Can Stream Order Be Used as a Proxy to Determine Stream Navigability?
GEOG 6161 — MSGIS Capstone Project — Utah Stream Access Coalition
Introduction
Many Utah waterways were historically used to transport logs and other materials prior to statehood. Historical use of waterways for commerce is important in determining legal navigability and, consequently, public access. However, directly investigating every stream in Utah is both time- and resource-intensive, leaving many waterways unclassified.
This capstone project, completed in partnership with the Utah Stream Access Coalition (USAC), investigated whether stream order could serve as a preliminary proxy for identifying potentially navigable waterways. Rather than replacing historical or legal determinations of navigability, the objective was to develop a GIS-based screening method that could help prioritize waterways for additional investigation.
The analysis focused on the Little Bear–Logan HUC-8 sub-basin in northern Utah. Stream networks were derived from high-resolution elevation data and classified using both the Strahler and Shreve stream-ordering methods. Resulting stream-order thresholds were then compared with established major waterways to determine whether particular rankings could effectively identify streams with characteristics associated with navigability.
Figure 1) Little-Bear Logan Sub-basin study area
Figure 2) Little Bear-Logan sub-basin symbolized by Strahler stream order rank.
Methods
High-resolution LiDAR-derived Digital Elevation Models (DEMs) were obtained from the Utah Geospatial Resource Center (UGRC). Multiple DEM collections covering the study area were downloaded using a Python API script, while NHD major streams were used as the primary validation dataset and a HUC-8 boundary defined the extent of the Little Bear–Logan sub-basin.
The DEM tiles were mosaicked and standardized to NAD 1983 (2011) UTM Zone 12N with a common 3-meter cell size. The datasets were filled to remove sinks, combined into a continuous elevation surface, and clipped to the watershed boundary. This preprocessing substantially reduced the size of the original LiDAR-derived datasets while maintaining sufficient topographic detail for hydrologic modeling.
Hydrologic analysis was then performed in ArcGIS Pro. Flow Direction and Flow Accumulation were calculated from the DEM, and a threshold of 1,000 accumulated cells, representing approximately 9,000 m² of contributing drainage area, was used to define the modeled stream network. Because this network was generated from terrain rather than contemporary streamflow observations, it represented underlying drainage patterns rather than current hydrologic conditions.
The resulting network was classified using both Strahler and Shreve stream order and converted from raster to polyline features. Multiple thresholds were tested against NHD major streams and imagery to determine the balance between including major channels and excluding smaller tributaries. Strahler thresholds of ≥5, ≥6, and ≥7 were evaluated alongside Shreve thresholds ranging from ≥300 to ≥1,700.
Finally, the results were prepared for ArcGIS Online and ArcGIS Dashboards, creating an accessible product that could allow USAC users to examine modeled stream networks and use the identified thresholds as supporting evidence when evaluating potentially navigable waterways.
Results
The DEM-based hydrologic model successfully generated a complex stream network throughout the Little Bear–Logan sub-basin. Stream-order rankings generally increased as tributaries converged and flow accumulation increased downstream. The model performed particularly well in mountainous areas with pronounced elevation gradients, where drainage pathways could be clearly identified from terrain. Model performance decreased around bodies of water and transitions between mountainous and urban environments. In developed areas, some modeled streams followed unrealistic straight paths through residential areas. These errors were likely caused by terrain-based modeling being unable to account for constructed drainage infrastructure such as culverts, gutters, and storm drains.
Threshold testing revealed substantial differences between the two stream-order methods. Strahler ≥6 produced the strongest balance among the tested Strahler thresholds. Strahler ≥5 substantially overrepresented the major stream network, while ≥7 excluded many major channels. Because Strahler contains relatively few possible ranks, small changes in the threshold produced large changes in the resulting stream network (Figure 3).
The Shreve method provided finer control over stream classification. Shreve ≥650 produced a total stream length approximately 91.56% of the NHD major-stream validation dataset, while Shreve ≥500 produced approximately 101.5%, making it the closest tested approximation by total stream length. Lower thresholds increasingly overrepresented the stream network, while higher thresholds progressively excluded major waterways (Figure 4).
The comparison chart summarizes the stream order threshold relationship particularly well. Shreve thresholds changed more gradually than Strahler thresholds, allowing the potential upstream extent of larger waterways to be represented more precisely (Figure 5).
Figure 4) Streams with a Strahler stream order greater than or equal to 6 compared to major streams.
Figure 3) Streams with a Shreve stream order greater than or equal to 500 compared to major streams.
Figure 5) Traditional Empirical Orthogonal Functions, Leading 3 modes (EOFs 1-3) explain over 65% of the variability in the dataset
Discussion
The results demonstrate that DEM-derived stream order has potential as a preliminary indicator of potentially navigable waterways, but it cannot establish legal navigability on its own. Both Strahler and Shreve successfully identified prominent channels within the watershed, although the two methods provided different advantages.
Strahler ordering provided a relatively straightforward classification, with Strahler ≥6 producing a stream network approximately 108% of the length of the NHD validation network. Shreve ordering, however, offered substantially finer threshold control. Shreve ≥500 and ≥650 produced the closest total-length relationships to the validation dataset, while Shreve's more gradual ranking system better represented how far potentially navigable channels might extend upstream.
These thresholds should not be interpreted as definitive classifications of public access. Legal navigability depends on historical use or susceptibility to use at the time of statehood, information that cannot be determined from a modern DEM. Instead, the methodology provides a screening and prioritization framework. USAC could use GIS-derived stream order to identify waterways warranting more detailed historical investigation, allowing limited research resources to be focused on the strongest candidates.
Future research could improve the model by incorporating additional hydrologic variables such as rainfall and soil permeability and by applying the methodology to additional Utah watersheds. Comparing threshold behavior across multiple basins would help determine whether the relationships identified in the Little Bear–Logan sub-basin are transferable statewide.
Overall, the project demonstrates how GIS can complement historical and legal research by providing a scalable method for identifying potentially navigable waterways. Historical documentation remains necessary for determining navigability, but spatial analysis can substantially narrow the search area and support more efficient allocation of resources for evaluating and protecting public stream access.
Skills/Tools
ArcGIS Pro • ArcGIS Online • ArcGIS Dashboards • QGIS • Python • LiDAR / DEM Processing • Hydrologic Modeling • Flow Direction & Flow Accumulation • Stream Network Extraction • Strahler & Shreve Stream Order • Raster Analysis • Spatial Validation • Cartography • Web GIS