Visualization Test Results of CSM Ocean Data for the CCLI Project

3/27/2000
Tim Scheitlin, NCAR Scientific Computing Division

Abstract

The CSM data was resampled to a regular XY grid,  converted into a Vis5D format and visualized with the VisAD SpreadSheet and the Vis5D application to determine a suitable data size for interactive exploration and whether or not a sub sampled data set contains  El Nino features that can be readily visualized.

Preliminary results show that a relatively small subset of the global CSM Ocean Model data can be visualized to show many prominent and recognizable features of the El Nino phenomenon.   Also, the tools we tested show promise as appropriate platforms for developing an interactive 3D visualization teaching tool.

The results presented on this page describe the data and show which El Nino features are prominent in the model data.  Also, the visualization tools are discussed along with the test platform and associated hardware requirements.
 

Data Description

The data were provided by Dr. Peter Gent (NCAR Climate and Global Dynamics Division) and  were generated from the ocean component of the Climate System Model (CSM)  run on a C90 computing system at NCAR.    The original data set is too large and voluminous to reasonably visualize on a PC class system, so the data were reduced to produce a domain that focused on the El Nino region during two peak historical El Nino events, 1982-1983 and 1996-1997. These time ranges (particularly the 96-97 event) are broad enough so that both normal and El Nino conditions are visible in the data. Also, the original data set only ran through the end of 1997 when the El Nino was at its peak. More current data may now be available, and the full 1997-1998 event cycle could be visualized.

The following table summarizes the characteristics of the original data as received from Dr. Gent and the characteristics of the reduced data set that was used to conduct the visualization tests outlined below:
 

	Original Data from Dr. Gent	Resampled Visualization Data
Real World Coordinates	Global 0-5600 meters below sea level	El Nino Region 30 N. Lat to 30 S. Lat 60 W. Lon to 240 W. Lon 0-232 meters below sea level
Resolution (lon x lat x depth)	152 x 173 x 45	90 x 30 x 15
Data Size	19 GB	1 MB/year (two 3D variables)
Time Resolution	Monthly Averages	Monthly Averages
Time Period	1958-1997	1982-1983, 1996-1997
Data Format	NetCDF	Vis5D
Variables	Approximately 50 ocean variables	Potential Temperature, Temperature Anomalies*

*Temperature anomalies were derived by subtracting the average potential temperature computed between 1958-1997, from potential temperature data values.

The original data sets were resampled and  tested to determine if salient features commonly attributed to the El Nino phenomenon could be reproduced and  visualized. The following images represent the visual information extracted from the reduced data sets described above.   These images were  generated with Vis5D - an interactive tool that may or may not be an appropriate platform for this CCLI project.  Further testing and discussion is needed regarding this issue.

The information used to determine which El Nino features and characteristics  to visualize was taken from a paper, Children of the Tropics: El Nino and La Nina, by Kevin Trenberth (NCAR Climate and Global Dynamics), and Bob Henson (UCAR Communications).  This article highlights many of the well documented features of El Nino and other ocean and atmospheric circulation characteristics.  Some of these features are visualized in the images below.
 

1982-1983	1996-1997	Description	ENSO Features
Movie	Movie	Horizontal Slice of Sea Surface Temperature Anomalies	The most prominent and recognizable feature of El Nino is the temperature anomaly (shown in red) off the coast of Peru.  This feature is clearly evident in the data.
Movie	Movie	Temperature Anomaly isosurface.  Value = 2 degrees C.	The temperature anomaly shows up again as a 2 degree C 3D isosurface.
Movie	Movie	Horizontal Slice of Sea Surface Temperatures	Normally persistent Easterly trade winds cause upwelling along the equator in the East making the water there cold. In the West Pacific, surface water is much warmer. Under El Nino conditions, trade winds weaken and so does the upwelling causing the water in the East to become warmer.
Movie	Movie	Temperature Anomaly isosurface. View from SouthEast,  Value = 2 degrees C.	More views of the temperature anomalies.
Movie	Movie	Temperature Anomaly isosurface View from South, Value = 2 degrees C.	More views of the temperature anomalies.
Movie	Movie	Vertical Slice of Temperature anomalies. -5 degrees Lat.	These vertical slices stretch across the entire Pacific and, when animated, show the evolution of El Nino and the relative strength of the event.
Movie	Movie	Vertical Slice of Temperature.  -5 degrees Latitude.	Under normal conditions warm surface water collects in the W. Pacific, and the thermocline deepens to 100-200 m. In the E Pacific, the thermocline is around 40 m deep because of upwelling.  Under El Nino conditions,  upwelling weakens, and the thermocline gets deeper.  These features are evident in the animations, but not very dramatic.  If you blink, you might miss them.
Movie	Movie	Vertical Slice of Temperature Anomalies. 85 degrees W. Lon.	These slices show  temperature anomalies as seen looking West off the coast of Peru (North-South slice plane).
Movie	Movie	Vertical Slice of Temperature.  85 degrees W. Lon.	Ocean temperatures off the coast of Peru in these animations clearly show the deepening thermocline that is a result of reduced upwelling there. Warm surface waters interfere with this upwelling and prevent  the nutrient rich, cold, deep water from reaching the surface.
Movie	Movie	Isosurface of Temperature.  View from East.  Value = 20 degrees C.	These animations show the deepening thermocline viewed from the east as a 20 degree C isosurface.
Movie	Movie	Isosurface of Temperature Anomalies.  View from East.  Value = 2 degrees C.	These animations show a view from the east and depict the evolution of the El Nino temperature anomalies.

Tools

VisAD SpreadSheet

1. Speed:  It took approximately 30 seconds to calculate isosurfaces for 24 time steps, and then they animated at sufficient, but slow frame rates.  These are, arguably, acceptable performance measurements, but they pale in comparison to native applications that run at least an order of magnitude faster.
2. Interface: Admittedly a very subjective measurement, the interface was found to be clumsy to use and limited  in functionality.  For example, when moving from one context to the next, parameter settings would return to their default instead of retaining their most recent values.  Also, color map editing was non intuitive and awkward.  In general, I found the application difficult to use and not well documented.
3. Functionality: There is no support for displaying slice planes in the visualization.
4. Functionality: There is no method for scaling the dimensions of the data.  Every visualization is displayed as a cube.
5. Functionality: There is no obvious way to add a map to georeference the data.
6. Functionality: There is no easy and automated way to save images.

Vis5D

The  main limitations include:

1. Platform independence:  It is not a platform independent application.  It is written in C and therefore must be compiled for each architecture on which it runs.
2. Licensing Issues:  The PC version of Vis5D  is not currently available for third party release.

Unidata Gridded Data Visualizer (GDV)

Some of GDV's limitations include:

1. File Support: Currently GDV can only read NetCDF files using NUWG, CSM or vmd conventions.  GDV could not read the test data sets.
2. Limited Functionality:  There is no support for 3D isosurfaces or vertical slice planes.
3. Animation: There is no support of automated animation, although the tool does load multiple time steps.  The user must manually move through them.

Hardware Platform