Re: [ferret_users] Re: fill and modulo problem

[Ansley Manke <ansley.b.manke@xxxxxxxx>]

Hi Jaison,
Ferret's FILL/MODULO is intended to check whether the longitudes as 
given in the file will cover the requested longitude region, and if not, 
then overlay the data at plus or minus 360 degrees until the region is 
covered. Grids whose longitude values span a range larger than 360 
degrees are not unusual, so we want Ferret to be able to automatically 
take care of such a case.  Clearly the algorithm for detecting whether 
it's filled,  is having trouble with this particular curvilinear grid, 
and in a way that seems dependent on the data values -- Patrick's 
example data is on the same longitude and latitude curvilinear grid, but 
has slightly different data values. 

So this is a bug in Ferret. Thanks for sending the script; it does help 
make things a bit clearer.

Ansley

jaison@xxxxxxxxxxxxxx wrote:
> Hi Patrick and Ansley,
>                         It seems to be a problem with non-conventional
> longitude values in all of those NetCDF files. Typically, we have longitudes
> varying either 0E to 360E or 180W to 180E (in this case x-values will be
> -180 to 180). In your dataset, longitudes vary from -280 to 360 and Ferret
> gets confused with such a longitude range. Once you apply a correction to this,
> both shade/fill with levels=15 or 10 will work just fine. It is quite interesting
> that, even with confusing longitude values, FILL works comfortably for certain
> values for /levels & SHADE works always!!!.....i have no clue.
>
> Please have a trial with the attached demo file (together with your NetCDF
> files). Please let me know if you have any questions.
>
> Thanks and Regards,
>
> Jaison
>
>
>   
> > Hi Patrick,
> > Thank you for the reports. We'll investigate this and let you know when
> > we have a solution!
> >
> > Ansley
> >
> > Brockmann Patrick wrote:
> >     
> > > Hi all,
> > >
> > > I have encountered a problem with fill command when I use the /modulo
> > > qualifier.
> > > Only a part of the plot is drawn.
> > >
> > > The stranger is that when I use /lev=10 rather than /lev=15 it works.
> > > It works also if I use shade command rather than fill command.
> > >
> > > The bug is available from:
> > > http://dods.ipsl.jussieu.fr/brocksce/ferret_bugs/err608_fillmodulo.jnl
> > > using the file:
> > > http://dods.ipsl.jussieu.fr/brocksce/ferret_bugs/err608_fillmodulo.nc
> > >
> > > I would appreciate some help on this behaviour.
> > >
> > > Patrick
> > >
> > >