high-dimensional, nonlinear, and non-Gaussian models with unknown parameters. Focusing on ensemble-based particle and Gibbs approaches, we highlight several interesting examples, including on-line model selection, parameter inference, and assimilation of non-Gaussian observations.

 Friday, 28 October 2016, 10:00 AM

Refreshments 9:45 AM
NCAR-Foothills Laboratory
3450 Mitchell Lane
Bldg. 2, Small Seminar, Room 1001

First Name: 
Caroline
Last Name: 
Haws
Phone Extension (4 digits): 
8189
Email: 
haws@ucar.edu
Building:
Room Number: 
1001
Host lab/program/group:
Type of event:
Calendar Timing: 
Friday, October 28, 2016 - 10:00am to 11:00am

Helical tropical cyclogenesis: a modern look based on cloud-resolving numerical analysis of self-organization of moist convective atmospheric turbulence

 Galina LevinaSpace Research Institute, Russian Academy of SciencesMoscow, Russia

Recent results of our collaborative Russian-American efforts on how a notion of helicity can be applied in the atmospheric research to tropical cyclone (TC) investigations will be presented. Briefly recalling the role of helical turbulence in the formation of large-scale structures in magnetohydrodynamics and general dynamics of non-conducting fluids, we make an accent on the existence of threshold for large-scale instabilities in all cases. To bring together the notion of helicity and TC formation, we emphasize one of the very first achievements obtained by near-cloud-resolving numerical simulation of tropical cyclogenesis. This is the discovery of vortical nature of atmospheric moist convection in the tropical zone – rotating cumulonimbus clouds, which were dubbed ‘Vortical Hot Towers (VHTs)’ – and their crucial role in TC formation (Hendricks et al., 2004; Montgomery et al. 2006). As it was noted by Molinari and Vollaro (2010), “VHTs are helical by definition because they contain coincident updrafts and vertical vorticity”.

For the first time in TC research, we highlight the inherently helical tropical cyclogenesis. This implies the role of a special topology of the newly forming mesoscale vortex and the contribution of motions of cloud scales – VHTs – to provide such topology. Our works of 2010-2016 examine helical self-organization of moist convective atmospheric turbulence during TC formation and offer a way to solution of one of the most intricate enigmas of meteorology on tropical cyclogenesis by diagnosing a time when cyclogenesis commences as well as allow to consider an idea on controlling the formation of hurricanes at the very early stage of their evolution.

 

Thursday, 20 October 2016, 3:30 PM

Refreshments 3:15 PMNCAR-Foothills Laboratory3450 Mitchell LaneBldg 2, Main Auditorium, Room 1022

Building:
Room Number: 
1022
Will this event be webcast to the public by NCAR|UCAR?: 
Calendar Timing: 
Friday, October 21, 2016 - 3:30am to 5:00am

Helical tropical cyclogenesis: a modern look based on cloud-resolving numerical analysis of self-organization of moist convective atmospheric turbulence

 Galina Levina
Space Research Institute, Russian Academy of Sciences
Moscow, Russia

Recent results of our collaborative Russian-American efforts on how a notion of helicity can be applied in the atmospheric research to tropical cyclone (TC) investigations will be presented. Briefly recalling the role of helical turbulence in the formation of large-scale structures in magnetohydrodynamics and general dynamics of non-conducting fluids, we make an accent on the existence of threshold for large-scale instabilities in all cases. To bring together the notion of helicity and TC formation, we emphasize one of the very first achievements obtained by near-cloud-resolving numerical simulation of tropical cyclogenesis. This is the discovery of vortical nature of atmospheric moist convection in the tropical zone – rotating cumulonimbus clouds, which were dubbed ‘Vortical Hot Towers (VHTs)’ – and their crucial role in TC formation (Hendricks et al., 2004; Montgomery et al. 2006). As it was noted by Molinari and Vollaro (2010), “VHTs are helical by definition because they contain coincident updrafts and vertical vorticity”.

For the first time in TC research, we highlight the inherently helical tropical cyclogenesis. This implies the role of a special topology of the newly forming mesoscale vortex and the contribution of motions of cloud scales – VHTs – to provide such topology. Our works of 2010-2016 examine helical self-organization of moist convective atmospheric turbulence during TC formation and offer a way to solution of one of the most intricate enigmas of meteorology on tropical cyclogenesis by diagnosing a time when cyclogenesis commences as well as allow to consider an idea on controlling the formation of hurricanes at the very early stage of their evolution.

 

Thursday, 20 October 2016, 3:30 PM

Refreshments 3:15 PM
NCAR-Foothills Laboratory
3450 Mitchell Lane
Bldg 2, Main Auditorium, Room 1022

First Name: 
Caroline
Last Name: 
Haws
Phone Extension (4 digits): 
8189
Email: 
haws@ucar.edu
Building:
Room Number: 
1022
Host lab/program/group:
Calendar Timing: 
Thursday, October 20, 2016 - 3:30pm to 5:00pm

Pages