Hydraulic model (HEC-RAS) of the Upper San Saba River between Fort McKavett and Menard, TX

This is a 2D Hydraulic model (HEC-RAS) for the Upper San Saba River between Fort McKavett and Menard, TX. Model geometry is based on USGS 3DEP data (2018), with underwater bathymetry “burned” in using cross-sections sampled in the field in 2018. The model was calibrated based on water surface and velocities measured during data collection.

Methods
The following is a summary of available data utilized for developing a bathymetric terrain for 2D hydraulic modeling using HEC-RAS. Data available for model calibration and validation is also discussed.

Available Data
Cross-section data was collected at approximately 350-foot increments with the following devices:

Table 1. Equipment used and their accuracy for field data collection.

Parameter

Equipment

Unit Accuracy

Location

GPSMap 64 Handheld GPS

10-50 feet

Velocity

Hach Velocity Meter (Model FH950.1)

0.1 feet/second

Depth

An adjustable “ruler” stick with feet as units

0.1 feet

Wetted Width

Laser Technology Inc. TruPulse 360r

3 feet to nonideal (natural) target

The data was collected from July 10 to 26, 2018. The USGS Gage on the San Saba River at Menard, TX (08144500) shows discharges varied from 8 to 18 cfs during the time period.

USGS 3DEP data was used in the HEC-RAS terrain. These data were collected at a 1-meter resolution over a period of two months (February 14 to April 22, 2018) over a range of discharges from 16 to 25 cfs.

Bathymetric Areas
3DEP represents areas where the terrain is inundated as a hydro-flattened surface (hydrosurface). These flat areas are apparent relative to the surrounding terrain. The 3DEP terrain was imported into HEC-RAS 6.2. The field data collection featured 200 cross-sections. HEC-RAS was utilized to create a bathymetric surface by interpolating 1-D cross-sections.

The field data were projected as a shapefile and brought into HEC-RAS for cross-section placement. Cross-sections were drawn from the field-measured location. Given that the 3DEP data collection was at a similar discharge, wetted widths at the field-measured location were compared to the width of the hydrosurface areas by the following procedure:

Hydrosurface areas are delineated with a shapefile.
1D XS from the HEC-RAS are exported to ArcMap.
The 1D XS are “clipped” to the hydrosurface extent using the hydrosurface shapefile.
The 1D XS lengths are calculated using the Geometry Calculator.
Wetted widths are compared to the clipped 1D XS lengths.

1D XS placement was shifted downstream or upstream from the field-measured location to try to conform to the wetted widths measured in the field. At the first iteration, 80% of the XS were able to be placed within 5 feet of the field-measured wetted width. Locations that diverge greater than 5-feet from the field-measured data were flagged for further inspection.

For the field-measured cross-sections, three depths and a mid-channel velocity were measured in the field. Given that the discharges and the wetted depths for the 3DEP data and the field-measured data collection are similar, a first iteration of bathymetric data formulation assumed measured depths were equivalent to the distance from the hydro-flattened surface on the 3D to the channel bed.

Three depths were measured in the field per cross-section, on channel left, right and mid-channel. The exact locations of these depth measurements were not recorded. Cross-section depths within 0.25 feet of each other, and average depth was used for the 1D cross-section formulation in these locations. For areas where the standard deviation of measurements per cross-section exceeded 0.25 feet, all three depths were used in delineating the cross-section.

Once the 1D cross-sections were processed, bank stations were shifted toward the center of the channel. This reduces high ridges during channel interpolation and ensures all channel points are beneath the hydrosurface. A TIF was created using HEC-RAS 6.2 from bank to bank (Channel-only option). The TIF was exported as a 3-foot resolution raster. The 1D interpolation captures channel centerline between measured cross-sections, but meanders and channel widening may not be covered by the interpolated channel. The channel raster was exploded in ArcMap using the Raster to Point tool. The points were then interpolated using the Inverse-Distance-Weighted interpolation tool (IDW). This creates a much wider-coverage terrain while maintaining fidelity to the points/channel-only raster, while also covering meanders and pools that conform to the energy grade between cross-sections.

The IDW surface is clipped to the hydrosurfaced area, and then mosaicked with the original 3DEP terrain to create a seamless bathymetric/topographic surface. Landcover was delineated using aerial photography. (4) types of landcover were identified: Dense woody vegetation, sparse shrub, channel areas, and grasslands.  Normal Manning’s n values, based on Chow 1959, were selected. The 2D HEC-RAS mesh was set to 40-feet square, with a breakline with 25-foot cell-spacing at the channel center.

The discharge event simulated for this segment was 15 cfs. A 1D computation of 15 cfs was used as a “hot-start” to fill the river channel, and then 15-cfs unsteady state simulation was run. The outlet boundary was observed for 15 cfs flow. The depth and velocity results for the timestep associated with 15 cfs were exported to ArcMap. The statistics (mean, minimum, maximum) for each 1D cross-section were pulled from the raster results using the Zonal Statistics as Table Tool.

The Table of 2D model simulation results was compared to the field-measured mid-channel velocities and average channel depth. If the field-measurement was within the range of the Zonal Statistics, then a 0 [unit] RMSE error was assumed. RMSE for the 2D model table was computed based on whether the model results were ‘High’ or ‘Low’ from the field-measurement. Then the error was computed from the Maximum and Minimum results respectively. 15 cfs is within the channel banks, this was also the dataset with most field-measured information. The Manning’s n for the channel was adjusted as a single roughness and then improved by adding more roughness types. Modification of the roughness type was dictated by 1) bed material recorded during the field survey, to justify the use of “Cobbly” bed roughness; 2) inspection of aerial photography and confirmation via site visits to justify areas as being “weedy and ineffective flow”.

The 3DEP data is available for areas of inundation calibration for low flows. The uncertainty of the areas of inundation may be quantified by delineating inundated areas from aerial photography (or from the 3DEP terrain). A discharge of around 120 cfs was captured with aerial photography in November 2018; a discharge around 45 cfs was captured with aerial photography in September 2020. These inundated areas can be used to calibrate/validate flows of a similar discharge magnitude. Inundated areas in the aerial photography were delineated with a shapefile polygon. Differences between the modeled inundated area at a similar discharge were inspected for corrections in landcover roughness and breakline delineations in the mesh for proper numerical computation.

Following calibration, the model has overall RMSE error in velocity of 0.26 ft/s and 0.56-ft depth error. Error per cross-section communicates the magnitude of possible error in specific areas of the reach.

Landcover Description

Chow 1959 Description, which has minimum/normal/maximum ranges (Manning's n Values (orst.edu))

Selected Roughness

Channel

(Main Channel or Mountain Streams)

Channel

sluggish reaches, weedy, deep pools (normal)

0.07

Channel2

clean, winding, some pools and shoals, some weeds and more stones (maximum)

0.05

Cobbly

no vegetation in channel, banks usually steep, trees and brush along banks submerged at high stages no vegetation in channel, banks usually steep, trees and brush along banks submerged at high stages (maximum)

0.07

Cobbly2

sluggish reaches, weedy, deep pools (maximum)

0.08

Ineffective Sec

sluggish reaches, weedy, deep pools (normal)

0.07

Ineffective Sec2

sluggish reaches, weedy, deep pools (maximum)

0.08

Ineffective Sec3

very weedy reaches, deep pools, or floodways
  with heavy stand of timber and underbrush (normal)

0.1

Weedy Reach

very weedy reaches, deep pools, or floodways
  with heavy stand of timber and underbrush (between normal and maximum)

0.12

Intermediate Zone

(Floodplains)

Grassy Floodway

Scattered brush/heavy weeds (maximum) or light brush and trees in summer (between normal and maximum)

0.07

Grassy Floodway2

Medium to dense brush (between normal and maximum)

0.1

Floodplain

(Floodplains)

Dense Woody

Dense willows, summer straight (minimum) or heavy stand of timber, down trees, little undergrowth (normal)

0.1

Sparse Shrub

Light to dense brush (various definitions, ranges from minimum to maximum)

0.08

NoData

Scattered brush, heavy weeds (between normal and maximum)

0.06

Summary of Assumptions:

Accuracy of discharge measurements of river gages (expect up to 10% error in discharge measurement). Data used for 2018 from the Menard, TX gage at San Saba River has been approved by USGS
3DEP terrains have a 1-m resolution.
Field-measured cross-section location (quality assurance/uncertainty related to the field-measured wetted width).
The data provided in 1D cross-section dimensions are minimal laterally, but rather high resolution longitudinally down the channel. River area between 1D cross-sections is interpolated. Bank to measured depth contours and measured depth to measured depth contours are uncertain.
Unmeasured pools and widening locations have assumed depths that conform to channel energy grade. Actual bathymetries are most uncertain in these locations relative to other channel areas, as channel slopes cannot be assumed to be constant in backwater areas.
The structure, number of culverts, and dimensions of low-water crossings were not collected nor input into the model.
Groundwater contributions and losses were not incorporated into the model.

Related Works:
Cushway KC. (2023). Go with the flow: impacts of high and low flow conditions on freshwater mussel assemblages and distribution. Master’s Thesis, Texas State University. 114 pp.

Cushway KC, Harris AE, Piercy CD, Mitchell ZA, Schwalb AN. (2024) Go with the flow: Impacts of high and low flow conditions on freshwater mussel assemblages and distribution. PLoS ONE 19(2): e0296861. https://doi.org/10.1371/journal.pone.0296861

Mitchell ZA. The role of life history strategies and drying events in shaping mussel communities: a multiscale approach [dissertation]. San Marcos (TX): Texas State University. 2020.

Mitchell ZA, Cottenie K, Schwalb AN. (2023). Trait-based and multi-scale approach provides insight on responses of freshwater mussels to environmental heterogeneity. Ecosphere 14(7): e4533. https://doi.org/10.1002/ecs2.4533

Mitchell ZA, Cottenie K, Scwhalb AN. (2023). Trait-based and multi-scale approach provides insight on responses of freshwater mussels to environmental heterogeneity. [Dataset]. Dryad Data Repository. https://doi.org/10.5061/dryad.msbcc2g3d