Weather Modeling for Improving National Security and Coupled-Model Applications (Cont.)

advanced

Specific Priorities

Specific priorities are spread across a broad spectrum of applications that all require a heavy emphasis on directed research and engineering, and a moderate emphasis on basic scientific research. The primary priority areas are listed below.  Sponsor’s budgets and specific operational requirements will fine tune these priority areas during this planning period.

Urban Meteorology

Urban meteorology has been a long-neglected area of study, in spite of the fact that most of the world’s population lives in metropolitan areas and this trend is increasing.  A primary approach is to better understand the influence of urban areas on the meteorology, on scales from buildings and street canyons to that of the entire city.  RAL will strive to both improve its general knowledge of urban weather as well as to allow more accurate calculations of the transport and diffusion (T&D) of hazardous material accidentally or intentionally released at the surface within or near a city.

On smaller scales, the buildings channel winds and generate turbulence, causing large departures between street-level motions and those on the larger meso- and synoptic-scales.  The thermal effects of the buildings are also important and must be taken into account.  Simulation of winds on the building scale must be done with Computational Fluid Dynamics (CFD) models; RAL’s objective is to build experience with these tools, both for operational prediction as well as for forensic analysis of historic cases

On larger scales, mesoscale models must be able to accurately represent the bulk effect of urban areas on the atmospheric dynamics and thermodynamics.  To accomplish this, small-scale processes, such as radiation-trapping in street canyons, must be parameterized using an Urban-Canopy Model (UCM).  RAL’s strategic approach will be to improve UCMs and make them an integral part of the WRF-based modeling system.

Because urban influences on weather are multi-scale, and these scales interact with one another, another strategic approach for RAL is to develop coupled mesoscale-model/CFD-model systems that treat scales from the bulk effect of the city to street-canyon flow.  Because of the typically different numerical and physical constructs of mesoscale and CFD models, this coupling will be especially challenging.

The CFD models, with grid increments as small as a few meters to allow resolution of wind patterns in street canyons, are generally extremely computationally intensive, which makes their use for real-time prediction problematic.  Increasing the computational efficiency is a strategic approach that RAL will pursue, and will be addressed in terms of finding faster alternatives to traditional CFD models, as well as making the CFD codes more efficient on parallel computing hardware.

Coupling urban-scale meteorology to transport and dispersion models

RAL is concentrating on developing defense, homeland security, and air quality applications that diagnose and predict the movement of hazardous materials that are released into the atmosphere. One of RAL’s central strategic approaches is to transition the successfully demonstrated Pentagon Shield building protection prototype to an operational system residing at the Pentagon.

Another approach is to support additional development of the Variational Lidar Assimilation System (VLAS) to improve its high resolution forecast capabilities. VLAS currently provides an analysis of the current non-building-aware wind field, but work needs to done on the specification of its lateral boundary conditions to improve its forecasts.

A challenging and needed capability is real-time CFD (e.g., 4-m resolution) flow analysis and prediction for urban areas, particularly for buildings. CFD models have been used to analyze wind flow around a single building for the Pentagon Shield project, and there are plans to extend this capability to the neighborhood scale.

A major effort will soon be launched to develop an urban transport and dispersion test bed that will provide an efficient infrastructure for developing and showcasing new transport and dispersion applications. This will include an Observing System Simulation Experiment (OSSE) capability that will allow for objectively analyzing applications and determining system sensitivity to sensor inputs.

Through the use of data fusion techniques it will be possible to improve source term characterization for initialization of transport and dispersion models. This will also improve the prediction of the plume’s movement and morphology in open terrain or urban environments.

Other plans include gaining additional expertise in a variety of T&D models, including ones such as CALPUFF, which is used primarily for air quality applications. Some of the models currently being applied across our projects have licensing and export limitations, so by exploring other models with fewer restrictions, other opportunities could present themselves, such as international technology transfer. Greater expertise in air quality modeling could also be useful for applications in programs such as the FAA JPDO.

Understanding and predicting the structure of boundary layers over a variety of surfaces

Boundary-layer (BL) properties are significantly influenced by various characteristics of the surface, such as vegetation cover, soil moisture and other substrate properties, and orography.  These surface factors will influence BL turbulence intensity, depth, static stability, and thermally forced circulations.  Such BL responses to surface heterogeneity are presently poorly understood and rarely validated in model predictions.  A specific priority of RAL’s research is to evaluate and understand the predictability of these surface-forced BL properties because they so strongly influence the transport and diffusion of contaminants released at the surface.  Of special relevance is predicting the urban BL because of the greater likelihood of an accidental or intentional release of hazardous material there, and the greater population density.  On the city scale, better understanding and modeling are needed of BL meteorology response to the heterogeneity associated with the contrast between the urban surface and that of the surroundings.  Of special importance is the nocturnal BL and how it differs between urban and rural areas as a result of substrate-property differences.  These landscape-forced effects on the BL will clearly influence the T&D on the scale of the city and its neighborhoods.

Because large population centers are often located near coastlines, special attention must be paid to the influence of land-water contrasts on BL structure throughout the diurnal cycle.  This has been well studied for simple coastlines, but many coastline configurations are complex, as are the substrate variations in littoral zones near the coast.  These factors can cause sharp contrasts in BL properties near coastlines, including internal boundary layers, making the T&D of hazardous material challenging to understand and predict. 

The knowledge gained about the sensitivity of the BL response to surface heterogeneity will allow developers to perturb land-surface conditions, within their range of uncertainty, in order to generate mesoscale-model ensemble members that contain sufficient spread near the surface (within the BL).  Developing this variability is essential for estimating T&D uncertainty with ensemble forecasts using coupled mesoscale and T&D models.

Model verification, targeted to the user

Verification is just as important for the applications here as it is for the aviation sector discussed earlier.  The deployment of very fine resolution models, in particular, is leading the research community to consider new ways to quantify their accuracy, since it is believed that the older methods of considering rms errors and two-by-two contingency tables fails to quantify the advantages.   The RAL plans in this area are presented in the later section “Model Verification and Quality Assessment.” 

Ensemble forecasting

A specific priority in this development area is to establish an advanced meso- and fine-scale data assimilation and forecast system by combining a number of cutting-edge achievements in data assimilation technology in order to address unique issues of mesoscale meteorological processes and mesoscale modeling.  This development is focused on producing the best possible current analyses and most accurate short-term forecasts of regional to local scale weather, and quantitative forecast uncertainties.

This problem will be addressed via a phased-in nudging/ensemble hybrid.  Steps include the following:

Stochastic parameterization in ensemble systems

The goal is to conduct a combination of isolated, off-line modeling experiments and full 3D ensemble prediction experiments to develop relevant stochastic approaches to parameterization schemes.

The closure problem for NWP models will remain despite computational advances facilitating fine-scale simulation and forecasting.  Current closure (parameterization) methods to represent processes such as PBL turbulence, moist convection, and cloud microphysics are built on the assumption that resolved and sub-grid scales can be separated, and that the sub-grid scale can be parameterized as a function of the resolved flow.  Advances in computational capabilities currently allow real-time simulation and forecasting at resolved scales that do not meet this assumption, and future parameterization will need to be much different than currently.  Because many of these processes have components that appear random, stochastic approaches are likely to be fruitful, but the research is immature, particularly so in the PBL, which is relevant to RAL model users. RAL will undertake a multi-year research and development effort to focus on stochastic parameterization in the PBL.

Mesoscale and convective-scale Quantitative Precipitation Forecasting (QPF)

Moist dynamics are important components in mesoscale analyses and forecasts. The moist process not only takes a dominant role in the convective weather life cycles, but also dramatically modulates other weather processes such as frontal precipitation systems and thermally-forced circulations by underlying heterogeneities. Mesoscale and convective-scale QPF relies critically on accurate mesoscale weather analyses of cloud and precipitations in the model initial conditions. Remote-sensing measurements of radars, satellites and some other instruments provide substantial information about the atmosphere cloud and precipitation properties, including their spatial distribution, densities and hydrometeor categories.   

RAL will build an effective cloud and precipitation analysis system that incorporates various non-conventional cloud and precipitation measurements to produce dynamically and thermodynamically balanced, moist atmospheric states and to use the analyses to initialize mesoscale NWP model and improve the mesoscale QPF.

Fine-scale NWP

Fine-scale weather processes are strongly controlled by the local-scale terrain and underlying thermal forcing. Properly resolving and representing fine scale terrain structures and the underlying land-soil properties is central to achieving accurate fine-scale NWP. In order to better resolve detailed terrain structures and land-soil properties, NWP models should make use of grid scale increments as fine as computation platforms allow. On the other hand, given a computing capacity and a reasonable model grid size, a very important and neglected research topic is how to represent the underlying forcing factors in the models to produce representative local-scale weather. 

A high priority within RAL is togreatly advance meso- and small scale NWP and data analyses in various geographic regions, especially over complex terrain. To provide valuable scientific understanding on and also technical tools to represent the predictability and uncertainties in meso- and small scale weather processes. This task will be implemented by developing advanced algorithms to adequately represent and simulate small-scale underlying forcing with fine grids.

Coupling Numerical Weather Prediction output to secondary applications

There are many situations where decision makers need real-time knowledge of processes that are forced by the atmosphere, and this information must be obtained from secondary-applications models that are loosely coupled (non-interacting, separate software packages) or closely coupled with the NWP models.  Secondary-application models that have been loosely coupled with MM5 and WRF have simulated sound propagation, ballistic trajectories, parachute drift, and stream flow. The suite of new secondary-application capabilities that will be used with WRF to serve future needs includes models for simulating

The aerosol and air-chemistry models driven by WRF will serve state and Federal air-quality and public-health communities. Because of the advantages of modeling two-way interacting physical systems, fully integrated capabilities such as WRF-Chem will be used where possible. In addition, the complete treatment of surface hydrological processes in the WRF Regional Water Cycle Model will be used for forecasting river discharge.

Leveraging latest software and hardware technology to deliver information-rich weather and climate information to end users and decision makers

One of RAL’s strengths is its commitment to a disciplined software engineering practice, and its tendency to balance projects with both scientific and engineering involvement, even when the latter is not specifically called for by sponsors. This breadth of expertise is somewhat unusual in the atmospheric sciences community, and should be considered a crucial asset for the future, as ready access to weather, climate, and decision support information is becoming coveted and demanded by many of RAL’s sponsors.

The future for the engineering aspects of projects in the atmospheric sciences community is to develop strategies and infrastructure to manage and exploit large volumes of model, observation, and geographical datasets, and to develop algorithms and display tools that reduce and communicate complex data and information in ways that are well-matched to a wide variety of end users. For example, atmospheric transport modeling scientists, who routinely examine plume concentration and dosage plots, normally assume that the information contained within such plots are intuitively obvious to anyone who needs to interpret them. But the gap is wide between such scientists and the rescuers who must decide within seconds what sort of chemical suit to don in response to an incident where certain key parameters may be unknown. The latter seeks a yes/no type of guidance based on the best analysis that science and technology can bring to bear.  This is yet another example of RAL trying to focus on the ultimate end users.