[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[IDV #PRY-290340]: averageOverTime method

Hi Bill-

> Full Name: bill fingerhut
> Email Address: address@hidden
> Organization: lyndon state college
> Package Version: 2.4 build date:2007-12-20 16:34 UTC
> Operating System: Windows XP
> Hardware: Java: home: C:\Documents and Settings\waf01060\Program 
> Files\IDV_2.4\jre version: 1.5.0_11 j3d:1.3.2 fcs (build12)
> Description of problem: Don,
> 
> To work around this missing data issue, I tried selecting times that
> appear to be okay. The dataset selected is on Motherlode, GFS Northern
> Hemisphere 381 km, latest. (The grids are in grib format.) Times were
> selected from the dataset properties, Times: 00 and 12 Z, every day
> from 0 to 84 hr forecasts. The field was selected: 3 D, geopotential
> height @ isobaric. The display was contour plan view at 250 mb;
> projection was North Pole. Following is what was displayed:
> time index    requested time  displayed time  comment
> 1     3/10 12 Z       3/10 12 Z       ok
> 2     3/11 00 Z       3/11 12 Z       ok, wrong time
> 3     3/11 12 Z       3/12  6 Z       no data, wrong time
> 4     3/12 00 Z       3/13 00 Z       ok, wrong time
> 5     3/12 12 Z       3/14 00 Z       ok, wrong time
> 6     3/13 00 Z       3/14 12 Z       no data, wrong time
> 7     3/13 12 Z       3/15 12 Z       ok, wrong time
> The 3rd and 6th maps have no contours; 2nd, 4th, 5th, and 7th maps
> have data from different times; 5th, 6th and 7th maps have data not
> requested (beyond 84 hour forecast).

We'll look into this.  I suspect the problem is that the indexing
is off because the data source shows all times in the file.  For
a particular parameter, the list of times might be different.  If
you select the times in the data subset panel instead of for
the data source, do you get the correct times?

> What does this mean? I don't see how I can do what I want from this dataset.


> Next, I repeated the experiment with _netcdf_ grids from Motherlode.
> Everything worked as it should, including the averageOverTime method !

The netCDF files have only one time dimension and all grids are
dimensioned by that whether they have data or not.

> So, I am now going to stay away from the grib grids. What is going on?

Why are you using the 381 km grids instead of the global grids?  

Don

> > -----Original Message-----
> > From: Unidata IDV Support [mailto:address@hidden]
> > Sent: Thursday, March 06, 2008 5:55 PM
> > To: Fingerhut, William A @ LSC
> > Cc: address@hidden
> > Subject: [IDV #PRY-290340]: averageOverTime method
> >
> > Hi Bill-
> >
> > > Full Name: bill fingerhut
> > > Email Address: address@hidden
> > > Organization: lyndon state college
> > > Package Version: 2.4 build date:2007-12-20 16:34 UTC
> > > Operating System: Windows XP
> > > Hardware: Java: home: C:\Documents and Settings\waf01060\Program
> > Files\IDV_2.4\jre version: 1.5.0_11 j3d:1.3.2 fcs (build12)
> > > Description of problem: I am having trouble use the system library
> > method, averageOverTime, with GFS data
> > > from the northern hemisphere. No contours are drawn and the range is
> > from infinity to
> > > negative infinity.
> > >
> > > I copied the method to my library and did some experimentation. The
> > following snipet:
> > >
> > > for t in range(domainSet.getLength()):
> > > cnt=cnt+1
> > > rangeValue = field.getSample(t)
> > > print "   ",cnt,t,rangeValue[1]
> > >
> > > confirms that a few time levels are missing. This seems to produce the
> > failure.
> > >
> > > The following test on rangeValue[1] would work if I knew what to test
> > for - None is not
> > > correct.
> > >
> > > if(cnt == 1):
> > > current = rangeValue;
> > > if(rangeValue[1] == None):
> > > print "      is None";
> > > else:
> > > print "      is not None"
> > > current = current + rangeValue;
> > >
> > > What do you think ?
> >
> > Are the timesteps completely missing or is just part of the grid missing?
> >
> > The downside I see to the checking just the first point is that it could
> > be missing, but there could be valid points in the grid.  The RUC grids
> > are like this where there are NaN's along the edge.  There is a method
> > GridUtil.isAllMissing(grid) which checks for all missing values.  You
> > could call that instead of checking rangeValue[1].
> >
> > However, what would you want the value to be at a point if in
> > one timestep it was missing, but all the other values were okay,
> > e.g.:
> >
> > 30,20,31,28,NaN,25,27
> >
> > Adding those up and dividing by 7 would give you a NaN (missing).
> >
> > Don
> >
> > Ticket Details
> > ===================
> > Ticket ID: PRY-290340
> > Department: Support IDV
> > Priority: Normal
> > Status: Open
> 
> 
> 
> 


Ticket Details
===================
Ticket ID: PRY-290340
Department: Support IDV
Priority: Normal
Status: Open