Summary of 30 August 1995 NUWG Meeting

• We considered the re-mapped satellite case, most similar to the gridded data case, and the navigation variables conform to the gridded data conventions. Andrew Sundelin presented an example CDL for the remapped AWIPS/ISPAN satellite data. The NUWG group approved this example, with the addition of a nav_model="GRIB1" variable to specify the navigation model, deletion of redundant "missing_value" attributes, and addition of valid_min and valid_max attributes to the image variable.

• We discussed whether the default for byte data should be signed or unsigned. Current documentation of the horribly-named signedness attribute says the default is unsigned, but Harvey Davies has pointed out that this conflicts with the current default _FillValue for netCDF byte data, which is -127. For now, data providers may use valid_min and valid_max attributes to indicate the range of values for byte data.

• We discussed possible values for the nav_model navigation variable. for parameterizations of georeferencing information. Previously we had considered the WMO GRIB edition 1 GDS specification and the U.S. Federal Geographic Data Committee (FGDC) Content Standards for Digital Geospatial Metadata available in text, Postscript, or PDF forms.

  Russ Rew promised to send information on the new GeoTIFF specification.

  GeoTIFF represents an effort by over 140 different remote sensing, GIS, cartographic, and surveying related companies and organizations to establish a TIFF based interchange format for georeferenced raster imagery.

  Code to build libgeotiff is freely available, including Niles Ritter's public-domain GeoTIFF key parsing library resides, with example programs and makefiles.

  Someone should look at this and evaluate whether GeoTIFF is useful for a navigation model of our georeferencing conventions. We discussed whether it would be better if we recommended a single navigation model in the NUWG conventions, rather than providing a list of usable navigation models, thus requiring applications or the udnav library to be able to handle all of them.

• We decided to table until the next meeting a discussion of a new proposed convention for coordinate reference variables (John Caron's terminology) that has some advabntages over the current dimension attribute convention. Details will be made available on a web page soon.