search

NSF-Polar-Cyberinfrastructure / datavis-hackathon

http://nsf-polar-cyberinfrastructure.github.io/datavis-hackathon
42 stars 11 forks source link

Exploring Data with Time as a 4th Dimension #78

Closed allenpope closed 9 years ago

allenpope commented 9 years ago

Imagine this not-so-hypothetical scenario: It's exciting - I'm a researcher and have found some ocean model output to help understand changes to the ice sheet I've seen. The problem is I'm used to working with elevation change and glacier albedo, so this data cube is a new type of data for me and I'm not sure how to interact with it easily. What is a research to do?

This session will come up with a solution to this problem - how to explore and interact with a netcdf "cube" or data where each variable (zonal velocity, meridional velocity, potential temperature) has 4 dimensions (x, y, z, and time). I'm thinking something which lets the user select which variables/dimensions to show, sliders to go through the data, editable color bars, etc.

Soren Scott at NSIDC suggested potentially monkeying around with WebWorkers? Or other web-examples like: • http://tools.pacificclimate.org/dataportal/downscaled_gcms/map/ • http://www.globalcarbonatlas.org/?q=flux_maps • http://earth.nullschool.net/#current/wind/surface/level/orthographic=-90.00,0.00,330 Toni Rosati at NSIDC suggested that Vapor (https://www.vapor.ucar.edu/) might be a good start, too.

I also have and example dataset ready to go (it's MIT-GCM output). ~2.7GB in .nc - once I know where to put it / how to do so...

In addition, when more developed, something like Toolmatch could help solve this sort of issue in the future: http://toolmatch.esipfed.org/#

allenpope commented 9 years ago

Here's some ocean model output in the Bellingshausen Sea in West Antarctica that could be used as an example dataset. Temperature, Meridional, and Zonal Velocity at 0.25 degree resolution, month since 1979. Available at: ftp://sidads.colorado.edu/pub/incoming/apope/ (2.6 GB)

allenpope commented 9 years ago

Thanks to @flamingbear for pointing out Panoply (http://www.giss.nasa.gov/tools/panoply/) - it looks like this might solve most of the issues I laid out, but I'm happy to discuss more if people have thoughts on the use cases they encounter...

chrismattmann commented 9 years ago

always good to have "easy wins" and good examples of products at the workshop - so even if it solves it (seemingly) let's try it out and see if we can build a cool product of the hack.

sgpearse commented 9 years ago

Vapor instructions:

Video Tutorial: http://www.vis.ucar.edu/~pearse/VaporWebinar_3_26_2014.mp4 Installation (requires registration): https://www.vapor.ucar.edu/page/vapor-download Sample Data: https://www.vapor.ucar.edu/page/vapor-download#Example

allenpope commented 9 years ago

Summary from this session:

-A lot of people looking for answers on how to visualize data, but time is a hard thing to visualize, and the nature of static representations seem archaic. -Rasters vs multivariate data, each has unique challenges -Contours sometimes helpful instead of colormaps, easies to see change sometimes, especially across particular thresholds -John Morton (LDEO) suggested cesium.js -Scott (NCAR) showed Vapor can see flow as well as volumes -We went through datasets in Panoply, too (see Session #77 ),

Keep an eye out and share good examples into the future Started working on AWS Session #3 next....