Ben Brown, 1/22/2011
Ben primarily uses VAPOR for ASH code data (spherical grid).
Resampling spherical to rectlinear grid presents a number of problems:
- Explosion of data (7x to 10x). Simulation grid is a spherical shell of convection zone. For sun, inner two thirds are empty
- Undefined data are mapped must be mapped to valid data values which results in a number of visualization artifacts
Ben is current computing at NICS, but considering move to TACC in spring
Interested in generating a time animation of convection simulations. Need 4TBs of space
Wish list:
- Native support for spherical data
- Animation needs:
- Zoom, Pan
- Rotate
- Control timestep advance
- Animate region
- Animate TF
- Efficient rendering with missing values (e.g. exterior of sphere)
- Gaussian interval (over time) to control field-line opacity
- Interested in ray tracing
- Transfer functions need more values, not uniformly spaced.
- Logarithmic TF
- Would like to see curve of TF colors in a color wheel
- Save and restore field line color/opacity map
James Done, 1/26/2011
James does regional climate simulations (NRCM work)
Uses VAPOR primarily for producing visuals for communicating with public. Doesn't use it for analysis
Model for Prediction Across Scales (MPAS): new weather model under development by Joe Klemp
- Targets weather and regional climate
- Adaptive, non-rectilinear grid used for computation. James speculates interpolation to regular grids will be supported for output
- long term plan is to switch from WRF to MPAS. Who?
Wish list:
- Lossless image output format (e.g. PNG)
- Easy way to produce AVIs
- Import 2D geo-referenced data from various sources
- Bring in data from other models (on different grids). This could be done more easily if we supported import from NCL, since James already has NCL code that would reinterpolate data.
- Animation control: Primarily zoom, pan, rotate (around a point of interest)
- Smaller things:
- Output uncompressed tiff animation frames
- Provide color scale for all visualizers
Pablo Mininni & Duane Rosenburg, 1/27/2011
Wish list:
- Include 3D geometry in scene. Would benefit many communities, e.g. molecules built from cones and balls could be placed inside density fields.
- Key framing. Control camera path (+ zoom, pan, rotate), time step, transfer function
- Eventually would like to animate from scripting language
Currently working on simulations with 3072^3 grid points, soon moving to 4096^3. Interested in direct output to VDC2
Leigh Orf, 1/28/2011
Leigh uses both VisIt and VAPOR. Uses VisIt for parallel capabilities.
Running George Bryan's code (CM1). For cloud simulations only 15 - 20% of the data is of interest... ROI selection is valuable
CM1 uses a stretched grid
Wish list:
- support for stretched grids
- key framing - change viewpoing and time step
- Turn on and off spin animation
- Make tooltips in the Transfer Function editor less sluggish.
Nic Nelson & Kyle Auguston, 2/3/2011
Also running ASH code.
Same basic problems and wish list as Ben Brown
Dynamo simulations. Looking at how sun spots form (1024x512x193), chevy-chev grid
Using ASH code and another code (Kyle's home grown) that uses spherical grid -
Nick computes at TACC on Ranger. Kyle elsewhere
Nick runs IDL routines on Spur at TACC
found structures rising due to boyancy using vapor - have to zoom in on portions of data
problem: too much data - can't save time steps frequently enough
Nick interested in getting Longhorn account.
Interested in extensibility - may write their own flow integration in spherical coordinates
desired features
+ more support (e.g. probe) for spherical coordinates
+ advect the rake using the average velocity within the box (may be other workarounds)
+ mask for masking off interior and exterior values resulting from resampling sphere to box, and other support for missing values
+ support for spherical data
+ key framing (camera path, color map, region of interest)
Sasha Brun, who was visiting that day, mentioned the need to select a particular field line and color it uniquely. Also would like Auto-stereo display.
Janice Cohn, 1/26/2011
Desired features:
- Wind barbs on model layers, especially on bottom layer
- Would be OK if there's an option of positioning the barbs based on distance above the ground
- Needs to use 2d variables as well as 3d variables for vector components
Sherrie Fredrick, 2/4/2011
Sherrie put together a collection of requirements from the WRF community.
They frequently need the following derived variables (currently available in NCL and/or RIP):
- Sea level pressure
- Dewpoint
- Relative humidity
- Equivalent potential temperature
- Reflectivity
- Maximum reflectivity in vertical column
- Wind speed
- Potential vorticity
- Cloud top temperature
- Wind Shear
- Cape
- Cin
The following color maps (from the NCL Colortables website) are most useful:
- testcmap
- wh-bl-rg-ye-red
- BlRe
- First 4-5 maps in NCL colortable page
- Small maps: all with names starting with “precip”. Probably as discrete, not smooth mapping.
Would like to position wind barbs on pressure layers (P+PB isosurfaces)
Sherrie thought it would be quite valuable to be able to import 3d radar data into vapor scene. Mentioned that Jay Miller (ljmill@ucar.edu) developed Sprint and Cedric tools. Jay may be able to tell us how to use Sprint to get radar data into WRF grid). Sprint can reinterpolate (Nexrad et al ) radar data to a cartesian grid. Cedric is a vis tool for radar data.
Zbigniew Piotrotrosky and his associate Andrzej (Andy) Wyszogrodzki March 1, 2011
Andy uses Vis5D and would like vapor to have some of its capabilities:
- Arbitrary planes of wind arrows. Interactively rotate, position the plane.
- Animation scripting (the only parameters he wanted to animate were the viewpoint, orientation and position of wind-barb plane and probe). He would like to be able to write scripts in Python to animate parameters in vapor.
- Would like to graph a variable's values along a line (especially a vertical line) Handle multiple variables with multiple colors in the graph.
Would like to know how to have vapor read their data as if it were a WRF dataset:
- Short term, by renaming dimensions and variables. Should also tell users what variables and dimensions are necessary in WRF-like data.
- Longer term, by having VAPOR recognize CF-compliant conventions.
Other things:
- Add geometric objects into scene.
- Be able to make wind 0 where there are missing values in data
- Export data values (not just colors) from probe. Nearest value is OK. Smoothing or trilinear interpolation may also be OK.
- Would like to run without a graphics card (e.g. using mesa) on frost, bluefire, etc. Possibly not render in the visualizer, just in the tabs.
- Would like the GUI to hide complex non-default settings unless user requests them.
- Need to be able to handle missing values