Re: NGSETI error in vector plots

Hi Steve (and Rick),

The problem is simply that the value passed to the NGSETI routine was not being initialized properly if
it turned out that no vectors were eligible (for whatever reason) to be plotted. In Dave Allured's case
it was because the minimum magnitude was set greater than all the vectors. In your case it is
because no vector pass through your where function without getting set to a fill value.

I have fixed it and will check it in very soon. However, I am still thinking about Dave Allured's suggestions
for better reporting of this condition. Sorry that it was overlooked for so long. Note however, that this has
not ever been considered a true error condition. It is a zero field, although I can see that could be argued
when it results from the user eliminating otherwise eligible vectors.
 -dave

On Mar 22, 2011, at 2:48 PM, brownrig@ucar.edu wrote:

> Hi Steve,
>
> I've tracked down where in the code this emanates, but that of course
> yields no insight as to why. How much trouble is it for you to
> provide a script and data that exhibits the problem?
>
> Rick
>
> On Tue, 22 Mar 2011 14:02:15 -0600 (MDT)
> Stephen Yeager <yeager@ucar.edu> wrote:
>>
>> Hello:
>>
>> This vector plot issue was brought up in a 2008 post by Dave
>> Allured, and
>> I'm not sure what the status is of any fix...
>>
>>
>> I get the following errors when attempting a vector plot:
>>
>> ERROR 2 IN NGSETI - Parameter value out of range for CT
>> fatal:VectorPostPlotDraw: SET - UNCLEARED PRIOR ERROR
>> fatal:VectorPostPlotDraw: error drawing vectors
>> fatal:VectorPlotDraw: draw error
>> warning:WorkstationDeactivate: workstation not active or not opened
>> fatal:VectorPlotDraw: draw error
>> fatal:PlotManagerPostDraw: error in plot post-draw
>> fatal:_NhlPlotManagerDraw: PostDraw error
>>
>>
>>
>> This is generated by the following code sequence
>> tmprcu = where(prob(i,:,:).lt.0.9,rcu@_FillValue,rcu(i,:,:))
>> tmprcv = where(prob(i,:,:).lt.0.9,rcv@_FillValue,rcv(i,:,:))
>> copy_VarMeta(rcu,tmprcu)
>> copy_VarMeta(rcv,tmprcv)
>> plotvec = gsn_csm_vector(wks,tmprcu,tmprcv,resv)
>> overlay(plot(i),plotvec)
>>
>> The vector plot is failing because I'm masking out the vector
>> components
>> whereever probability is below 0.9. Apparently, this results in too
>> many _FillValues, which gsn_csm_vector can't handle. The sequence
>> works in other situations where probabilities are high and there is
>> relatively little masking. Similarly, I have no problems if I skip
>> the masking step.
>>
>> gsn_csm_vector() should be able to handle an array full of
>> _FillValues,
>> shouldn't it?
>>
>> Thanks,
>> Steve
>>
>>
>>
>>
>> ++++++++++++++++
>> Stephen Yeager
>> yeager@ucar.edu
>> (303) 497-1721
>> ++++++++++++++++
>> _______________________________________________
>> ncl-talk mailing list
>> List instructions, subscriber options, unsubscribe:
>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Tue Mar 22 15:10:46 2011