Process narrative for construction of 30 m. WRB DEM

1. Obtaining data

As part of its Baseline97 collaborative project, the Oregon State Service Center for GIS coordinated assembledge of 30 m., 7.5 min. digital elevation models into a statewide database. The DEM's are organized in one degree blocks as described in this file. Data for the nine, one degree blocks contained within the green line in this image were transfered via ftp to the Sun Sparc20 at the UO Data Services Laboratory.

2. Preparing the data

  1.  Uncompress one degree blocks (e.g.):  unzip 43121.zip

  The quads are seven directory layers below the current one after this step:

   extracting: HOPPER/BASELINE/DIRECTOR/LATTICES/45122/H/8#/5122H8LG.ZIP
   
  each of which must be individually decompressed in the next step.

  2.  Uncompress 7.5 min. quads:  

        find . -name \*.ZIP \-exec unzip {} \;

  This creates 64 ArcInfo export files per one degree block, e.g:

        Archive:  ./HOPPER/BASELINE/DIRECTOR/LATTICES/45122/H/8#/5122H8LG.ZIP
  exploding: 5122H8LG.E00

  These are then moved to a convenient working directory in the next step.

  3.  Move 7.5 min quads to working directory:  

        find . -name \*.E00 \-exec mv {} /scratch/ise/ \:

  4.  The Arc 7.0.4 IMPORT command will not recognize export files whose 
names contain upper case characters.  The first step is to create a file
containing the names of these files:

        ls *.E00 >> names  

  The second step is to process the export file names using an awk 
script processed by the gnu awk interpreter:

        gawk -f fnm.awk names

  This changes a name like 5122C2LF.E00 to 5122c2lf.e00

  5.  Then perform the Arc imports using another gawk script in two steps:  

  The new set of import file names are, like the original ones, placed into a
file via the ls command:

        ls *.e00 >> newnames

  This file is used as input to the demimp awk script:

        gawk -f demimp.awk newnames

3. Assembling the quads

 1.  Merge quads using LATTICEMERGE command:  

  This command has a limit of 49 input files per execution requiring the merge
to be done in multiple passes.  This AML is an example of the method used.
 

4. Evaluating the results

    These results were examined for defects some of which are evident in this image of the 45122 one degree block. The size of these gaps will change depening on which quads are selected in a particular merge operation, but they are present both when processed by the Arc latticmerge command and the GRID MERGE function. In this map the red lines show the edge gaps and the green line is the basin boundary. In this low-relief detail area north of Eugene, the larger east-west gap is roughly 750 m. wide. A similar situation is present in areas of high relief in which a less conspicuous horizontal banding is also visible.

5. Solving gap problems

    The lattices provided by SSCGIS have been projected to a standard Lambert projection adopted for the state. Edge matching difficulties in merging the grids arise because of the use of this non-rectangular projection. Back- projection of the Lambert grids to UTM, for example, is not a potential solution because the grids produced have differing cell center spacings.
    The solution adopted here is to use the 'raw' ASCII DEM'. Even when produced by the same agency, however, these files have differing units, feet (1) and meters (2), and some, identified in the z_un_raw field as 0 and 3, have values not described in the acompanying documentation. DEMLATTICE treats the 0 as no unit, and places %2% in the Z UNITS field of the projection file for an input Z unit code of 3.
    For DEM's with elevation values in feet (1), the z_min and z_max values provided in the SSCGIS documentation are expressed in meters, whereas the actual values in the header records of these files are in feet. This, and the 0 and 3 Z units problems means that all of the DEM's must be checked and corrected before processing. Since multiple versions exist for some of the quadrangles, we began the process by using the USGS version in any case in which one was available, the BLM version next, and USFS last. Other data quality issues included the presence of negative elevation values in some grids, and anomalies in slope data derived from the DEM.