[thredds] a question about "forecastModelRunSingleCollection"

[Knut Arild Lisæter]

I am having problems with the "forecastModelRunSingleCollection" in  
thredds catalogs - this collection is based on netcdf files with names  
of the type

FORDAILY_startyyyyMMdd_dumpyyyyMMdd.nc

Each file has a time axis of length 1. The dataset is set up like this:

<datasetFmrc name="Test scanfmrc 4" path="fmrc/test/4" >
    <metadata inherited="true">
       <timeCoverage>
          <start>2008-04-09</start>
          <end>2008-05-20</end>
       </timeCoverage>
    </metadata>
    <netcdf xmlns="http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2 
" enhance="true">
       <aggregation dimName="runtime"
          type="forecastModelRunSingleCollection"
          timeUnitsChange="true" >
          <scanFmrc location="/opt/local/share/java/tomcat5/content/ 
thredds/public/topaz/"
             regExp=".*FORDAILY.*[0-9]{8}_.*\.nc"
             runDateMatcher="#_start#yyyyMMdd"
             forecastDateMatcher="#_dump#yyyyMMdd"/>
       </aggregation>
    </netcdf>
</datasetFmrc>

A collection of files such as this:

FORDAILY_start20080430_dump20080430.nc
FORDAILY_start20080430_dump20080501.nc
FORDAILY_start20080430_dump20080502.nc
FORDAILY_start20080430_dump20080503.nc
FORDAILY_start20080430_dump20080504.nc
FORDAILY_start20080430_dump20080505.nc
FORDAILY_start20080507_dump20080507.nc
FORDAILY_start20080507_dump20080508.nc
FORDAILY_start20080507_dump20080509.nc
FORDAILY_start20080507_dump20080510.nc
FORDAILY_start20080507_dump20080511.nc
FORDAILY_start20080507_dump20080512.nc

Gives me a 2D forecast time and runtime like this:
Dataset {
     Float64 valtime[runtime = 2][valtime = 6];
     String runtime[runtime = 2];
} fmrc/test/4/Test_scanfmrc_4_fmrc.ncd;
---------------------------------------------
valtime[2][6]
[0], 0.0, 24.0, 48.0, 72.0, 96.0, 120.0
[1], 168.0, 192.0, 216.0, 240.0, 264.0, 288.0
runtime[2]
"2008-04-30T00:00:00Z", "2008-05-07T00:00:00Z"
Seems like the offset is calculated relative to the first "start" date  
only?  In addition the 2D forecast selection is the only one which  
actually gives a data set - all the others (best estimate, constant  
offset etc etc) present me with no opendap access link.  I should also  
mention that the above datasetfmrc will fail if I remove the  
timecoverage metadata - then it seems to fail  when extracting the  
daterange from the data set:

[16/May/2008:14:44:23] ERROR thredds.catalog.InvDatasetFmrc : Error  
making catalog for fmrc/test/4
java.lang.NullPointerException
         at  
ucar 
.nc2.thredds.MetadataExtractor.extractDateRange(MetadataExtractor.java: 
251)
         at  
thredds.catalog.InvDatasetFmrc.makeCatalog(InvDatasetFmrc.java:302)
         at  
thredds.catalog.InvDatasetFmrc.makeCatalog(InvDatasetFmrc.java:151)
         at  
thredds 
.servlet.DataRootHandler.makeDynamicCatalog(DataRootHandler.java:1234)
         at  
thredds.servlet.DataRootHandler.getCatalog(DataRootHandler.java:1202)
         at  
thredds 
.servlet.DataRootHandler.processReqForCatalog(DataRootHandler.java:1117)
         at thredds.servlet.CatalogServlet.doGet(CatalogServlet.java:78)
...


Any hints on how to fix this would be appreciated.


Best Regards,
Knut Lisæter

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.unidata.ucar.edu/mailing_lists/archives/thredds/attachments/20080516/0f7ec712/attachment.html