[IDV #COT-972342]: Loading 3D Fields in IDV

Hi Josh-

> I suspect the post processing issue is the solution.  Whenever I converted the
> data over to Vis5D format using ARWpost, the 3D fields loaded pretty quickly.
> And you are right on with the derived parameters taking the longest amount of
> time.  So that answers the original question.

Okay.  That's what I figured.

> The only issue with converting using ARWpost is IDV won't create a topography
> from the ARW height variable once it's converted from NetCDF to Vis5D.  It
> comes up with an error that says something about the units being unknown.  I
> think this is an issue with Vis5D.  Do you have any suggestions on that by
> chance?

When you say height, are you talking about geopotential height or
surface height.

Can you provide the Vis5D file on an FTP or HTTP server for me
to take a look at?

> Thanks very much!

Have a good weekend.  

Don

> On Fri, Apr 11, 2008 06:01 PM, "Unidata IDV Support"
> <address@hidden> wrote:
> >
> Hi Josh-
> >
> >> Full Name: Josh Gelman
> >> Email Address: address@hidden
> >> Organization: PSU
> >> Package Version:
> >> Operating System:
> >> Hardware:
> >> Description of problem: Hello,
> >>
> >> Whenever I try to load a 3D field in IDV using a WRF ARW dataset, the time
> >it takes to load is extremely slow (up to a half hour sometimes).
> >For 2D fields it's nearly instantaneous.  This only seems to happen with my 
> >WRF
> >ARW datasets.  The rest load pretty fast.  I am curious if there is any way 
> >to
> >resolve this issue?
> >
> >When you say "only seems to happend with my WRF ARW datasets", are
> >you
> >comparing them to standard (e.g. GFS) grids or WRF NMM datasets?  Are
> >you
> >loading single levels or are you loading in the entire volume and trying
> >to do cross sections or isosurfaces?
> >
> >The problem with the WRF netCDF files (NMM and ARW) is that they
> >grids are staggered.  To handle this, we need to hold both the
> >native grid domain and a transform to pressure in memory.  So
> >at the outset, just loading in the grid is a double wammy.  Your
> >performance is going to depend on the size of the grid and number
> >of timesteps as well.  What are the dimensions of your grid?
> >Any of the derived parameters have an added overhead of resampling
> >the data from one point of the domain to another.  For example,
> >wind speed or wind vectors require resampling the V to the U domain.
> >Those are grids that I've seen really bad performance on and the
> >culprit is the resampling.  Straight parameters like U wind or
> >vapor pressure shouldn't take much longer than a standard unstaggered
> >high resolution NAM grid.
> >
> >In general, we suggest that users postprocess the WRF netCDF data
> >using WRF-POST or the Vis5D converter.  These destagger the grids
> >and output them on pressure levels as well as compute standard
> >parameters like  sea level pressure.
> >
> >There are also some steps you can take to improve performance
> >listed at:
> >
> >http://www.unidata.ucar.edu/software/idv/docs/userguide/misc/PerformanceTuning.html
> >
> >> The computers I'm running IDV on have 16 GB of memory and 3 GHz processor 
> >> speed.
> >
> >When the data is loading slowly, is the memory monitor showing that
> >you are maxing out on memory?  Have you increased the default
> >amount of memory that the IDV gets allocated?  If not, see the
> >performance tips above.  Depending on your operating system, you'll
> >only be able to use a subset of the total memory.
> >
> >> Thanks for any suggestions.
> >
> >I'm guessing that the slowdown is due to resampling which is painful
> >for the staggered grids.  But if you are seeing really slow loading
> >on one of the main fields in the grid, then there is some other
> >issue.
> >
> >Don Murray
> >
> >Ticket Details
> >===================
> >Ticket ID: COT-972342
> >Department: Support IDV
> >Priority: Normal
> >Status: Open
> >
> >
> >
> >
> 
> 
> 
> 
> 
> 


Ticket Details
===================
Ticket ID: COT-972342
Department: Support IDV
Priority: Normal
Status: Open