Microphysical Processes in Weather and Climate Models II
Presiding: D O Starr, NASA Goddard Space Flight Center; J G Hudson, Desert Research Institute
A42A-01 INVITED 10:30h
A New Multi-moment Microphysics Scheme
Microphysical processes affect directly the formation and distribution of precipitation. In numerical models, these processes are parameterized. In bulk schemes, the size spectrum of each hydrometeor category is often described by a three-parameter gamma distribution function. Two-moment schemes generally treat the intercept and slope as prognostic parameters while holding the dispersion parameter alpha constant. We investigated the role of alpha and its effects on sedimentation and microphysical growth rates. It is found that the size-sorting mechanism, which exists in a bulk scheme when different fall velocities are applied to advect the different predicted moments, is significantly different depending on whether a single-momnet, two-moment, or three-moment formulation is used. In addition, the instantaneous hydrometeor growth rates are also changes with the dispersion parameter. Motivated by these results, a new mixed-phase microphysics scheme, with complexity ranging from a one-moment version to a three-moment version, has been developed. Experiments on precipitation development in the context of a kinematic updraft were carried out. It is shown that the three-moment version demonstrates superior performance in terms of the vertical distribution of the hydrometeor mass content and surface precipitation rate.
A42A-02 10:45h
An efficient mixed-phase cloud and precipitation scheme for use in Operational NWP Models
An efficient cloud microphysical parameterization that includes mixed-phase processes has been running in the operational Eta model at the National Centers for Environmental Prediction (NCEP) since November 2001. This scheme is also being run in the Non-hydrostatic Mesoscale Model version of the Weather Research and Forecasting (WRF-NMM) model as part of NCEP's production suite. It is intended to contain some of the functionality of more sophisticated microphysics packages used in cloud-resolving models and high resolution mesoscale models (e.g., Rutledge and Hobbs, 1983, 1984; Lin et al., 1983; Reisner et al., 1998), while remaining computationally efficient for use within operations. Within the microphysics subroutines, the prognostic variables are mixing ratios of water vapor, (nonprecipitating) cloud water, rain, and ice. The ice is a composite category composed of small, nonprecipitating ice crystals ("cloud ice") and precipitating ice particles ("precipitation ice"). Exponential size distributions are assumed for rain and precipitation ice, however, their intercepts and slopes are variable unlike in many of the other, more sophisticated schemes. Throughout the rest of the forecast model code outside of the microphysics routines, the prognostic variables are specific humidity and total condensate. Within the microphysics driver after the microphysical calculations, the updated cloud water, rain, and ice mixing ratios are summed to give the total condensate. Local arrays in the microphysics driver store the fractional contributions of each hydrometeor class to the total condensate. These arrays are then used to extract the mixing ratios of cloud water, rain, and ice from the condensate upon entry into the microphysics driver at the next physics time step. This approach assumes that changes due to advection in the relative composition of cloud water, rain, and ice from the previous time step are small within each grid column. Although some errors are bound to occur with this approach, advecting only a single variable saves considerable computing time. Different closures are assumed in this scheme compared to other bulk parameterizations, most notably in the size distributions of precipitation ice particles. These approaches, together with some methods of optimization, will be summarized at the meeting. Supercooled mixed-phase conditions are considered in the scheme, and it includes such microphysical processes as cloud water condensation and evaporation, deposition and sublimation of ice, conversion of cloud water to rain, riming of cloud water and accretion of rain onto ice, melting of ice, freezing of rain, and ice nucleation. In addition, more sophisticated physics are considered, such as shedding of rain from melting ice particles accreting cloud droplets, condensation of cloud water onto melting ice (instantly shed to form rain), and evaporation of liquid water from melting ice. Work will soon be underway to incorporate partial cloudiness effects into the scheme following the work of Sundqvist et al. (1989), which can be important in better representing cloud-radiation interactions.
A42A-03 11:00h
Sensitivity Of Cloud Microphysical Processes On Hurricane Intensity
We have been carrying out sensitivity studies on hurricane intensity to cloud microphysical parameters. These are model studies that make use of the fifth generation non-hydrostatic Pennsylvania-National Center Atmospheric Research (PSU-NCAR) high-resolution cloud resolving mesoscale model (MM5) in a multiple nested manner, having the inner most domain with 3km horizontal resolution and with fully explicit representation of moist process. This study elucidates the role of various production and conversion terms (viz. melting, evaporation, condensation, accretion) of hydrometeors (i.e. graupel, snow and ice) in the microphysical parameterization scheme in determining the intensity changes of the hurricane. We examine the impact of over production of these frozen hydrometeors, which lead to water loading problem and also explore sensitivity of other source of uncertainties (e.g. intercept parameter, slope parameter, fall speed) of hydrometeors, which are ultimately influencing (directly or indirectly) on the precipitation and intensification forecast of Hurricanes. The high resolution global model data sets are being used for the creation of initial and boundary conditions for the model's coarsest domain, subsequently finer inner domains are constructed using the initial and boundary conditions from their respective domains. The findings of this study provide us useful information on the deficiencies of the empirical formulations of microphysical parameterization schemes for hurricane simulations.
A42A-04 11:15h
Improving Regional High-impact Weather Forecasts with a High-resolution Cloud-resolving Numerical Model
High impact weather is one of the greatest threats to human society, while an accurate forecast of these events remains a challenge for current operational models. Very often the temporal and spatial evolution of these high impact weather systems is dominated by small convective-scale motion and complicated microphysical processes. For improvements in the forecasting of such weather systems, a high-resolution cloud-resolving model with improved model physics is required. In this study, a cloud-resolving mesoscale forecast model, the Cloud Resolving Storm Simulator (CReSS), was employed to simulate observed cases lake-effect snowstorms, an important source of high impact weather over the Great Lakes region. One of the challenges in simulating these events is a correct representation of the cloud-scale processes that result in the quasi-2-dimensional clouds that are characteristic of these storms. The simulation had a very high spatial resolution, 500m in the horizontal and a stretched vertical grid with a minimum spacing of 30m, and was run in a domain of 800 km by 650 km that encompassed much of the Great Lakes region. Such a simulation remains a computational challenge and we made use of CReSS's efficient parallelization capabilities to complete the simulation using 880 CPUs on the Earth Simulator. The 8-hour simulation was completed in 2.67 hours and CReSS attained a sustained rate of 1.6 Tflops. The model simulated results are in good agreement with the field observations from the Lake-Induced Convection Experiment (Lake-ICE), in particular, the model explicitly resolved the convective roll clouds, which is crucial for the model to successfully represent this high impact weather event. Indeed, synthetic radar data derived from the model's hydrometeor fields was in good agreement with observations. The results of this study highlight the improvements in regional-scale forecasts that can be made with cloud models.
A42A-05 11:30h
Explicit Simulation of Ice Crystal Habits in 3D Eulerean Numerical Weather Prediction Model
Current operational weather prediction models simulate the ice microphysics based on Kessler/Orville approach, or categorization of solid hydrometeors, for its computational efficiency. The scheme may not be appropriate to study cloud microphysics not only due to the prescribed decision tree and arbitrary categories, but also the inability to reflect the growth history of solid hydrometeors. In addition, along with more available computational power, recent remote sensing technology needs proper representation of phase, distribution, shape, and density of solid hydrometeors in radiation transfer calculation. This research describes a methodology to simulate ice crystal habit (or shape), and density of frozen hydrometeors explicitly which can be used in 3D Eulerean Numerical Weather Prediction (NWP) Model. In Eulerean NWP model the prognostic variables are predicted using a mass conservation equation requiring the predictive variable to be extensive. To predict the evolving habit structure of cloud physics variables in such a model framework, we introduce extensive variables that contain the essential information necessary to reconstruct the probable geometry of the ice crystals: mass components and concentration-weighted length. At any given point, we can retrieve the geometry of ice crystals, such as dendrite, plate, column, needle or rosette from these variables. This method will be implemented in a bin model which represents all the solid hydrometeors by one mass distribution. Idealized orographic storms will be simulated by University of Wisconsin Nonhydrostatic Modeling System with the bin model. The result will be discussed in the presentation.
A42A-06 11:45h
On the use of ICE/SAT Lidar Space-Born Observations to Evaluate the Ability of MM5 Meso-Scale Model to Reproduce High Altitude Clouds Over Europe in Fall.
The description of ice clouds in meso-scale models has progressed significantly those last years in including improved microphysical schemes based on recent physical parametrizations deduced from observations. Recently, the first lidar in space has collected a valuable dataset that allows reaching an important step in our knowledge of occurrence and macrophysical properties of thin upper troposphere ice clouds. This study aims at evaluating MM5 capability to reproduce high altitude thin ice clouds using the Ice/SAT October-November 2003 dataset. MM5 is forced at large scale with NCEP thermodynamical profiles and ran hourly over the European continent with 0.5° spatial resolution. Space-borne lidar profiles are simulated from model outputs and compared to the observed ones at the same location and time. One month of simulations / observations comparisons show that the model reproduces correctly the clouds structures in average, but misses a significant quantity of high altitude thin clouds. When high clouds appear in the model, their ice water content seems to be overestimated. A complete analyze of this simulated / observed dataset will be presented.