Lloyd Treinish's home page

Lloyd A. Treinish
IBM Thomas J. Watson Research Center
Yorktown Heights, NY 10598
914-945-2770 (voice)
914-945-3434 (fax)
lloydt@us.ibm.com
 

I am a research staff member in the Advanced Computing Technology Center of the Deep Computing Systems Department at the IBM Thomas J. Watson Research Center in Yorktown Heights, NY.  I am in a group focusing on systems, techniques, meteorology and applications of numerical weather prediction, which is part of the Deep Computing Institute at IBM Research.  Previously, I was in the Mathematical Sciences Department, and before that, the Visual Analysis group in the Visual Technologies Department.  This work included techniques, architectures and applications of scientific data visualization and methods of scientific data management.  My research interests range from visualization systems, visualization design, data fusion, data models, and perceptual rule-based tools to study of atmospheric and space physics phenomena, and cartography.  I have been at IBM Research since 1990.

Earlier I did related work for over a decade at NASA/Goddard Space Flight Center in Greenbelt, MD.  In some sense, I have been moving closer to home over the last 25 years.  I started off doing radio astronomy (VLBI) and planetary work in school, and initially at Goddard.  Then I moved into space physics and upper atmospheric studies.  Throughout this period, I did work in applied computer sciences, particularly in visualization, data management, and data and information systems because efforts in these areas were needed to more effectively pursue the physical sciences work.  Here at IBM Research, these activities have continued, but with new applications in the troposphere and on the ground as well as in other disciplines.  To illustrate some of these efforts, I have included a number of representative visualizations below.  If you would like to examine any of the images in more detail, click on each to display a higher-resolution JPEG-compressed 24-bit image instead of the in-line, 8-bit quantized GIF.  Many of them are accompanied by other media such as animations with pointers to more detailed information.

I head an effort with colleagues here at IBM Research on local weather forecasting tailored to weather-sensitive business operations, which has been dubbed Deep Thunder.  This was an outgrowth of earlier work done in collaboration with the National Oceanic and Atmospheric Administration.  An example appears to the left, which was used by the National Weather Service to help accurately forecast a prediction of no rain during the closing ceremonies of the 1996 Olympic Games in Atlanta, particularly via animation.  Currently, I am looking at applications and extensions of the visualization techniques and the precision weather forecasting in a number of areas including emergency management, surface transportation, aviation, agriculture, broadcast, energy and insurance.  You can learn more about the overall effort by visiting: http://www.research.ibm.com/weather. If you would prefer to start by examining details about just the visualization portion of this effort, then go directly to a site that discusses the subject.  It contains a few papers, one of which focuses mostly on the problems of designing effective visualizations by considering the prerequisite tasks using the application of weather forecasting as an operational testbed.  You can learn more by visiting the following site:

http://www.research.ibm.com/weather/vis/w_vis.htm  

You can also start by experimenting with this example visualization as either VRML or as a PanoramIX scene.

You can also try a little "demo" of some of these ideas in task-specific visualization design as applied to applications of mesoscale forecasts as well as to physics simulations.
Part of the work is to put all of these ideas into operational practice, which also serves as a testbed for further efforts in high-performance computing, systems integration, automation and visualization as well as various business applications.  In that regard, we put together an operational prototype initially for the the New York City metropolitan area to 1km resolution.  An example appears to the left, which shows an operational prediction of a thunderstorm.  The results were available about 15 hours before the actual storm, and were confirmed very well with radar observations from the National Weather Service.  This operational capability provides several interactive visualization capabilities within our weather laboratory here at the IBM Research as well as products automatically disseminating on the web, a sample of which you see at http://www.research.ibm.com/weather/NY/NY.html .  This prototype has been extended with operational forecasts for the Baltimore, Chicago, Kansas City and Washington, DC metropolitan areas to 2km resolution.  In addition, experimental forecasts for the San Diego metropolitan area at 1km resolution and the Miami-Fort Lauderdale area at 1.5km are produced. 

I try to stay somewhat active in the academic community, and have published papers in a variety of different forums.  I am a member of the IEEE Computer Society (IEEE-CS), the IEEE-CS Technical Committee on Graphics and Visualization, the IEEE-CS Visualization Conference Committee (e.g., Visualization '95, '98 and '99 program co-chair), the Association for Computing Machinery (ACM), ACM SIGGRAPH, the Planetary Society, and the American Geophysical Union.  I was also a co-editor of the Visualization Viewpoints in IEEE Computer Graphics and Applications

An important aspect of developing visualization applications is in the design of the actual pictorial content.  I have been working with another colleague here at IBM Research to create more efficient methods to design visualizations, by introducing rules to the process under user control and interaction.  To date, we have focused on perceptual rules, particularly in color.  In the spirit of clever acronyms, from which you fill in the blanks, we call this approach PRAVDA for Perceptual Rule-based Architecture for Visualizing Data Accurately.  We have developed a rule-based advisory tool for designing colormaps using these principles.  Consider, for example, the image below, which shows two views of the same data set of topography and bathymetry observations.  The only difference between them is the choice of colormap.  The one on the left uses a popular "rainbow" colormap while the one on the right was derived from our PRAVDA efforts.  Which picture do you think best represents the underlying data?

If you would like to learn more about our work and these principles, then you can begin by reading the following three papers:

http://www.research.ibm.com/dx/proceedings/pravda/truevis.htm

http://www.research.ibm.com/dx/proceedings/pravda/index.htm

Why Should Engineers and Scientists Be Worried About Color?

One of my favorite topics is cartography and the use of cartographic methods in the visualization and analysis of data.  In particular, I have spent some time on the applications of these ideas to correlative visualization and studying global stratospheric ozone.  To illustrate this, consider the image to the left or a related animation.  These show the data in an orthographic map projection for the earth's southern hemisphere.

Alternatively, consider a spherical projection of similar data.  Both of these representations show the annual depletion of ozone in the Antarctic spring (i.e., the ozone hole).

You can also view an animation similar to this representation or a simplified VRML representation of this image.  In addition, you can view a video  (3:17 in length) that discusses this work.  A second video (4:47 in length) discusses an application with ozone and atmospheric dynamics data.  Both videos are MPEG-1 encoded at 800Kbps, and include my voice-over narration of the animations that describe the data, the visualization techniques and the results.

The spinning globe at the beginning of this page was created using these cartographic methods.  If you click on it, you can see it at higher resolution as an MPEG animation.  Alternatively, if you would like to play with a geometric representation of the globe, you can do so via VRML.  This geometry, however, has been simplified from the original 30-minute resolution data in order to make the file of more manageable size.  You can learn more about cartography and some of my work with ozone and other upper atmospheric science data by visiting the following two sites:

http://www.research.ibm.com/dx/proceedings/cart/index.htm

http://www.research.ibm.com/dx/proceedings/cart/ozone.htm

You can also see some other examples of my earlier work, which utilizes many of the ideas discussed in the sites I have already referenced.  The first utilizes rain gauge data to gain a qualitative understanding of mesoscale storms induced by El Niño in northwestern Peru, as illustrated in the image below.

You can learn more by visiting the following site:

http://www.research.ibm.com/dx/proceedings/peru/index.htm

You can also view an animation of these data or a VRML representation of this image.  In addition, you can view an MPEG-1-encoded video (2:03 in length) that discusses this work, which includes my voice-over narration of the animation.

The second example discusses a sounding rocket experiment observing the mesosphere for responses to energetic particle precipitation as illustrated in the image below.

You can learn more by visiting the following site:

http://www.research.ibm.com/dx/proceedings/mae/index.htm

You can also view a VRML representation of this image.

All of this work has involved the use of IBM Visualization Data Explorer.  This widely-used, leading-edge software package continues to evolve and address some of the most interesting problems in visualization.  I came to IBM Research in 1990 to continue work in visualization from where I started at NASA/GSFC.  The efforts by the group that I joined became Data Explorer.  The opportunity to participate in such a project and to work on such exciting problems has been very gratifying.  If you would like to learn more about Data Explorer, you can start with a paper that discusses the software.  It is dated at this stage, since the package has considerably greater capabilities than are discussed there.  But it will give you some basic information about Data Explorer.  To learn more about the current version of the software, see what people are doing with it, or even try it out, visit the Data Explorer home page.  If you did not already know, Data Explorer is no longer a commercial software product from IBM, but is an open source software package.  You can learn more by visiting the following sites:

http://www.opendx.org and http://www.research.ibm.com/dx

http://www.research.ibm.com/dx/proceedings/dx_paper/index.htm

Finally, one of my long-standing areas has been in scientific data management and data models, structures and formats for the representation, access and utilization of scientific data.  I have provided some background material, so that you can learn more about this often neglected but very important topic for the effective and appropriate use of data.  This background material focuses primarily on data models.  I have recently extended some of these ideas to define a new, higher level data model.  A paper that discusses this approach is available for you to read.  There are a few applications of these ideas that may be of interest.  I have already referenced two of them, rule-based advisory tools and visualization systems.  Another application is in the utilization, particularly remotely, of archives of scientific data that have been collected or computed.  If you are interested in this topic, you may wish to read my paper on Interactive Archives of Scientific Data.
 

If you are interested in seeing some other publications, I have prepared a list of those papers and other material.


lloydt@us.ibm.com            20 December 2005