Fall 2007- Spring 2007

Scientists use many different tools to study the ice sheets including radar, seismic sensors, and satellites to collect the data they need.  The Center for Remote Sensing of Ice Sheets (CReSIS) is developing a UAV (unscrewed aerial vehicle) to measure ice sheets using synthetic aperture radar. A single flight from the UAV can generate more than one terabyte of data.  Real-time field processing will allow the field team to adjust location and flight patterns to collect optimum data sets.  The large amounts of data create a significant problem for scientists who wish to process their data while in the field. 

The PolarGrid project is developing a 64-core computer cluster to allow for storing, and processing of the data collected in the field in real-time.  We will be doing research that directly benefits the PolarGrid project.  The topics include the network infrastructure that is used for clusters, file systems to utilize all the nodes of the cluster, the software that is currently in use to help in the development of parallel programs, and Beowulf clusters. 

There are currently three main software packages that we will look at that are used for parallel computing: MPI (Message Passing Interface), Condor, and BOINC (Berkeley Open Infrastructure for Network Computing).  We will look at the differences between these systems and how and what type of programs they are used for.  The advantages and disadvantages of the systems will help us to determine what software should be run on the cluster. 

The file system is an important factor in determining performance and availability of data on a cluster.   The PVFS (Parallel Virtual File System) and Lustre file systems are open source projects that have been used for linux clusters.  We will look at these file systems and others to determine what file system best fits the needs of the PolarGrid project. 

A Beowulf cluster is a specific type of cluster that was designed with the use of readily available, inexpensive hardware.  This type of cluster has been around for a long time and there is a lot of documentation as well as an online community to help in the building and maintaining of these clusters.  These resources will be very helpful for guiding our research. 

Summer 2007

PDF
Research Page
Powerpoint

Modeling and Simulation of the Seismic TETwalker Concept

Abstract
The objective of this project was to adapt the design of a robot that was originally created at NASA Goddard Space Flight Center called the TETwalker, in a computer simulation software program to demonstrate the collection of seismic data of ice sheets in Antarctica and Greenland. We will take their design and adapt it for seismic data collection by placing seismic sensors (geophones) in each ground node of the tetrahedral structure, or in the center node for deployment. Seismic methods are analyzed in order to determine which design could possibly be more efficient and reliable in polar environments in terms of geophone deployment and environmental characteristics.

Fall 2006 - Spring 2007

PDF
Research Page
Powerpoint

Correlations Between the Concentrations of Chlorophyll a in Surface Waters and Dissolved Oxygen in Bottom Waters of the Northern Gulf of Mexico

Abstract
Hypoxia is an environmental condition of low concentrations of dissolved oxygen that is detrimental to marine animals. It is a result of excessive input of fluvial nutrients to the coastal environment that leads to rapid growth of phytoplankton. When phytoplankton die and sink, they consume dissolved oxygen during decomposition. Water column stratification, especially a strong stratification in summer, is essential for the development of hypoxia. Seasonal occurrence of hypoxia has been observed in many coastal waters. The most prominent hypoxia in the United States was observed in the Northern Gulf of Mexico. For more than two decades, scientists have been monitoring the hypoxia in the Northern Gulf of Mexico by conducting annual ~ 2-week-long cruise in summer and developing strategies for reducing the flux of nutrients and the size of hypoxia. There are many shortcomings in the current monitoring method. The major shortcoming is poor temporal and spatial resolutions of these field studies.

We conducted the following research project to overcome some of these shortcomings and to lay a foundation for high resolution monitoring of hypoxia with satellite remote sensing. The project was based on the hypothesis that there is a linear correlation between the concentrations of chlorophyll a in surface waters and dissolved oxygen in bottom waters. To test this hypothesis we : (1)* obtained SeaWiFS data and processed it for sea surface chlorophyll a; (2) obtained dissolved oxygen data from NOAA’s National Coastal Data Development Center (NCDDC); (3) selected chlorophyll a data that coincided with dissolved oxygen data; (4) correlated the concentrations of dissolved oxygen in bottom waters to that of chlorophyll a in surface waters; (5) correlated the water column apparent oxygen utilization (AOU) to concentrations of chlorophyll a in surface waters. If our hypothesis is proven true, the correlations to be derived will enable monitoring of the future evolution of hypoxia in the Northern Gulf of Mexico with satellite remote sensing. This will significantly increase both temporal and spatial resolutions of hypoxia mapping. The correlations could also be used to evaluate the past evolution of the hypoxia in the Northern Gulf of Mexico (i.e. from 1997 when SeaWiFS was launched).

Summer 2006

PDF
Research Page

Airborne Measurement of Snow Thickness Over Sea Ice

Abstract
Snow cover on sea ice plays an important role in the climate of the polar regions. Snow on the sea icereduces the heat exchange between the ocean and the atmosphere by its high albedo (reflectivity) and low thermal conductivity. The lower the albedo, the less solar energy is reflected back into the atmosphere. This energy is absorbed into the ocean. The warmer water will melt more sea ice, and eventually the warmer atmosphere above the warmer water will melt more of the sea ice in the polar regions. Better data on the extent and thickness of snow cover are therefore needed to understand thecondition and future behavior of sea ice.

Up until recently, the only practical means of observing snow cover over sea ice was by satellite remote sensing. The Advanced Microwave Scanning Radiometer (AMSRE) onboard NASA’s Aqua satellite does precisely this. To validate the measurements made by AMSR-E, the University of Kansas developed an ultra-wideband frequency-modulated continuous-wave (FM-CW) airborne radar to measure snow thickness over sea ice. This system was flown over the Arctic sea ice in March 2006 to measure the snow thickness.

Fall 2005- Spring 2006

Research Page
Powerpoint

Using Passive Network Discovery to Fingerprint Vulnerabilities within Ethernet Broadcast Frames.

This paper examines how open source embedded network tools were used to perform persistent internal audits of Ethernet Local Area Network broadcast traffic. The initial requirements to define the project phases were developed based on the analysis of each open source learning stage. Open Source UNIX version, Unbuntu, was selected as the platform to prototype because of its ease of use and usable business productivity, internet, drawing and graphics applications. To understand why hosts within the ONR LAB were experiencing a decrease in system performance and transmission speed. A Passive Network Discovery of Ethernet Broadcast Frames was captured and analyzed to determine if Local Area Network traffic between the local and foreign hosts is malicious or valid. The identification of remote active nodes and their system information was collected to build a resource map of all remote hosts requesting services from hosts within the ONR Lab and listing of local hosts listening ports and services running on those ports. The passive analysis approach was selected by the ONR UNIX Network Administration Team, because the collection of active LAN traffic would be not impact ECSU’s LAN/WAN assets. Moreover, this paper goal is to show that persistent packet monitoring of Ethernet traffic can identify weaknesses that reduce LAN performance and possibly harm valuable assets used to support major and/or general support systems.