Fwd: IDV visualizing NYHOPS III files
Richard Signell
I wanted to pass along this message from Nickitas Georgas at Stevens
about his use of the Unidata IDV, which is a free java viewer for
CF-compliant data.
http://www.unidata.ucar.edu/software/idv/
It has a bit of a steep learning curve, but there are lots of
tutorials and support.
-Rich
---------- Forwarded message ----------
From: Nickitas Georgas <Nickitas...@stevens.edu>
Date: Tue, Jun 9, 2009 at 3:15 PM
Subject: RE: Model Data and DMAC tasks
To: Avijit Gangopadhyay <agango...@umassd.edu>, Rich Signell
<rsig...@usgs.gov>
Cc: John Wilkin <wil...@marine.rutgers.edu>, asch...@umassd.edu,
Nickitas Georgas <Nickitas...@stevens.edu>, Josh Kohut
<ko...@marine.rutgers.edu>, Scott Glenn <gl...@marine.rutgers.edu>,
Daniel Holloway <d.hol...@gso.uri.edu>, David Ullman
<d.ul...@gso.uri.edu>, "Howlett, Eoin" <ehow...@asascience.com>,
Dave Runnels <David....@stevens.edu>, Alan Blumberg
<Alan.B...@stevens.edu>, os...@marine.rutgers.edu,
ciossem...@ooici.ucsd.edu, suppo...@unidata.ucar.edu, liang
kuang <kuangl...@gmail.com>
MARCOOS and OSSE colleagues,
IDV is really powerful in visualizing CF1.4-compliant NetCDF files.
The link below is a 3D IDV movie of the passage of Tropical Storm
Ernesto (~2Sep2006) as simulated by NYHOPS Version III.
What you will see:
As the depression comes through from the SW over the southern NY Bight
it pushes high pressure to the NE. The barometric gradient creates
very strong winds. These, in turn, generate waves up to 6 meters high.
A cyclonic surface current is formed, and both initial near-shore
salinity and offshore temperature stratifications collapse over the
DE/NJ shelf in less than two days.
Movie link: http://hudson.dl.stevens-tech.edu/maritimeforecast/google/TS_Ernesto_NYHOPS_Vs_III_Nickitas.zip
(you will need to unzip the avi movie).
Details:
Upper level shows 10-m wind barbs and shaded barometric pressure.
Surface level shows wave contours (meters) and surface current vectors.
Transects show salinity (nearshore) and temperature (offshore).
Only the NY Bight NYHOPS subdomain is shown, and the NYHOPS currents
and wind barbs are subsampled by a factor of 1-3.
Here is a screen capture of the first frame and the legend:
Tech notes of caution (lessons learned…):
IDV does not seem to want to capture legends in movie mode without
obstructing the view. Or it does, and I just don’t know how to.
Currently, IDV has limits on displaying “unstructured” grids: The full
NYHOPS Version III curvilinear grid has a lot of land cells that
create a non-contiguous array with cavities. In order to make NYHOPS
fully render in 3D (2D is ok) we had to limit our much larger NYHOPS
domain (that includes the Sound, the Harbor, the Hudson, etc.) to a
water-only NY Bight one. IDV folks know about this, and are
experimenting on how to display grids with cavities. It is a very
tricky issue and may be “insolvable” in IDV. Maybe griddata can solve
this on the OpenDAP/THREDDS level?
We have been unable to make IDV 2.7 beta work with our data – it only
registers 3D vars and then does not read them. IDV folks are working
on the release 2.7 version which should have all that fixed. IDV 2.6
is stable (I used 2.6u2), but the 3D choices are reversed: you have to
select “all levels” to read and display the surface level only, and
you have to select the surface sigma level to read the full 3D grid.
Also, in order to make 2 days of 10-min averages render (this is the
new NYHOPS temporal resolution), even for the much smaller subdomain
displayed, we had to increase the Java VM memory that IDV sees to its
highest allowed for 32bit Windows (1.6MB, set by editing the
runIDV.bat file). Hopefully, 64bit will work better, esp. with Windows
7 (no commission from MS yet…)
Finally, the full NYHOPS version III native output is CF-1.4
compliant, and this is immensely helpful: Grid-oriented currents can
be rotated to E-W/N-S through built-in grid formulas, all variables
are stored at cell centers, vertically stretched sigma levels register
at instantaneous cell mid-points, as they should, times and units are
understood automatically and allow for conversions. After reading the
water-only file with IDV, one can easily do volume rendering, drop
virtual probes and get vertical profiles, create dynamic cross
sections, calculate advection of scalars, vorticity and divergence,
and a whole multitude of things that I have not been able to do yet.
Acknowledgment: Thanks to Liang Kuang, who wrote a very nice automated
Matlab2008b subroutine to subsample our full NYHOPS domain variables
to the water-only grid displayed, assuming CF compliance of the input
file.
-Nickitas
Nickitas Georgas
Senior Research Engineer
NYHOPS: New York Harbor Observation and Prediction System
Center for Maritime Systems
Davidson Laboratory at Stevens Institute of Technology
711 Hudson Street
Hoboken, NJ 07030
www.stevens.edu/maritimeforecast
To resist progress is to perish.
John Kenneth Galbraith
--
Dr. Richard P. Signell (508) 457-2229
USGS, 384 Woods Hole Rd.
Woods Hole, MA 02543-1598