EDNA Stage 1 Blind Pass Processing Steps

Science Center Objects

Using well-tested GIS techniques, the NED database is being processed into a Stage 1 EDNA product. The bulk of this work is being done by the National Weather Service's Severe Storms Laboratory (NSSL) in Norman, Oklahoma. Additional processing is currently being provided by the USGS WRD office in Sacramento, California.

The NED is processed using blind-pass techniques to remove spurious depressions from the DEM. The resulting flow direction and flow accumulation data sets are also being developed, as well as an initial synthetic streamlines database. The Stage 1 reach catchments (drainage areas corresponding to each "link" in the synthetic streamline coverage) are also being derived. These catchments will be immediately useful in the NWS's AMBER implementation. This national catchment database is being attributed with catchment identification numbers according to Pfafstetter,which allows for efficient upstream aggregation and downstream tracing.

This blind-pass development provides the Stage 1 drainage basin delineations and synthetic streamline coverages which will be transferred to stakeholders involved in the Stage 2 assessment. QA/QC to be done at this stage includes initial conflation of the synthetic streams with the NHD and comparison of derived drainage basin boundaries with existing digital basin boundary coverages.

EDNA Stage 1 Processing Steps

The first stage in the development of the EDNA dataset is a highly automated blind-pass procedure. In order to optimize the blind-pass production, the Cataloging Unit (8-digit Hydrologic Unit) is being used as the processing unit. With the exception of the selection of the seed points necessary to approximate the Cataloging Unit, these procedures require no user intervention.

The selection of the Cataloging Unit as the processing unit was made for two reasons. First, one of the goals of the EDNA development is to assist in producing improved Cataloging Unit boundaries for the country (along with watershed and subwatershed subdivisions). Using the Cataloging Unit as the processing unit allows us to perform blind-pass quality checks on the resulting watershed boundaries. Secondly, the size of the data files produced from the 30-m resolution NED requires that an efficient processing unit be selected. By having the processing unit somewhat consistent with the hydrographic divides, the problems resulting when the data sets are merged together are minimized.

The EDNA dataset is being developed in the national Albers Equal Area projection with the following projection parameters:

  • 1st standard parallel:  29  30  0
  • 2nd standard parallel:  45  30  0
  • central meridian:  -96  0  0
  • latitude of origin:  23  0  0
  • false easting:  0
  • false northing:  0

Albers Equal Area Projection Parameters

1st standard parallel29 30 02nd standard parallel45 30 0central meridian-96 0 0latitude of origin23 0 0false easting0false northing0

The procedures for Stage 1 production were developed in ARC/INFO's AML. These steps are further described in the Stage 1 flow chart. The Stage 1 steps done for each Cataloging Unit are:

  1. The vector boundaries for the Cataloging Unit are extracted from the USGS 1:250,000-scale Hydrologic Unit coverage (HUC250k). This coverage is projected into Albers and buffered by 25,000 meters.
  2. The portion of the current NED covering the buffered region is extracted and projected into Albers with a cellsize of 30 meters. The associated NED metadata is also extracted.
  3. All depressions in the DEM are filled using the ARC/INFO FILL function. (NOTE: In areas with significant naturally occurring closed drainages, the large depressions in the DEM will need to be examined to identify natural vs. spurious sinks. This does require a second iteration with user intervention and will occur following Step #5 )
  4. The appropriate 1:24,000-scale DRGs are extracted. A shaded relief representation is made of the filled DEM.
  5. A polygon coverage is created of all the sinks filled in Step 3. The large sinks should be examined to ensure that naturally-occuring sinks are not inappropriately filled.
  6. Flow directions and flow accumulations are derived from the filled DEM.
  7. Synthetic streamlines are derived from the flow accumulation grid using a threshold of 5,000 pixels. With the 30m cellsize of NED, this corresponds to an upstream drainage area of 4.5 km2 or about 1.7 mi2. This threshold has been found to be a good approximation of the desired drainage density for NHD as well as providing an adequate basis for derivation of initial subwatershed boundaries. The synthetic streamlines are attributed with the flow accumulation values.
  8. The appropriate 1:100,000-scale NHD is extracted. (If the NHD is not available for the Cataloging Unit, the DLG can be used, but processing of the Cataloging Unit could, also, be delayed until the NHD is completed).
  9. The Stage 1 synthetic streamlines are digitally compared with the NHD and areas of discrepancy are flagged for review in Stage 2. Measure of the percent of agreement is produced.
  10. Candidate seed points for replication of the existing Cataloging Unit boundaries are generated by intersecting the existing 1:250,000 Cataloging Unit vector boundaries (Step 1) with the synthetic streamlines from Step 7. The user interactively edits the seed points to best recreate the Cataloging Unit boundaries.
  11. The Stage 1 EDNA based Cataloging Unit boundary is derived and checked for goodness-of-fit. Areas of deviation from the existing unit are flagged for examination in Stage 2.
  12. The reach catchments (drainage basins corresponding to each reach of synthetic streamlines) are produced.

Stage 1 Flow Chart

Click on Prerequisites for EDNA level 1 processing or Interactively edit seedpoints for further information.

Edna Methodology Stage 1 Flow Chart

Edna Methodology Stage 1 Flow Chart

Prerequisites for EDNA level 1 processing

Pre-existing geodata

  • 1:250,000 WRD digitized HUC polygon coverage (HUCS)
  • 1:24,000 one arc-second NED tiles in arc GRID format (56 gigabytes)
  • 1:100,000 DLG/NHD hydrography for reference and conflation
  • 1:24,000 DRGs (tifs) for reference/verification (~ 400 gigabytes)

Setup paths to scripts and data

These steps are not essential, but will eliminate having to specify the path when executing amls.

In setup.aml 

  • set .datapath1 variable to parent directory of covers, aml, ...
  • Define &amlpath in $HOME/.arc file to point to (common_directory)/aml

To begin processing

  • arc: &r nedh < cu# >
    • Example: arc: &r (common_path)/nedh <11140103>

Intermediate automatic EDNA processing

Pre-existing geodata

  • 1:100,000 DLG/NHD hydrography for reference and conflation
  • 1:250,000 WRD digitized HUC polygon coverage (HUCS)
  • 1:24,000 one arc-second NED tiles in arc GRID format (56 gigabytes)
  • 1:100,000 DLG/NHD hydrography for reference and conflation

To begin processing

arc: &r nedh < cu# >

  • Example: arc: &r (common_path)/nedh <11140103>
  • The first part of the aml runs without further user input
  • Extract digitized HUC polygon (11140103) and buffer it by 25000 meters
  • Project HUC_POLY coverage to geographic to determine which NED tiles need to be extraced
  • Extract and merge NED tiles for buffered area
  • Project elevation data to albers
  • Fill sinks in elevation data
  • Create polygon coverage of sinks
  • Create shaded-relief image(s) for background
  • Clip and project DRG for background
  • Create flow direction grid
  • Create flow accumulation grid
  • Create synthetic streamlines and attach attributes
  • Extract 1:100K DLG hydrography or NHD data for reference/conflation

Generate watershed seed points

Pre-existing geodata

  • Local HUC polygon coverage (HUC_POLY)
  • Synthetic streamline coverage
  • 1:100,000 DLG/NHD hydrography for reference and conflation

To begin processing

arc: &r getseedpts < cu# >

  • Example: arc: &r getseedpts 11140103
  • The first part of the aml runs without further user input
  • Synthetic streams (STREAM_LINES) and HUC (HUC_POLY) coverages are intersected
  • A new point coverage (HUC8_SEEDS) is created from the intersections
  • At the end of the non-interactive processing you will see an arcedit screen like this
  • Select and delete unwanted points
    • In this example, all obviously unnecessary points are at the extreme ends of a reach. They are shown as blue squares in this screen.
  • Zoom in to remaining points
  • Select and delete point on the short tributary
  • Select point from contributing huc (shown as blue square)
    • Arc command: calc huc# = 11140104 - This will be the start of a new polygon for the contributing HUC, effectively "pinching off" the current HUC (11140103)
  • Final graphic showing the final two seed points
  • When finished editing points, arc: &ret to complete the arcedit session

To create watershed boundary

  • &r (path)/wshed <11120101>