Justin R. Davis

On the Design of Virtual Machine Sandboxes for Distributed Computing in Wide-area Overlays of Virtual Workstations

With recent advances in virtual computing and the revelation that compute-intensive tasks run well on system virtual machines (VMs), the ability to develop, deploy, and manage distributed systems has been ameliorated. This paper explores the design space of VM-based sandboxes where the following techniques that facilitate the deployment of secure nodes in wide-area overlays of virtual workstations (WOWs) are employed: DHCP-based virtual IP address allocation, self-configuring virtual networks supporting peer-to-peer NAT traversal, stacked file systems, and IPsec-based host authentication and end-to-end encryption of communication channels. Experiments with implementations of single-image VM sandboxes, which incorporate the above features and are easily deployable on hosted I/O VMMs, show execution time overheads of 10.6% or less for a batch- oriented CPU-intensive benchmark.

Architecture of a Community Infrastructure for Predicting and Analyzing Coastal Inundation

The Southeastern Universities Research Association (SURA) has advanced the SURA Coastal Ocean Observing and Prediction (SCOOP) program as a multi-institution collaboration to design and prototype a modular, distributed system for real-time prediction and visualization of the coastal impacts from extreme atmospheric events, including hurricane inundation and waves. The SCOOP program vision is a community “cyberinfrastructure” that enables advances in the science of environmental prediction and coastal hazard planning. The system architecture is a coordinated and distributed network of interoperable, modularized components that include numerical models, information catalogs, distributed archives, computing resources, and network infrastructure. The components are linked over the Internet by standardized web-service interfaces in a service-oriented architecture (SOA). The design philosophy allows geographically disparate partnering institutions to provide complementary data-provider and integration services. The overall system enables coordinated sharing of resources, tools, and ideas among a virtual community of coastal and computer scientists. The distributed design builds on the notion that standards enable innovation, and seeks to leverage successes of the World Wide Web by creating an environment that nurtures interaction between the research community, the private sector, and government agencies working together on behalf of the nation.

Circulation and Flushing in the Lagoonal System of the Guana Tolomato Matanzas National Estuarine Research Reserve (GTMNERR), Florida

Abstract Circulation and flushing inside the lagoon system of Guana Tolomato Matanzas National Estuarine Research Reserve (GTMNERR or GTM) have been studied using a three-dimensional numerical circulation model CH3D. The lagoon system includes two tidal inlets (St. Augustine Inlet and Matanzas Inlet). To ensure accuracy in model results, we extend the model domain to include a large portion of the coastal water and the Ponce de Leon Inlet in the south. Water levels on the open boundaries are provided by the East Coast (2001) ADCIRC Tidal Database. Model simulations of barotropic and baroclinic circulation from April 1 to May 31, 2004, produced reasonable water levels at numerous stations inside the GTM. Simulated salinity results are not as good because of the lack of freshwater inflow data inside the GTM and salinity data offshore. Using the simulated flow fields, we solve the three-dimensional transport equations for conservative species, and determine the flushing characteristics inside the GTM in terms of the 50% renewal time of the conservative species within each of eight segments, which are selected by considering the geographical features and proximity to tidal inlets and rivers. The flushing results indicate that tide is the most dominant flushing mechanism, while river and salinity are important flushing mechanisms for segments that are far from the tidal inlets. The normalized flushing times, defined as the 50% renewal time divided by the volume of the segment, are calculated for the eight segments and compared with each other. Comparing the “normalized” flushing time at all segments, a relative flushing ranking (RFR) is generated that ranks the “normalized” flushing time from the shortest to the longest as follows: segment 2 (includes St. Sebastian), segment 8 (near Ponce Inlet), segment 3 (includes Fort Matanzas), segment 7 (near Ponce Inlet), segment 1 (includes Pine Island), segment 4 (includes Pellicer Creek), segment 6 (includes High Bridge Road), and segment 5 (includes Bings Landing). This quantitative ranking of flushing characteristics inside the GTM is made possible because of the use of a three-dimensional numerical circulation and transport model that incorporates the effect of hydrodynamics on flushing. These results provide much more quantitative information than the simple empirical residence time indices (1–4) developed for the GTM in a previous study. CH3D was also applied to simulate the circulation during January 25 to February 3, 2006. Simulated currents inside the St. Augustine Inlet on February 2, 2006, compare favorably with the currents measured by Acoustic Doppler Current Profiler (ADCP). Model simulated flow rates through St. Augustine and Matanzas inlets reproduced measured data on July 1 and June 16, 2004, respectively. Model-simulated currents and water levels improved when the flooding-and-drying version of the model was used. The three-dimensional modeling approach can be used to provide sguidance on resource management and the development of sampling strategies for several ongoing and prospective biogeochemical studies in the GTM.

Toward the Probabilistic Simulation of Storm Surge and Inundation in a Limited-Resource Environment

Abstract To create more useful storm surge and inundation forecast products, probabilistic elements are being incorporated. To achieve the highest levels of confidence in these products, it is essential that as many simulations as possible are performed during the limited amount of time available. This paper develops a framework by which probabilistic storm surge and inundation forecasts within the Curvilinear Hydrodynamics in 3D (CH3D) Storm Surge Modeling System and the Southeastern Universities Research Association Coastal Ocean Observing and Prediction Program’s forecasting systems are initiated with specific focus on the application of these methods in a limited-resource environment. Ensemble sets are created by dividing probability density functions (PDFs) of the National Hurricane Center model forecast error into bins, which are then grouped into priority levels (PLs) such that each subsequent level relies on results computed earlier and has an increasing confidence associated with it. The PDFs are...

A Real-Time Forecasting System for Hurricane Induced Storm Surge and Coastal Flooding

Hurricanes cause property damages and loss of lives in the U.S. every year. In 2004, four major hurricanes caused more than 5 billion dollars in damage each in Florida alone. Katrina in 2005 caused more than

A Regional Testbed for Storm Surge and Coastal Inundation Models - An Overview

200 billion dollars in damage. Major damage caused by a hurricane is usually associated with storm surge and coastal flooding. This paper presents CH3D-SSMS, a robust storm surge and coastal flooding forecasting system for tropical and extratropical storms. The basic forecasting system couples a storm surge and coastal flooding model (CH3D) with a shallow water wave model (SWAN) in a high resolution (50–500m) coastal/estuary/inland model grid. The coupled surge-wave model receives open boundary conditions of surge and wave from a global surge model (ADCIRC) and a global wave model (WAVEWATCH-III) which cover the Gulf of Mexico and Western Atlantic. Currently high resolution grids have been set up for the East Florida Coast (including St. Johns River, Indian River Lagoon, and adjacent coastal water), Tampa Bay and Charlotte Harbor, Florida Panhandle, and Chesapeake Bay and adjacent coastal waters. The forecast wind and atmospheric pressure information are provided by a number of NOAA wind models including NAM (North Atlantic Mesoscale), GFDL-Hurricane model, and an analytic model which uses GFDL hurricane parameters. Based on the forecast wind, the CH3D-SSMS produces an 84-hour forecast of water level, wave height/period/direction, flow field, and maximum of maximum (MOM) water level and inundation, every 6 hours. Prior to forecasting, the model performs a 24-hour nowcast using analysis wind. Performance of the CH3D-SSMS forecasting system has been demonstrated using results from 2003 (Isabel) and 2004 (Frances and Ivan) hurricanes. Forecasting of the impact of Hurricane Wilma on the Charlotte Harbor region is illustrated to reflect the current capabilities and weaknesses of the CH3D-SSMS as well as hurricane wind model. To improve the forecasting system, more accurate and efficient hurricane wind models and coupling of models of various processes and scales are needed. Hurricane-ground interaction should be incorporated into the hurricane wind model to produce more realistic near ground wind after hurricane landfall. Cyberinfrastructure should also be employed to facilitate regional and community collaboration.

Comparing Standard to Feature-Based Meteorological Model Evaluation Techniques in Bogotá, Colombia

Since 2008, a Regional Testbed has been comparing storm surge models in terms of historical storm simulations and coastal inundation maps, e.g., Flood Insurance Rate Maps and Surge Atlas. The models include two structured grid (CH3D and POM) and two unstructured grid (ADCIRC and FVCOM) models. During 2008, the storm surge and coastal inundation in the Chesapeake Bay and the Outer Banks of North Carolina during Hurricane Isabel was simulated and the results compared in an independent but non-interoperable effort by partners. In 2009-2010, an additional model SLOSH was added, and all five models were used to simulate the storm surge and coastal inundation in southwest Florida during Hurricane Charley and the results compared. Model inputs and outputs were designed in an interoperable fashion, using common model input data, parameterization and coefficients, common model output formats using a common model data grid. Thirty scenarios were developed to test the sensitivity of the models to bathymetry, storm forcing, wind drag coefficient, bottom friction, Coriolis, 2D vs. 3D formulation, etc. Various types of model products, including time series of storm surge and maximum inundation over the entire model domain, were compared to each other and measured data. The detailed model simulations and comparisons required considerable computational and analysis time, but resulted in the discovery of how model features affected the model accuracy, leading to an overall improvement of all the models used. Testbed results showed differences in storm surge elevation and coastal inundation during both Isabel and Charley. While the simulated water level at the observed stations generally did not differ by more than 20% and no model appears to be consistently superior / inferior to any other model, there are more significant differences in the produced inundation maps. The computational efficiency differs considerably among the various models. Additional simulations of a large number (20+) of storms and domains are needed to better define the relative importance of different model parameters and to sort out the causes for subtle differences among the model results. More in-depth model inter comparison results will be forthcoming in a future paper.

Integrated-Process and Integrated-Scale Modeling of Large Coastal and Estuarine Areas

AbstractStandard meteorological model performance evaluation (sMPE) can be insufficient in determining “fitness” for air quality modeling. An sMPE compares predictions of meteorological variables with community-based thresholds. Conceptually, these thresholds measure the model’s capability to represent mesoscale features that cause variability in air pollution. A method that instead examines features could provide a better estimate of fitness. This work compares measures of fitness from sMPE analysis with a feature-based MPE (fMPE). Meteorological simulations for Bogota, Colombia, using the Weather Research and Forecasting (WRF) Model provide an ideal case study that highlights the importance of fMPE. Bogota is particularly interesting because the complex topography presents challenges for WRF in sMPE. A cluster analysis identified four dominant meteorological features associated with air quality driven by wind patterns. The model predictions are able to pass several sMPE thresholds but show poor performa...

“Grid”-based Particle Tracking in Florida Bay

To predict the response of coastal and estuarine ecosystems to anthropogenic and natural changes, it is necessary to conduct integrated-process and integrated-scale modeling of large coastal and estuarine areas. This paper presents an integrated modeling system, CH3D-IMS, which includes models of circulation, wave, particle trajectory, sediment transport, water quality dynamics, light attenuation, and seagrass dynamics. The CH3D-IMS has been and continues to be validated with data from various estuaries in Florida. A 3-D variable-density groundwater flow model and a fishery model are being coupled to the CH3D-IMS. This paper presents example applications of the CH3D-IMS including: (1) simulation of the Indian River Lagoon and trajectory of Shuttle Columbia debris in North and Central Florida Atlantic Coastal water; (2) simulation of storm surge in Tampa Bay, Sarasota Bay and adjacent Gulf of Mexico; and (3) simulation of circulation in Charlotte Harbor and adjacent Gulf of Mexico water. As the integrated modeling system continues to be applied to ever more complex problems over increasingly larger coastal areas, it requires more computational resources and disciplinary expertise which are often unavailable in any single institution. To facilitate integrated-process and integrated-scale modeling by multiple institutions, the development of an infrastructure, a regional modeling “grid” is proposed.

Multi-Objective Optimal Design of a Building Envelope and Structural System Using Cyber-Physical Modeling in a Wind Tunnel

In recent years, grid computing has emerged as an attractive approach to solve large problems through the coordinated sharing of resources. While successful in other fields, e.g. high energy physics, few studies focusing on coastal and estuarine problems have been performed; thus, a pilot study was performed to illustrate how coastal and estuary modeling could use grid computing technologies. To this end, several costal and estuarine models were dynamically coupled and then executed in a local computational grid environment composed of Silicon Graphics computers connected by a fast ethernet network. The models: a robust hydrodynamic model (CH3D), a particle tracking model, and a visualization model communicated through the network using a grid-enabled version of MPI, MPICH-G2. The grid-enabled, coupled modeling system was then tested on a real estuary, Florida Bay, USA. Using simulated hydrodynamics (Node 1), the trajectories of 2000 particles were simulated by two independent instances of the particle tracking model (Nodes 2 and 3). Updated particle positions were then visualized (Node 4) and compared with those calculated with a statically coupled version of the modeling system running on a single node. The dynamically coupled system was shown to reproduce the particle trajectories exactly with some speedup due to the two particle tracking models running in parallel. A graphical interface to the modeling system was then created using the advanced web portal and grid middleware software In-VIGO. The robustness of the results shown along with relative ease of converting an existing modeling system into a dynamically coupled system running in a grid environment demonstrated the feasibility of using grid technologies on coastal and estuarine modeling.