--- Welcome to the official ADCIRCWiki site! The site is currently under construction, with limited information. ---

For general information, see the ADCIRC site, ADCIRC FAQ, or the ADCIRC Wikipedia page. For model documentation not yet available on the wiki, see the ADCIRC site. New content is being continuously added to the ADCIRCWiki, and material from the main ADCIRC site will be gradually transitioned over to the wiki.

Visualization: Difference between revisions

From ADCIRCWiki
Jump to navigation Jump to search
mNo edit summary
(→‎Kalpana: added link)
Line 37: Line 37:


==Kalpana==
==Kalpana==
Kalpana (by Rosemary Cyriac) produces either a shapefile or a Google Earth kml file of ADCIRC NetCDF output or mesh bathymetry. Kalpana is a Python script that generates color contours using the mesh triangulation via the 2D plotting library Matplotlib, then converts the contours to either shapefile via the Fiona/Shapely libraries, or to kmz via the Simplekml library. Both the shapefile option and kml option produce polygons that enclose areas with data values that fall into one contour bin.  
[https://ccht.ccee.ncsu.edu/kalpana/ Kalpana] (by Rosemary Cyriac) produces either a shapefile or a Google Earth kml file of ADCIRC NetCDF output or mesh bathymetry. Kalpana is a Python script that generates color contours using the mesh triangulation via the 2D plotting library Matplotlib, then converts the contours to either shapefile via the Fiona/Shapely libraries, or to kmz via the Simplekml library. Both the shapefile option and kml option produce polygons that enclose areas with data values that fall into one contour bin.  
*Like many Python scripts, Kalpana should be run in a virtual environment. Setting up this virtual environment can be simple or frustrating depending on existing system installations on your HPC. Jason Fleming wrote a great post on his experience in installing the various Python libraries within a virtual environment on the Kalpana website.   
*Like many Python scripts, Kalpana should be run in a virtual environment. Setting up this virtual environment can be simple or frustrating depending on existing system installations on your HPC. Jason Fleming wrote a great post on his experience in installing the various Python libraries within a virtual environment on the Kalpana website.   
*Kalpana runs on an HPC, so you don’t need to download data to your desktop to plot.
*Kalpana runs on an HPC, so you don’t need to download data to your desktop to plot.

Revision as of 18:19, 18 March 2019

In the course of running ADCIRC you may wish to visualize or plot data pertaining to your model. Visualizing data can help you fine tune the mesh design, conduct QA/QC on results, and communicate key facets of the modeling to others. Examples of datasets you might plot include mesh node elevation; nodal attributes such as Manning’s n or wind stress reduction coefficients; or modeled results such as water surface elevation. For visualizing these quantities on a 2D flexible mesh, ADCIRC users often choose to plot them as color contours that follow the mesh triangulation. Options abound for generating such plots, and the best tool for the job depends on project goals and available computational resources. Some factors to consider in choosing a visualization tool include:

  • Interactive vs. static image – do you intend to zoom and pan around your visualized data and/or change contour intervals on the fly, or would you prefer to batch process image files at fixed zoom extents and contour limits?
  • Data transfer – do you intend to plot on your HPC directly, or do you prefer to transfer data first to your desktop to plot?
  • Desired processing speed – does the speed of plotting matter to you? If you’re generating hundreds of images to stitch together in an animation, speed may become important.
  • Effort – do you prefer a DIY, high customization approach or to use a ready-to-run program? Somewhat related question: do you want to buy software or leverage open source libraries?
  • NetCDF vs. ASCII data – while ADCIRC mesh and nodal attributes input files must take ASCII form, you can specify ADCIRC output in either ASCII or NetCDF format. Because some plotting tools can only work with one or the other, the plotting requirements of your project might influence which tool you choose. A few considerations: multiple applications can read a NetCDF4 file simultaneously, which could save considerable time if you can run your plotting tool concurrently on multiple processors within one compute node. Also, NetCDF ADCIRC output files self-contain mesh node, connectivity, and boundary information.
  • Projection – do you need the plotting tool to accurately project geographic coordinates, or can it treat latitude and longitude as though they are cartesian coordinates? ADCIRC runs in geographic coordinate system, which references a spherical object - but plots are flat objects. Some visualization tools cannot project geographic coordinates accurately, which may distort the image if plotting near the poles or over a large domain.
  • Which files do you need to visualize? Do you need to plot nodal attribute (fort.13) data, or just ADCIRC output?
  • Customization – do you require additional information in your plots beyond color contours? Your wish list might include: overlaying a vector field showing wind magnitude and direction or wave direction; adding background aerial imagery; auto-calculating the contour maximum within the specified zoom extents; or overlaying line shapefiles showing county boundaries, inshore waterways, storm track, etc.

Now that you’ve contemplated your visualization requirements, let’s discuss some frequently used options.

SMS (Surface-water Modeling System)

Aquaveo’s SMS software features a suite of many GUI-based pre- and post-processing tools that allow for interactive visualization. SMS requires a license, but out of the box you can easily open and view an ADCIRC mesh and any associated data, including nodal attributes files. SMS plots mesh data as a triangulated irregular network (TIN) surface, allowing the user to change contour intervals as desired. You will need to download files from the HPC you run ADCIRC onto your desktop to plot them with SMS. A few notes about SMS:

  • ADCIRC output NetCDF support is “coming soon” as of version 12.1. SMS supports other types of NetCDF data but cannot open a *.63.nc file.
  • When ADCIRC output files are opened, SMS automatically converts them from ASCII to binary HDF5 format. This format allows data to load quickly – anecdotal reports estimate this speed to be faster than the same data displayed as a TIN in ArcMap.
  • SMS can...
    • project data in geographic coordinates properly.
    • download zoom-specific aerial imagery to display as a background layer.
    • supports vector overlays for fort.74-style ADCIRC data (u v format only, so you need to reformat swan_DIR.63 before displaying as vector field).
    • display shapefiles, but with limited options for changing their symbology.

FigureGen

The FigureGen (by Casey Dietrich of NC State) program batch produces images of model data. FigureGen is a Fortran program that uses the Generic Mapping Tools (GMT) library. To use FigureGen, you must first install GMT and any of its auxiliary programs that don’t already exist on your system. When you compile FigureGen, you can use flags to enable Google Earth output, NetCDF support, and parallel processing. You’ll then edit a text file of input parameters, including zoom extent and required output snaps (for time-varying ADCIRC output file), and FigureGen will generate the series of requested images. A few notes about FigureGen:

  • FigureGen is a triple threat – it’s fast, easy to use, and free (but please cite as appropriate!)
  • You can run FigureGen directly on your HPC so no need to download large ADCIRC output files in order to visualize the results.
  • GMT supports map projections so FigureGen plots geographic coordinates correctly.
  • FigureGen features options for overlaying a vector field (fort.74-style data) including automatically finding the maximum vector magnitude present within the zoom extent.
  • FigureGen can...
    • automatically find the maximum data value present within the zoom extent to set the contour maximum.
    • overlay a GSHHS shoreline. If your mesh includes detailed inshore waterways, this shoreline may not include sufficient resolution.
    • add a time bar.
    • plot the difference between two files.
    • plot nodal attributes data (except for elemental_slope_limiter, but that is easy to add yourself)
    • produce several types of image files, and the user can set resolution to control file size.

Kalpana

Kalpana (by Rosemary Cyriac) produces either a shapefile or a Google Earth kml file of ADCIRC NetCDF output or mesh bathymetry. Kalpana is a Python script that generates color contours using the mesh triangulation via the 2D plotting library Matplotlib, then converts the contours to either shapefile via the Fiona/Shapely libraries, or to kmz via the Simplekml library. Both the shapefile option and kml option produce polygons that enclose areas with data values that fall into one contour bin.

  • Like many Python scripts, Kalpana should be run in a virtual environment. Setting up this virtual environment can be simple or frustrating depending on existing system installations on your HPC. Jason Fleming wrote a great post on his experience in installing the various Python libraries within a virtual environment on the Kalpana website.
  • Kalpana runs on an HPC, so you don’t need to download data to your desktop to plot.
  • There exists a file size and speed trade-off to the two file format options – the kml option is slower to generate, but the file size is relatively small; conversely, the shapefile option generates very quickly (nearly as fast as FigureGen) but the file size is on the order of the size of the file it’s plotting. Also note that the speed of the kml generation depends on the number of lat/long bins you set up in the command line, and the number of bins depends on mesh extent and resolution.
  • The shapefiles and kml files produced allow the user to zoom and pan around, but not to dynamically change contour intervals on the fly – the polygons represent an instance or a snapshot of the data.
  • Both shapefile and kml are popular file formats with the outside world, so they’re good choices for communicating your model results to others.
  • Neither file format supports vector fields. If you want vector fields, you should edit the Kalpana script to skip the Shapely/Fiona or Simplekml parts and use Matplotlib directly to add the vector field and save the Matplotlib plot as a jpg.
  • Kalpana is free (but cite as appropriate!)
  • No support for plotting nodal attributes parameters yet.
  • Shapely/Fiona and Simplekml support map projections and thus properly plot geographic coordinates.

ADCIRC.io

ArcGIS

Esri’s classic software contains no canned tools to read or write ADCIRC data. The user must develop scripts to read ADCIRC files via ArcMap's integrated Python package (ArcPy). Once you’ve imported ADCIRC mesh geometry, nodal attributes, and model results data into a geodatabase, you can easily create a TIN using the mesh triangulation as hard breaklines to view your data in a similar fashion to SMS. Note that the license for ArcMap costs significantly more than an SMS license. So why the heck would you ever consider using ArcStuff for your visualization needs?

  • Esri is everywhere – you might already have access to a network license or to colleagues with advanced GIS knowledge.
  • While ArcMap doesn’t feature canned tools to read ADCIRC data, it does feature canned geoprocessing tools that can aid greatly in assigning nodal attributes data or interpolating mesh node elevations. For example, you can map data showing land use/land cover in raster or polygon format onto your nodes to assign Manning’s n, or you can create custom schemes for interpolating DEM data onto your mesh nodes.
  • You must download ADCIRC output data from your HPC to wherever your ArcGIS license exists.
  • ArcMap allows for a very high level of customization in your image layout. North arrows, scales, legends, infinite symbology and overlays – wow!

Matlab

Paraview

External Links