The Challenge

The question was how four major tributary basins within the Upper South Fork New River watershed compare using terrain, drainage-network, and basin-shape characteristics. Because those comparisons depend on how the drainage network and basin units are defined, the first task was to make sure the analytical units were appropriate before interpreting a ranking.

The initial Stream Link workflow produced 51 incremental link catchments with a combined area of 87.672 km². These were small local units rather than comparable tributary basins, so I did not use the initial catchments as the final ranking framework. I restructured the analysis around four major tributary basins and retained the intervening Mainstem Local area separately rather than forcing it into the tributary comparison. A later notebook QC step found that 2.988 km² of the final 30 m analysis extent had not been assigned to any of the 51 catchments; why those cells were left out was not determined.

Before ranking, the notebook compared the final zones' total area with the watershed boundary and tested for overlap among zones. During an October 12, 2026 portfolio review, I added footprint-level checks: coverage gap, excess area, pairwise overlap, and raster-to-vector correspondence. Those later measurements did not inform the original restructuring. All of these are internal consistency checks; they are not field validation or calibration of hydrologic performance.

"The first delineation produced 51 incremental link catchments, not comparable tributary basins. I did not use the initial catchments as the final ranking framework. I rebuilt the analytical units and required the area and overlap checks to pass before calculating the ranking."

Process

  1. 1
    Prepare the watershed and elevation model Clipped USGS SRTM 1 Arc-Second Global elevation data to the Upper South Fork New River watershed and projected the analysis to WGS 84 / UTM Zone 17N (EPSG:32617) so distance, area, and terrain measurements could be calculated in metric units.
  2. 2
    Build the D8 hydrologic model Filled DEM depressions, generated D8 Flow Direction and Flow Accumulation rasters, and applied a 1,000-cell accumulation threshold — approximately 0.9 km² of contributing area at 30 m resolution — to define the modeled drainage network.
  3. 3
    Extract and structure the stream network Created Stream Link and Strahler Stream Order products, converted stream links to vector features, and used the ordered network to distinguish the main channel from major tributary connections.
  4. 4
    Identify comparable tributary basins Used the ordered drainage network to identify the major tributary outlets — Strahler order-2 links whose downstream end lies within 45 m of the order-3 main channel — and delineated Basins 11, 14, 16, and 17 upstream of them. Residual cells associated with the mainstem/intervening drainage area were retained as Basin 9999, Mainstem Local, so they remained visible in the spatial accounting but were not treated as a fifth tributary basin.
  5. 5
    Calculate morphometric indicators Combined basin geometry, stream-network measurements, elevation statistics, and slope statistics to calculate drainage density, stream frequency, relief ratio, mean slope, bifurcation ratio, and circularity ratio for the four tributary basins.
  6. 6
    Check, then rank Before ranking, compared final zone total area with the watershed boundary and checked for overlap among zones. Then converted the six indicators to ordinal ranks and combined them into the composite factor. During final map production, a separate geometry check inspected the narrow Mainstem Local zone (ID 9999, one multipart feature of 44 parts).

How the Ranking Works

The ranking is a relative morphometric comparison of four basins under the assumptions described here. It is not an established intervention priority or a calibrated hazard assessment.

Each indicator was ranked across the four tributary basins. Higher drainage density, stream frequency, relief ratio, mean slope, and bifurcation ratio received lower numeric ranks, where rank 1 represented higher relative priority. Circularity ratio was ranked in the opposite direction. The six indicator ranks were equally weighted and averaged to produce the composite factor, so a lower composite factor indicates more high-priority ranks. The code assigns averaged ranks to tied values; no ties occurred among the four basins.

These directions follow the assumption recorded in the notebook code: higher values of the first five indicators, and lower circularity (a more elongated basin), are treated as indicating greater runoff/erosion susceptibility. The project files cite no literature for these directions or for equal weighting. They are implemented analytical assumptions, not validated relationships.

  • Drainage density — total modeled stream length per basin area (km/km²); higher values received higher priority.
  • Stream frequency — count of modeled stream links (Stream Link segments) intersecting the basin, per basin area (links/km²); higher values received higher priority.
  • Relief ratio — basin relief divided by basin length (longest upstream flow path), both in meters, so the ratio is dimensionless; higher values received higher priority.
  • Mean slope — average terrain slope in degrees; higher values received higher priority.
  • Bifurcation ratio — dimensionless mean ratio of stream-link counts in successive Strahler orders; higher values received higher priority.
  • Circularity ratio — dimensionless basin compactness measure (4πA / P²); lower values received higher priority in the implemented ranking.

Stream frequency and bifurcation ratio both count modeled links rather than whole Strahler-ordered streams. That implementation choice affects their values and interpretation, and results may differ from studies that count streams under another convention.

Rank sums across the six indicators were 13 for Basin 14, 14 for Basin 11, 15 for Basin 16, and 18 for Basin 17, giving composite factors of 2.167, 2.333, 2.500, and 3.000. Form factor and elongation ratio were calculated but left out of the composite because, like circularity ratio, they describe basin shape and would weight shape more heavily.

The four labels — Very High, High, Moderate, and Low — are assigned by position among the four basins (first through fourth by composite factor), not by calibrated thresholds. No sensitivity analysis was performed on the indicator selection, ranking direction, stream threshold, or equal-weight assumption. Mainstem Local was excluded because it is an intervening/mainstem reference zone rather than a directly comparable tributary basin.

Metadata & Documentation

Key project datasets were documented in ArcGIS Pro using FGDC CSDGM metadata so source, processing, coordinate reference, attributes, quality checks, limitations, and appropriate use are recorded with the data rather than existing only in project notes.

For the watershed boundary, provenance was traced to the USGS Watershed Boundary Dataset and HUC12 050500010201, Headwaters South Fork New River. For the final priority dataset, metadata documents its raster-to-vector derivation, dissolution by analytical zone, analytical field schema, conceptual-consistency review, completeness assessment, and appropriate-use limitations.

The ArcPy notebook, File Geodatabase, and CSV results are local project files and are not published with this case study. The notebook runs only inside the open ArcGIS Pro project, because it reads the current project and its active map; it requires the Spatial Analyst extension and expects the clipped SRTM DEM and the watershed boundary to already be loaded as map layers.

Metadata documented: Source & provenance · CRS / datum · lineage · geoprocessing history · attribute definitions · constraints · conceptual consistency · completeness QA/QC
Final Priority Zones — Processing Lineage Dataset lineage in ArcGIS Pro records the final priority layer's raster-to-vector derivation, geoprocessing steps, and analytical schema.
Watershed Boundary — Data Quality & QA/QC Dataset-level QA/QC documentation records conceptual consistency, completeness, review dates, evaluation methods, and the limitations of the assessment.

What the Analysis Revealed

90.633 km² USGS WBD watershed boundary area (HUC12 050500010201) used for area QA/QC
4 major tributary basins ranked in the final morphometric comparison
0.029% net total-area difference between the final raster-derived zone total and watershed boundary
  • Basin 14 — Very High: composite factor 2.167 across 16.227 km², the highest relative priority among the four tributary basins.
  • Basin 11 — High: composite factor 2.333 across 10.655 km².
  • Basin 16 — Moderate: composite factor 2.500 across 30.363 km², the largest ranked tributary basin by area.
  • Basin 17 — Low: composite factor 3.000 across 19.126 km².
  • Mainstem Local: 14.288 km² retained as an intervening/mainstem reference zone for spatial accounting, but excluded from tributary ranking because it is not a tributary basin delineated on the same basis as Basins 11, 14, 16, and 17.
  • QA/QC (notebook): the complete_zones raster contains 100,733 30 m cells, or 90.6597 km². The watershed polygon area is 90.6330809 km², a net total-area difference of 0.0266191 km² (0.02937%). The notebook's aggregate overlap check (sum of zone areas minus dissolved area) returned 0.0 km².
  • QA/QC (added in the October 2026 review): net agreement does not by itself establish footprint coincidence. Comparing footprints, 0.208 km² of the boundary lies outside the final zones and 0.235 km² of the zones lies outside the boundary, a combined footprint difference of 0.443 km² (0.489% of the boundary area); the net figure is small because these largely offset. These differences are concentrated within 45 m of the boundary, consistent with raster–vector boundary effects. Their precise cause was not independently isolated. Pairwise intersection of the five final zones found 0 m² of overlap.
Interpretation note: this is a relative morphometric comparison among four selected tributary basins, not a direct measurement of observed flood, erosion, sediment loss, or regulatory hazard. Results depend on DEM resolution, D8 routing, the 1,000-cell stream threshold, outlet/delineation choices, indicator definitions, ranking directions, and equal weighting. Mainstem Local is retained for spatial accounting but is intentionally excluded from the tributary ranking.
QA Decision
Settle the analytical units before ranking

The initial 51 incremental link catchments were not the comparable units the question needed, so I did not use the initial catchments as the final ranking framework. I rebuilt the units around four tributary basins, kept the residual mainstem area explicit, and gated the ranking on total-area and overlap checks. The footprint measurements reported above came later, from the October 2026 portfolio review, and did not inform that decision.