Post-processing of CFD simulations via flow visualization and data analysis techniques. Automated tools for overset grid generation.
Monitoring of flow solution convergence using automatic data extraction techniques and tools.
Assisted with proposal reviews, which led to an award of 4.75 million hours of supercomputing time to leading U.S. researchers.
Based on a highly competitive selection process, David was accepted into the NASA Leadership Development Program 2005-2006. As part of this one-year program, he completed two detail assignments outside of NASA. For one of these assignments, he served as a program manager for the Homeland Security Advanced Research Projects Agency, Science and Technology Directorate, DHS.
Collaborators:
In this project, we developed a surface wind simulator. It utilized NASA Earth science satellite
and buoy observations and a downscaling NWP model to provide boundary conditions
for fluid dynamic model runs. In addition to providing surface wind simulations other
climate parameters (temperature and precipitation) were generated.
Uncertainty Visualization
Collaborators:
NASA's Office of Earth Science sponsors research which generates multiple simulations or
measurements of the same physical field. Ensemble forecasts, outcomes from geophysical simulations,
or repeated measurements in an experiment all produce multiple instances of the same physical field.
This type of data is referred to as multi-valued data.
A level of uncertainty is implied by the existence of multiple datasets and it is important to know
what that level of uncertainty is. The situation is complicated by the fact that multiple variables
(eg. temperature, altitude, vegetation) may be involved, each with their own level of uncertainty.
Furthermore, the variable of interest may be a derived quantity (eg. Fire danger) that is not directly measured.
We investigated methods for visualizing uncertainty in multi-valued data sets.
3D Ultrasound Data Analysis
Collaborators:
In this project, we developed new techniques for visualizing 3D ultrasound data from patients
with cardiovascular diseases and conditions. The recent development of real-time 3D echocardiography
creates the opportunity to greatly improve the ability to guide minimally invasive procedures.
We evaluated the latest display technologies for a more intuitive assessment of echocardiography
for guidance of catheter-based procedures.
Investigated surface flow visualization using synthetic textures.
Surface flows are important for revealing flow phenomena near the
body surface. The challenge is to find effective methods for understanding
and representing flow features in large-scale data sets.
Comparing and Understanding Data Distributions in EOS Applications
(funded by NASA Intelligent Systems (IS) program),
2001-2004.
Collaborators:
For this research proposal, we've developed a set of techniques for understanding and comparing
data sets that are in the form of a distribution (e.g. a probability
density function) or a feature vector at each sample location.
The driving application in designing and testing our
techniques are NASA Office of Earth Science applications using
Earth Observing System (EOS)
images to derive 2D biophysical fields representing
snapshots or time-series of the Earth 's surface.
To learn more about this project, see our IS Project web site.
Scientific Visualization in Nanotechnology
Collaborators:
Researchers at NASA's Center for Nanotechnology has engineered a double-ring protein complex,
called chaperonin, that could bind specific inorganic materials and self-assemble into
two-dimensional arrays or wires in nano scales. The arrays and wires are being
investigated for use in both memory-based devices and sensors.
We developed advanced techniques to visualize the overall surface of the protein
complex as well as the interfaces between molecules within the complex using a real-time
volumetric display, and to facilitate the design of the proteins.
Protein Interactions Network Graph
Collaborators:
In this project, we developed visualization applications for exploring complex graphs of
protein interaction networks from the Protein Data Bank. There are several hundred proteins
(nodes) and thousands of protein-protein interactions (edges) in the network graphs and
each protein is categorized by its functional group.
Each edge connecting two proteins indicates that there is an interaction between the two
connected proteins.
Our application helps the scientist to control the placement of the edges connected to the selected
proteins. Since the network graphs are very complex in nature, the tasks of interactively selecting
specific protein groups and manipulating the network graphs would have been very difficult to
perform without our multimodal interface.
Innovative Cluster Analysis for Multivariate Astronomical Data
Collaborators:
We investigated a broad range of visualization methods and techniques to space science data.
This was a joint research effort with astronomers, statisticians, and computer scientists.
Some of the challenges in our research are the unstructured nature of the data,
the multi-variate fields at each multidimensional location, statistical nature of the data, and
the large-scale data size.
LidarVis is a software program for visualizing distribution data collected from lidar. LidarVis deals with the multivalue nature of the data. That is, at each grid cell, there are multiple values of a single variable. We refer to this data type as distribution data. Examples of a distribution lidar data set from forest provide information on forest structures, tree size, and density. LidarVis provides advanced query capabilities that allow the scientists to locate distributions with specific characteristics. LidarVis also allows the scientists to define the requirements of a significant bump/peak by the percentage of the area covered by the peak. Though, LidarVis was developed originally to deal with multi-return lidar data sets, it can be easily used for distribution data sets from other applications.
PDFVis is a software program for visualizing uncertainty, which can be represented by a probability density function
(PDF) located at each grid cell in a spatial domain. A full description of uncertainty at each grid cell is represented
using a probability distribution. The software allows scientists to visualize the uncertainty in the given spatial
domain by creating maps of first, second, and third order statistics summarizing the distributions.
Cursor3D is a software program for interactive tracking and selection of objects displayed
inside a 3D virtual environment. Unlike the conventional 2D mouse interface, where the user can
only move the cursor in a 2D plane, Cursor3D allows the user to move the cursor freely in 3D.
A voice and mouse input interface was implemented in Cursor3D. This multimodal interface
significantly increases the user performance in 3D interactive graphics applications.
Unsteady Flow Analysis Toolkit(UFAT) is a batch-oriented program for
post-visualization of time-dependent flow data. Scientists use UFAT to
generate streaklines, vortex cores, and other visualization.
Some of the impressive images include the descending delta wing and
the flow about the V-22 tilt-rotor aircraft. Worked with
David Kenwright,
who has implemented the algorithms for fast particle tracking and vortex
core detection in UFAT.
Graphical Line Integral Convolution (GLIC) is an interactive surface flow analysis tool
for visualizing surface flows from time-independent (steady) and time-dependent (unsteady) flow data sets.
Worked with Han-Wei Shen, Ling-Jen Chiang, and Aleksandra Kuswik on this software tool.
Animated Flow Line Integral Convolution (AFLIC) is a flow visualization program that generates
instantaneous surface flow patterns using synthetic textures.
Worked with Arthur Okada, who started as a summer intern, on this software program.
Program Comittee:
IEEE Pacific Visualization Symposium,
2008-2010
Program Committee:
International Conference on Information Visualisation,
2003-2005 and 2007-2010
Program Comittee:
First International Workshop on Super Visualization,
2008
Associate Editor:
IEEE Transactions on Computer Graphics and Visualization,
2003-2007
Program Committee:
IEEE Visualization Conference,
2001-2003 and 2006-2008
Advisor: Engineering and Applied Sciences,
National Research Council, 1998-2008
Subtopic Manager:
Small Business Innovation Research (SBIR),
Data Management and Visualization, NASA Earth Science Enterprise, 2002-2003
Co-Chair: Case Studies,
IEEE Visualization 2000 Conference, October 8-13, 2000
Guest Editor:
IEEE Computer Graphics and Applications Journal,
September/October 2000 Special Issue on Visualization
Co-Chair: Case Studies,
IEEE Visualization 1999 Conference, October 24-29, 1999
Adjunct Professor: Santa Clara University, Taught advanced computer graphics
courses in the Computer Engineering Department.
Paper Reviewer: IEEE Visualization conferences, ACM SIGGRAPH,
IEEE Transactions on Visualization and Computer Graphics,
and IEEE Computer Graphics and Applications
High End Computing Capability (HECC) Program
Past Projects
National Leadership Computing System (NLCS)
NASA Leadership Development Program
Surface Wind Simulator
Surface Flow Visualization
Software Tools
OVERSMART is a new software tool that provides a comprehensive report of solution convergence of flow computations over large complex grid systems. The new tool produces a one-page executive summary of the behavior of flow equations residuals, turbulence model equations residuals, and component forces and moments. The current implementation has been targeted for the OVERFLOW flow solver but the framework allows easy extension to other flow solvers. The software has been demonstrated to rapidly process large residual history files with millions of lines of data.
Professional Activities
Program Comittee:
Fourth International Symposium on Visual Computing,
2008-2009
Publications
