Re: [las_users] las: LAS does not produce longitude time plots

[Martin Schmidt <martin.schmidt@xxxxxxxxxxxxxxxxx>]

Hi Ansley,
thanks for your answer. I have done so and reinstalled the LAS with the 
latest ferret.

The first finding was, that
go/geo_borders_intermed.nc
go/geo_borders.cdf
are missing but required somewhere by LAS. I have copied them from an 
old version of ferret and LAS starts working.

Nevertheless, this does not resolve the problem with the missing values. 
Vertical sections can be displayed only if there
are no land points included.

The missing values in the geolon_t/geolat_t fields are from MOM. MOM is 
able to mask land-only processors
when running massive parallel. Combining the results into a single file, 
the variables for the missing tiles are
filled with"missing". This is fine for normal variables, but happens 
also for the curvilinear coordinates - you can see the result
with LAS in the variable "tracer latitude".

However, also when I try to display a section that includes some land 
but not an area, where
coordinates are missing (20°S, 0 - 13.8E) , I get the error.

Bailing out of external function "samplexy_curv_nrst":
    Arg 2, curvilinear X coordinates, contains missing values


This is true, but I have excluded the region with missing values.
On the other hand,

shade/L=1/y=-20 temp, geolon_t, z[gz=temp]

works!

My data depend on general curvilinear coordinates, in the netcdf-file 
this is to be seen
from the attribute:
                temp:coordinates = "geolon_t geolat_t" ;

However, the lat/lon grid (yt_ocean, xt_ocean) is identical with the 
curvilinear coordinates
and the curvilinear representation is pretty general but not needed here.

Am I right, when removing the attribute temp:coordinates = "geolon_t 
geolat_t", ferret would fall back to a
simple lon/lat representation? In this case it should help to remove the 
attribute?

How to proceed best?

Greetings,
Martin


Am 22.08.2014 19:09, schrieb Ansley Manke:
> Hi Martin,
> Starting with Ferret v6.9 we allow missing data in coordinate 
> variables for graphics commands:
> http://ferret.pmel.noaa.gov/Ferret/documentation/release-notes/version-6-9-release-notes#missing_coords 
>
>
> Ferret v6.9 should work fine with LAS v8, so I think if you upgrade 
> the Ferret executable, this data will plot in LAS.  I tried making 
> plots from the dataset and they are successful.
>
>    yes? use 
> "http://phy-50.io-warnemuende.de:8080/thredds/dodsC/genus/genus_run_81_5day_ocean.nc";
>    yes? shade/L=1/k=1 temp, geolon_t, geolat_t
>
> Ansley
>
> On 8/22/2014 7:37 AM, Martin Schmidt wrote:
> > Hi,
> > I am using LAS v 8.0 with ferret 6842. With Longitude-time of 
> > Longitude-depth plots I get the error:
> >
> > Bailing out of external function "samplexy_curv_nrst":
> >     Arg 2, curvilinear X coordinates, contains missing values
> >
> > Re-defining viewport FULL_VIEW
> > Re-defining viewport FULL_VIEW2
> >  **ERROR: error in external function
> > SHADE/KEY=CONT/LEVELS=v/TRIM/LEVELS=v/TITLE="  Practical Salinity   
> > (psu)" sampled_var0
> > Command file, command group, or REPEAT execution aborted
> > ERROR RUNNING SCRIPT
> >
> >
> > The error occurs only, if land points are included in the plot, but 
> > also, if the line ends
> > near a coast.
> >
> > The data are in geographical coordinates, but the structure is more 
> > general. From ncdump
> >         float temp(time, st_ocean, yt_ocean, xt_ocean) ;
> >                 temp:long_name = "Potential temperature" ;
> >                 temp:units = "degrees C" ;
> >                 temp:valid_range = -10.f, 500.f ;
> >                 temp:missing_value = -1.e+20f ;
> >                 temp:_FillValue = -1.e+20f ;
> >                 temp:cell_methods = "time: mean" ;
> >                 temp:time_avg_info = 
> > "average_T1,average_T2,average_DT" ;
> >                 temp:coordinates = "geolon_t geolat_t" ;
> >                 temp:standard_name = "sea_water_potential_temperature" ;
> >
> > geolon_t and geolat_t are 2D-versions of yt_ocean and xt_ocean.
> > I am not sure, if ferret (or the las) recognises this dependency. 
> > This may be a problem, since
> > geolon_t and geolat_t have missings, but only away from the coastline 
> > in the inner land.
> > For example  list/x=6.21:14.25/y=-22.58 geolon, geolat_t shows no 
> > missing value, the LAS
> > error shows up when a line covers only ocean points x=6.21:13.25.
> >
> > So my questions:
> >
> > - is LAS able to recognise such a dependency?
> > - it would be a big deal to remove these coordinates from the 
> > original file. Is it
> > possible to remove the "coordinates" attribute within a THREDDS or 
> > the LAS?
> > - is there possibly another source of this failure?
> >
> > The LAS and the data are public:
> >
> > http://phy-50.io-warnemuende.de:8080/las
> >
> > http://phy-50.io-warnemuende.de:8080/thredds/dodsC/genus/genus_run_81_5day_ocean.nc 
> >
> >
> > Any suggestion would be of great help.
> >
> > Martin Schmidt