Re: [thredds] aggregation error - illegal range for dimension?

Hi Erich,

[I'm a bit rusty on NcML aggregations and John is out of the office this
week. But here goes ...]

The "joinExisting" section of the NcML aggregation page [1] mentions that

    "[a]ny variable that has [the aggregation] dimension as
    its outer dimension is an aggregation variable. The
    dimension must be the outer (slowest varying) dimension."

I always forget the order of the dimensions in an ncdump. But I notice
that all the examples on the NcML aggregation page [1] show the
aggregation dimension ("time" in this case) as the first one listed in
the ncdump output, e.g.

   double P(time=31, lat=3, lon=4);

Whereas your ncdump output below shows the aggregation dimension ("day")
last in the ncdump output

   float wind_speed(lat, lon, day) ;

What is the order of the dimensions in the datasets that you are
successfully aggregating?

If that doesn't seem to pan out, can you send us (or point us to) some
sample data files? Five or six consecutive year files should do.

Hope that helps,

Ethan

[1]
http://www.unidata.ucar.edu/software/thredds/current/netcdf-java/ncml/v2.2/Aggregation.html#joinExisting


On 8/30/2013 2:30 PM, Seamon, Erich wrote:
> Hello all – Im trying to aggregate a series of nc files that have ‘day’
> as the time dimension – with 365 entries in each file – representing
> daily readings. 
> 
> This is a piece from my catalog.xml
> 
> <dataset name="MET Maximum Reflectance 1979-2012 Aggregated"
> ID="agg_met_rmax_1979_2012_WUSA" serviceName="all" harvest="true"
> urlPath="agg_met_rmax_1979_2012_WUSA.nc">
> 
>     <netcdf xmlns="http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2";>
>       <aggregation dimName="day" type="joinExisting" timeUnitsChange="true">
>         <scan location="/xxxx/xxxx/xxxx/MET/" regExp="rmax.*\.nc$"
> subdirs="false" />
> 
>       </aggregation>
>     </netcdf>
>   </dataset>
> 
> 
> When I attempt to access a subset more than one year – I get a
> javaexception – ‘illegal range for dimension 2:  last requested 753 >
> 365 max’  
> 
> Ncdump of a file that im trying to aggregate: 
> 
> netcdf vs_2012 {
> dimensions:
>         day = 366 ;
>         lon = 1386 ;
>         lat = 585 ;
> variables:
>         float lon(lon) ;
>                 lon:units = "degrees_east" ;
>                 lon:description = "longitude" ;
> 
>         float lat(lat) ;
>                 lat:units = "degrees_north" ;
>                 lat:description = "latitude" ;
> 
>         float day(day) ;
>                 day:units = "days since 1900-01-01 00:00:00" ;
>                 day:calendar = "gregorian" ;
>                 day:description = "days since 1900-01-01" ;
> 
>         float wind_speed(lat, lon, day) ;
>                 wind_speed:units = "m/s" ;
>                 wind_speed:description = "Daily Mean Wind Speed" ;
>                 wind_speed:_FillValue = -9999.f ;
>                 wind_speed:esri_pe_string =
> "GEOGCS[\\\"GCS_WGS_1984\\\",DATUM[\\\"D_WGS_1984\\\",SPHEROID[\\\"WGS_1984\\\",6378137.0,298.257223563]],PRIMEM[\\\"Greenwich\\\",0.0],UNIT[\\\"Degree\\\",0.0174532925199433]]"
> ;
>                 wind_speed:coordinates = "lon lat" ;
>                 wind_speed:height = "10 m" ;
>                 wind_speed:missing_value = -9999. ;
> }
> 
> 
> --
> 
> 
> I have other nc files that are very similar, and they aggregate fine. 
> The main difference is, the files that aggregate correctly have a ‘time’
> dimension, not ‘day’, and they use the ‘noleap’ calendar.
> 
> The files that do NOT aggregate have ‘day’ as the time dimension and use
> the ‘gregorian’ calendar.
> 
> --
> 
> Im thinking that my aggregation xml is not structured right, maybe I
> should use JoinNew and declare ‘day’ as a coordinate variable?  It seems
> that JoinExisting would work given that the ‘day’ dimension exists in
> all files to aggregate.
> 
> Anyway, I would appreciate any help, and can provide other log
> information as necessary. 
> 
> Im running the most recent stable version of TDS, on linux/red hat,
> under tomcat6
> 
> Thanks!  
> 
> Erich

-- 
Ethan Davis                                       UCAR Unidata Program
edavis@xxxxxxxxxxxxxxxx                    http://www.unidata.ucar.edu



  • 2013 messages navigation, sorted by:
    1. Thread
    2. Subject
    3. Author
    4. Date
    5. ↑ Table Of Contents
  • Search the thredds archives: