Watershed Delineation and Hydrologic Analyses

Project Overview

Delineate a watershed of choice using a publicly-available DEM
Find stream network and stream order from flow accumulation
Summarize watershed characteristics

Tutorial for this exercise by the Utah Geospatial Consortium can be found here. Data was sourced from the USGS National Map. All maps are plotted in NAD 1983 UTM Zone 12N.

Watershed Map

Reference map showing the location of Cedar Breaks National Monument and proximity of the watershed to Cedar City in southwestern UT.

I decided to delineate the watershed of Cedar Breaks National Monument, which is located south of Brian Head, UT. The Monument was carved from ancient stream channels, leading to a dramatic landscape of terraces, hoodoos, and exposed red rock.

The watershed boundary was determined based on flow direction, flow accumulation, and a designated pour point. Flow direction and flow accumulation were derived from a 10m 3DEP DEM. The pour point was manually selected to represent the ultimate outlet of the drainage basin.

A higher resolution version of this map can be found here.

Photograph of Cedar Breaks National Monument by Mac Cutler.

Stream Network

Table 1 - Stream network characteristics.

The stream network was defined using flow accumulation. A break value of 1000 was used to isolate cells with high amounts of incoming flow. This value was selected after comparison of results with tributaries represented in the National Hydrography Dataset. Length values were rounded to the nearest 10th meter since the DEM resolution was 10m.

Watershed Characteristics

Table 2 - Landscape characteristics within the watershed boundary.

Photograph of Cedar Breaks by Rob Whitmore (NPS).

The results for slope appear drastic at first glance but make sense when considering the landscape characteristics of Cedar Breaks. A slope of 0 likely exists somewhere within the watershed boundary due to the flat tops of terraces (see picture on the left). The maximum slope value of 80 also makes sense because the drop angle between some terraces at Cedar Breaks is nearly straight down. A mean slope of 20 indicates that in general, most slopes in this watershed are gentle. This value makes sense given that Cedar Breaks (and its steep hillslopes) only represents a small portion of the watershed as a whole.

I am more suspicious of the mean aspect value even though I did calculate the average accounting for circularity. A mean aspect of 359 indicates that the average hillslope in this watershed faces almost exactly due north. I'm not sure how to tell if this is accurate or not, but this value seems unlikely to me after reviewing the topography and raw aspect data. I also noticed that when I ran the aspect tool, "North" was represented by aspect values 0-22.5 and 337.5-360, which is nearly 20 degrees larger in range than other directions.

Mastery Obsverations

I checked that flow accumulation ran properly by overlaying it as a transparent layer on the DEM hillshade. The accumulation points were clearly located within channels. This indicated that the tool was running as I intended it to.
When I initially ran the watershed tool, the lower left portion of the watershed was clearly clipped off. To fix this problem and incorporate the whole watershed, I redid my raster clip to include a larger area. I ran the tool again with the larger DEM and it worked great!
I played around with cutoff values for the accumulation break and settled on 1000. This seemed like a good cutoff value based on a comparison with NHD for the area.
I am suspicious of my mean aspect values. I ran the tool a bunch of times, converted the raster to an integer, and still got the same value. Unsure if I did anything wrong or if the average aspect really is 359 for this watershed.
I noticed after running everything that my hillshade looked a little wonky (there were light diagonal lines across the whole landscape). I realized at that point that I had done something wrong when I reprojected my raster. It turned out that I had run nearest neighbor sampling instead of bilinear interpolation. Once I reprojected, the lines went away and the hillshade looked great. But...I did have to redo the whole exercise with the updated DEM after that. Oops.

Page updated

Google Sites

Report abuse