Search Results for El Niño Data

Tim Scheitlin, NCAR Scientific Computing Division
5/15/2000

A couple new data resources have been located since the last set of visualizations were published. One contains sea surface height data and another, atmospheric variables that lend themselves to flow and vector visualizations for demonstrating global temperature and wind patterns.

Issues and concerns related to the data are discussed here, and examples of visualizations of the data are presented.

Data Issues

Choosing Data and Variables

Several distinct sets of data will be required to demonstrate the varied climatalogical features that we wish to present to students. In some cases the features are most clear when viewing monthly or even tri-monthly averages. In other cases the data are better represented when visualizing hourly time scales. Also, features like El Nino warming, show up best when viewing anomaly fields rather than the data in its original form. Atmospheric wind trajectories, on the other hand, require data sets with a time resolution on the order of hours, and the wind vector patterns show up most clearly when viewing tri-monthly averages. Also in visualizations where we wish to compare data from the atmosphere and the ocean, it will be necessary to account for the large differences in range of motion over time, possibly through time averaging or interpolation (Atmospheric data varies much faster than ocean data).

Having multiple data sets with different time and perhaps spatial resolutions will require that we provide a mechanism for managing the data for the user and helping them decide which data and variables to visualize based on which feature is being discussed. This may be as simple as providing a single data set with each visualization exercise, or it may be more advanced by providing a method for allowing the user to query different data sets and choosing, on their own, which variables to visualize.

Choosing Visualization Parameters

It is sometimes difficult to identify the El Nino or Normal Year climatalogical features unless the colormaps are carefully adjusted or isosurface values are selected appropriately. Even then, the features, are often transient, only appearing in a few frames. I found that the vector and trajectory visualizations, in particular, were often the most difficult to correlate with the El Nino features. For example, the ocean and wind vectors, at times, seemed to flow in opposite directions from one another in a narrow band along the equator. Is this expected? In general, I have found that many of the characteristics that we attribute to the El Nino Phenomena, or other climatalogical issues, are small relative to the rest of Earth System, or even a reduced-size domain, and require a careful eye to find them.

We may need to heavily prescribe how the visualization parameters are chosen or present some other way of providing cues and hints that will allow the user to make choices that will lend themselves to meaningful and clear visualizations. This issue will require careful consideration so that the interactive appeal of an exploratory tool is not diminished by over-prescribing the steps needed to reach a discovery.

Performance

Some of the visualizations such as the trajectory animations will be difficult on a PC-Class system. Also, it is questionable whether or not any of the high time resolution data sets will be able to run on a PC-Class system.

The challenge this presents is that we will need to reduce the data resolutions significantly in some cases and at the same time preserve the salient climatalogical features that are being presented.

Note: I was unable to test the high resolution data sets on the PC system - the software would not load the data. I am currently investigating this issue.

Other Issues

Given that climatalogical features are not always readily apparent in some visualizations without proper adjustment of colormaps, isosurface values or other visualization parameters, it may be necessary to heavily prescribe many of the user actions necessary to reveal the pertinent features in the data. My concern is that if the user's actions are, by a large degree, externally directed, the usefulness of an interactive exploratory tool becomes questionable and could become a liability rather than a teaching asset.

So, in cases where data may be too difficult to visualize interactively because the features are not readily apparent or performance issues become an overriding concern, an alternative will be to provide high quality, pre-computed visualizations in an MPEG, Quicktime, or other movie format. The pre-computed animations could be much more visually engaging than the Vis5D-class, interactive graphics, and they would not be limited by domain size or time resolution. Also, there would be no problems with cross-platform compatibility - the animations would run on a Mac, IBM PC, or even a Unix system and could easily be "reused" for other teaching and presentation purposes.

As an example of the difference in graphics quality, a comparison is shown below between Vis5D graphics and higher quality pre-computed animations. The following animations show the same features (Precipitable Water (TOP) and Sea Surface Temperature Anomalies (BOTTOM)) using interactive Vis5D-Class graphics (LEFT) and a high quality, high time resolution, pre-computed animation (RIGHT).

Interactive Graphics	Pre-computed Graphics

Data Sources

Dr. Peter Gent (NCAR CGD): CSM Ocean Data
Gohkan Danabasoglu (NCAR CGD): Sea Surface Height Data
NCAR MSS Archives: NCEP Reanalysis Atmospheric Data

Data Sizes

Ocean Domain (30 x 90 x 15), 1 year (12 timesteps), 1 variable = 2 MB/year
Atmospheric Domain (73 x 144 x 12), 1 year (12 timesteps), 1 variable = 6 MB/year
Atmospheric Domain (73 x 144 x 12), 1 month (124 timesteps), 1 variable = 62 MB/month

Note: The atmospheric data has not yet been sampled down to the same domain size as the ocean data.

Visualization Examples

The following table compares data from a "normal" non-El Nino year (1996), with the last large El Nino event on record (1997). Particular effort was made to visualize the specified "5D Visualizations" from the "Ideal Inquiry Path" published by Ken Hay.

Variables and Phenomenon	Dataset Description	Normal (1996)	El Nino (1997)
Relative Humidity and Precipitable Water. Heavy rain coming from moist air in New Zealand.	Resolution : 144x73x12 1000-100mb Global lat x lon	Relative Humidity (monthly) Relative Humidity (tri-monthly) Jan. Precipitable Water (6-hourly) Precipitable Water (monthly)	Relative Humidity (monthly) Relative Humidity (tri-monthly) Jan. Precipitable Water (6-hourly) Precipitable Water (monthly)
Vertical Wind Component. Lifting in the equatorial Pacific	Resolution : 144x73x12 1000-100mb Global lat x lon	Pressure Vertical Velocity (tri-monthly)	Pressure Vertical Velocity (tri-monthly)
Water Vapor and wind direction. Moist air is not blowing in from other locations to NZ.	Resolution : 144x73x12 1000-100mb Global lat x lon	Relative Humidity Wind vectors at surface (tri-monthly) Relative Humidity Wind vectors at surface. (6-hourly)	Relative Humidity Wind vectors at surface. (tri-monthly) Relative Humidity Wind vectors at surface. (6-hourly)
Ocean temperature. Warm equatorial Pacific. Moist air forms from evaporation in W. Pacific.		Sea Surface Temperature (monthly)	Sea Surface Temperature (monthly)
Solar absorption. Sun's radiation on equatorial Pacific does not cause hot ocean.		No dataset located yet.	No dataset located yet.
SST, current direction, and SSH. Water is pushed in from other sources. Hot ocean is cause by water pushed along equator and pooled up by NZ.		Sea Surface Height (monthly) Sea Surface Temperatures and Ocean Current vectors. (monthly).	Sea Surface Height (monthly) Sea Surface Temperatures and Ocean Current vectors (monthly).
Wind and Water direction and speed. Wind pushes water.		1000mb Horizontal Wind vectors January (6-hourly). 1000mb Horizontal Wind vectors (monthly) Surface ocean current. (monthly).	1000mb Horizontal Wind vectors September (6-hourly). 1000mb Horizontal Wind vectors (monthly) Surface ocean current (monthly).
Current direction and speed. Water forms a cycle.		Vertical ocean slice at equator. Temperature and Current vectors (monthly).	Vertical ocean slice at equator. Temperature and Current vectors (monthly).
Horizontal component of wind. Consistent wind moving from E. to W.		1000mb horizontal wind vectors (tri-monthly)	1000mb horizontal wind vectors (tri-monthly)
Ocean Temperature and Photophakton. Peruvian cool surface water, nutrient rich cold water rising to the surface.		No dataset located yet.	No dataset located yet
Wind direction and speed in upper atmos. Walker cell - wind moving from E to W at surface and W to E in upper atmosphere.	Resolution : 144x73x12 1000-100mb Global lat x lon	January Trajectories at Equator. Color maps Temperature. View from S. (6-hourly) 300 mb horizontal wind vectors. view from top. (tri-monthly)	January Trajectories at Equator. Color maps Temperature. View from S. (6-hourly) 300 mb horizontal wind vectors. View from top. (tri-monthly)