- APPLICATION OF 1D AND 2D HYDRAULIC MODELING AND MAPPING WITH HEC-RAS 5.0 BETA David R. Markwood, PE [email protected] February 2015
Aug 12, 2015
-
APPLICATION OF 1D AND 2D HYDRAULIC MODELING AND
MAPPING WITH HEC-RAS 5.0 BETA
David R. Markwood, PE
February 2015
-
2
(Santa Clara River near Los Angeles)
-
3
Table of Contents
Introduction ............................................................................................................................................... 4
Methodology ............................................................................................................................................. 7
Results ..................................................................................................................................................... 15
Discussion ............................................................................................................................................... 18
References ............................................................................................................................................... 22
-
4
Introduction
There are a number of commercially available 2D hydraulic modeling software packages available; HEC-RAS
Version 5.0 Beta equips all hydraulic modelers with combined 1D and 2D capabilities, in addition to terrain
modeling using existing RAS model geometry, and mapping tools for interpreting and presenting results. HEC-RAS
is thoroughly vetted software that has been enhanced consistently since its inception, and remains available for
unlimited distribution. Some masterful hurdles maneuvered by RAS 5.0 for 2D simulations, as described by Mr.
Gary Brunner, author and leader of the HEC-RAS development team, and by the Resource Management Associates
group, include the ability to specify quite large cell sizes for representing 2D computational meshes while
minimizing impact to accuracy, unrestricted cell geometry for a 2D mesh and its boundaries, and partial wetting and
drying of cells during simulations (Brunner, 2014). These concepts are demonstrated in this investigation by
experimenting with multiple 2D mesh cell sizes for combined 1D and 2D modeling of a split flow loop network
system near the Gulf of Mexico, on the Mississippi coast. Remedies to unsteady flow and 2D modeling challenges
encountered are also discussed.
For the purposes of this report, the split flow loop network studied is called Flat Bayou, its watershed spanning
about 13 square miles of coastal plains with flood elevations around a dozen feet above sea level. The loop network
analyzed is about a quarter mile upstream of the mouth, and becomes increasingly active during runoff events.
Backwater events from downstream surge conditions, and the like, were not considered; the stream in fact drains to
another bayou before reaching the coast. Figure 2 shows a map of the split flow loop network, on readily available
imagery within RAS Mapper. The main stem of Flat Bayou is to the east, while the reach to the west drains Flat
Bayou Tributary and contributes relief drainage for the main stem during large events.
Figure 2. RAS Mapper schematic of Flat Bayou Split/Loop Flow System
The majority of the hydraulic reaches of the Flat Bayou network have been channelized. A 1D approach with a
single hydraulic reach and cross sections spanning the entire floodplain may well be sufficient for certain
applications with this system. However, an important objective here was to evaluate the variable water surface
elevations and depths at the bridge bisecting both reaches of the network, applying a combined 1D and 2D approach.
-
5
In either case, 1D or combined 1D and 2D, there are a number of areas upstream of the loop that would provide
some attenuation for an event loading the system. The figure below shows many of these storage areas. Ineffective
areas were applied in order to model these overbank areas as providing no active conveyance, but rather attenuation.
Figure 3. Attenuating Storage Areas Available Upstream of Split Flow Loop Network
Despite channelization, because of unnatural conditions occurring near the upstream junction during overflows, it
becomes increasingly difficult to figure the distribution of conveyance as a flood wave loads the system. The figure
below was taken from the River Hydraulics Engineer Manual (USACE 1110-2-1416, 1993), regarding a 1D
approach and depicting the need for using multiple hydraulic reaches in order to determine variable water surface
elevations along those reaches. Specifically, the image on the left shows the limitations of modeling (ignoring) split
flow situations using only a single reach and cross sections spanning the floodplain. The schematic to the right
shows a more realistic representation, with variable geometry and velocity and water surface elevations along each
reach and corresponding floodplain.
-
6
Figure 4 (adapted from USACE 1110-2-1416). Schematic of 1D Limitations for Split Flow Modeling (left) and a 2D
Representation (right)
For this evaluation, a large runoff event was considered, approximately a 1% annual chance event, and the system
response evaluated using multiple methods. The movement of a flood wave through the natural and developed
floodplains was of greater interest than the behavior of flow through the main corridors. A combined 1D and 2D
approach provided another, if not improved, idea of the split in conveyance and variation in flood elevations along
each reach that may be expected for a large event.
First, a 1D split flow optimization subcritical steady-state approach using junctions to tie the loop network together,
with natural cross sections to model flooding along both reaches. An alternative could have been to code in a lateral
structure(s) using the topography where the split, or spill, occurs. The 1D split flow system could also be modeled
under unsteady conditions. Yet another alternative, as was selected for this report, is using a combined 1D and 2D
hydraulic model. The representation of the system could have even been extended further by adding another 1D
hydraulic reach and connection to the 2D Flow Area, simulating contributions of Flat Bayou Tributary on the
system, which is directly connected to the auxiliary channel to the west.
Cross section spacing and orientation are perhaps the most important user inputs of a 1D HEC-RAS model for
minimizing violations to Gradually Varied Flow and other assumptions. The concern is virtually eliminated by
defining 2D Flow Areas in HEC-RAS 5.0 Beta. Because of the unique features for handling terrain and unsteady 2D
modeling, discussed in detail later in the report and here in summary, detailed hydraulic properties of each cell and
cell face within a 2D mesh can essentially be very well-defined merely by well-defined topography of the area—and
these cells can be quite large, irregularly shaped, and partially wet (Brunner, 2014).
Reasonable hydrologic inputs, hydraulic boundary conditions and runoff loadings play important roles in simulating
useful results. Hydraulic properties, much like those computed for a typical 1D cross section, are used in governing
the movement of water within and into and out of a 2D flow area, with the terrain providing much of the hydraulic
representation, as opposed to perhaps an input-intensive number of 1D model cross sections, requiring additional
assumptions such as average conditions between those cross sections, which may or may not be a good
representation. Fortunately for hydraulic modelers, LiDAR-based topographic data is becoming increasingly
available; terrain for this investigation was developed from LiDAR data available in the public domain.
-
7
The 2D modeling utilities offered in HEC-RAS 5.0 Beta have many potential applications in addition to those
discussed herein, including dam break modeling and emergency action planning, forensic H&H analyses, and in-
classroom and client demonstration. As Mr. Brunner states with refreshing enthusiasm in Combined 1D and 2D
Modeling with HEC-RAS, “The particle tracking visualization option…is extremely helpful in visualizing where
water is going, and the relative magnitude of the velocity. Try it out; it’s really fun and informative!!!” (2014). Sure
enough, the user-friendly software, combined with the upgraded terrain and combined 1D and 2D modeling and
mapping features extend the utility of this program to even a powerful teaching tool for young and experienced
engineers, alike. The reader is challenged to experiment with the particle tracing and other new features and not be
enthused; it really is fun.
Methodology
First, a steady flow analysis was performed using HEC-RAS 4.1 in order to compute flood elevations along the
study reach of Flat Bayou. Two separate reaches were defined for the 1D steady-state model, with junctions at the
upstream split and at the downstream confluence. This geometry consisted of multiple stream reaches, as shown in
Figure 5, allowing for variable water surface elevations along these reaches forming the loop network. The
floodplain geometry of the model cross sections was developed from LiDAR.
Figure 5. Flat Bayou 1D Split Flow Loop Network Schematic
Using junctions afforded the utility of the flow optimization routine at the upstream junction, improving the solution
for the split in conveyance (from an initial ‘guess’). It is worth noting the performance of this routine, available for
junctions where flow splits moving downstream through the floodplain in a steady-state analysis, is entirely limited
by the accuracy of the input, particularly the cross-sections nearest to and interfacing with the junctions bounding
the hydraulic reaches. A decent initial specification of the steady flow discharge distribution is also important for
narrowing the solution window and nudging the program towards a useful answer. Figure 6 shows a profile view of
the main stem of Flat Bayou in the 1D loop model, broken into three reaches and tied together by two junctions (tied
also with the western auxiliary reach), as well as a plan view of the 1pct floodplain (right).
-
8
Figure 6. Flat Bayou Main Stem Profile View and 1pct Floodplain Plan View
The 1pct discharge used for the main stem in the steady-state model, approximately 5,800 cfs, was developed using
the 7-parameter nationwide urban regression equations, as described in Flood Characteristics of Urban Watersheds
in the United States (USGS, 1984). The initial discharge used in the split reach was around 20% of the main stem
(total) discharge. The Calculation Tolerance settings (found via the Steady Flow Analysis Options menu) were
adjusted to ensure maximum allowable iterations for both general open channel and split flow optimization
computations, 40 and 60, respectively. The Flow Optimization option for the upstream junction was also activated.
Next, a combined 1D and 2D approach was implemented.
The 2D and combined 1D and 2D Flow Options, as specified within the HEC-RAS Unsteady Computation Options
and Tolerances menu, are shown in Figure 7. The maximum number of allowable iterations was set for all Flow
Options: 1D (40), 2D (40), and between 1D and 2D (20). Numerous tolerances were also relaxed, the sensitivities of
these changes deemed acceptable for this study. For specific example, the water surface tolerance for all three types
of computations was adjusted from the default hundredth of a foot to a larger tenth of a foot. The “Abort Tolerance”,
or maximum allowable error in water surface solution before aborting computations, was reduced to 20’ from the
default 100’. At this point, regarding the tolerances used, it is worth quoting a trailblazer of the industry, the late
Rooney Malcom: “We’re not making watches.” However, we are indeed aiming for accuracy, surely better than a
multi-story building. The reader is encouraged to reference the documentation provided with HEC-RAS 5.0 Beta for
more information regarding the dials to turn and adjustments available for troubleshooting and improving unsteady
models. Author Gary Brunner offers excellent in-depth discussion for the various options; not only how, but why,
they ought to be considered or deployed.
-
9
Figure 7. Unsteady Computation Options and Tolerances, 2D (left) and 1D/2D (right) Flow Options
The Geometry Preprocessor was run for the 1D split flow model geometry in order to highlight potential sources for
model instability prior to setting up the combined 1D and 2D unsteady flow simulations. Hydraulic properties
computed by the Geometry Preprocessor for all cross sections along the 1D reaches (of both the 1D split flow steady
model and the combined 1D and 2D model) were configured using the Cross Section Table Properties editor, as
shown in Figure 8. The resolution of the hydraulic property tables was refined, specifically at and near the elevations
of particular interest at cross sections, such as low chord or ineffective area trigger elevations.
Figure 8. Cross Section Table Properties Window, Hydraulic Table Parameters for 1D Cross Sections
Flat Bayou is low-lying in the coastal plain of Mississippi, so there are many areas along the overbank floodplains of
this study reach with enough relief to provide flood attenuation. In order to capture the effects of these storage areas
in the 1D split flow model, a number of multiple (or blocked) ineffective areas were specified for much of the 1D
model cross sections. The image to the left in the figure below shows discontinuities in hydraulic properties,
particularly storage area and flow area, computed by the Geometry Preprocessor for a particular cross section. This
cross section had ineffective flow areas defined in the overbanks set at elevations corresponding with these
discontinuities. In order to smooth the rapid transitions sure to cause instability, the ineffective flow areas for this
cross section were redefined as permanent (in fact, this change was not made until stability issues were later
encountered during unsteady simulations). Doing so maintained the area outside and below the ineffective elevations
for permanent storage, regardless of the simulated elevations. The result of this adjustment to the hydraulic
-
10
properties for this cross section is shown in the image to the right in the figure below; of particular note are the much
smoother Storage Area and Total Area curves.
Figure 9. Hydraulic Property Plots for a Riverine Cross Section – left figure shows discontinuities in hydraulic
properties, right figure shows smoothing of discontinuities
The Geometry Preprocessor also computed internal boundary (IB) curves for the bridges in all the hydraulic reaches
for the 1D split loop scenario, and for the only remaining bridge that was not replaced by the 2D Mesh in the
combined 1D and 2D scenario. The low chord of a bridge is typically a trigger elevation, at which computational
methods change rapidly. Without engineering consideration, flow assumptions can be violated entirely, simulated
results stable yet unrealistic, or more than likely, model instability dang near guaranteed. Unless the water surface
does not reach it, the low chord elevation is critical, along with bridge modeling method settings, overbank
geometry, contraction and expansion coefficients, and ineffective stations and elevations. The IB curves generated
for a structure depend on all of these inputs; rapid transitions or changes in hydraulic properties across various
tailwater elevations usually leads to computational instability. The hydraulic table parameters editor shown to right
of Figure 10 was used to smooth the IB curves computed by the Geometry Preprocessor for the bridge along the
main stem upstream of the split (left).
-
11
Figure 10. Bridge Geometry and Internal Boundary (IB) Hydraulic Property Tables Parameter Settings
The number of submerged curves—curves developed by the Geometry Preprocessor that provide headwater
elevations for a range of discharges and a single tailwater elevation—was maximized to 60. The number of points on
the free flow curve was increased, as were the number of points along the submerged curves. Maximum head- and
tailwater elevations were specified, with particular regard to, and therefore improved resolution near, flood
elevations expected from something like a 1pct runoff event. The maximum flow to be considered was also set.
These adjustments were made in order to further refine an IB curve near the elevations of interest, thereby narrowing
the solution window during unsteady simulations. Figure 11 shows curves computed by default (left), and the effects
of the adjusted Hydraulic Property Tables settings on the computed curves (right).
Figure 11. Internal Boundary Bridge Hydraulic Property Plots – image to the left shows discontinuities, image to the
right shows smoothing of computed IB curves, near the bridge low chord elevation.
The combined 1D and 2D hydraulic model developed for this analysis consisted of 1D hydraulic reaches up- and
downstream of a 2D flow area that encompassed the floodplain and channels of the loop network. An important step
in developing the combined 1D and 2D analysis was compiling the data to be used in modeling the behavior of the
2D flow area—specifically, a terrain model and spatial land roughness classifications—primary information used by
-
12
the software to compute hydraulic property tables for the cells and cell faces of a 2D mesh. HEC-RAS 5.0 Beta
allows a user to characterize the hydraulic roughness of the floodplain for a 2D mesh, specified by a polygon
shapefile of appropriate Manning’s n values (which is actually converted for use by RAS Mapper to a .tif of
specifiable resolution). A polygon shapefile with various roughness attributes, which was used in building the 1D
steady-state model, was used directly for defining the roughness of the 2D flow area in the combined 1D and 2D
model. The figure below shows the spatial roughness coverage (right) and attributed roughness (Manning’ n) values
(left). The program also allows for a global roughness value to be used for a 2D mesh.
Figure 12 Land Classification Roughness Editor and Land Classification Coverage (right)
A very cool terrain modeling feature available in HEC-RAS 5.0 Beta RAS Mapper allows a user to incorporate
geometric data from 1D model cross sections into a composite terrain model used for developing the hydraulic
properties within the 2D mesh. Figure 13 shows the prioritized terrain sources used, including a ‘channel’ terrain
model extracted from the 1D split flow model cross-section geometry.
Figure 13. RAS Mapper Terrain Builder Dialogue and Channel Geometry Terrain Model
-
13
Figure 14 shows a dialogue generated when building the terrain model with these prioritized sources and the channel
geometry terrain model (right). The significance of this tool is discussed in detail later in the document though
suffice it to say it is a powerful utility for leveraging available HEC-RAS geometry data. For example, surveyed
channel and structure sections, and in general, data that has already been compiled, can be integrated with other
terrain data into a more accurate terrain model.
Figure 14. RAS Mapper Terrain Builder Dialogue (left) and Channel Geometry Terrain Model (right)
Average cell sizes specified in this investigation for the 2D mesh of the Flat Bayou network included 200’ x 200’,
50’ x 50’ and 25’ x 25’. A 10’ x 10’ cell size was also implemented for the 2D flow area, though abandoned for this
analysis, as the computation time was prohibitive for any sort of iterating. The number of cells required to represent
the 2D mesh for 200’, 50’, 25’, and 10’ average cell sizes were 369, 6,152, 24,789, and 155,595, respectively.
Figure 15 depicts just how sharply a reduced cell size tends to increase the number of cells (i.e. computational
demand) in a simulation for a 2D flow area of a particular size; a similar depiction could be produced plotting the
computational times for geometry pre-processing, unsteady simulations, or post-processing for various cell sizes.
Figure 15. Plot of Number of Cells Required versus Cell Size for Modeling a 2D Flow Area
y = 2E+07x-2.018
R² = 1
0
20000
40000
60000
80000
100000
120000
140000
160000
0 50 100 150 200
Nu
mb
er
of
Ce
lls
Cell Size (feet)
-
14
In general, a computation interval (time step) ought to be fine enough to capture the changing flow conditions and
hydraulic characteristics during a simulation. Unsteady flow simulations perform much better in the absence of rapid
variations in hydraulic properties between cross sections. For the unsteady analyses, a 15-minute hydrograph using
the steady-state peak discharge was developed, fit to the SCS dimensionless unit hydrograph distribution, and used
as the upstream boundary condition and loading of the system. The HEC-RAS Version 4.1 Users Manual describes a
useful measure of the upper limit for what a particular computation interval should be—the duration of the rising
limb of the flood hydrograph divided by 20. For this analysis, the rising limb of the hydrograph is around 2.5 hours
(as shown in Figure 16), suggesting a time step of 7 minutes or less.
Figure 16. 1pct Upstream Boundary Condition Hydrograph
The computation interval suggested by “a second way of computing the appropriate time step” was about 2 or 3
minutes, based on the Courant condition (Brunner, 2014, p.89). Model instability prompted further reductions upon
subsequent iterations to a 30 second computation interval. It is worth noting, the Courant condition implies the
smaller the cell size, the finer the computation interval required. For this analysis, a 1 minute interval was used for
the 200’ cell size 2D mesh, 30 seconds for 50’, and 10 seconds for a 25’ x 25’ mesh.
Figure 17. Computation Messages for 1D/2D Simulations using 25’ (left) and 50’ Average Cell Sizes
-
15
The total time required for the combined 1D and 2D unsteady analysis—geometry pre-processing, unsteady
simulation, and output post-processing—was around 25 minutes using a 25’ cell size for the 2D mesh. The unsteady
simulation accounted for close to 99% of this time; the unsteady simulation was performed in less than 2 minutes
when using a 50’ cell size for the 2D flow area. Figure 17 shows the summary computation messages of each
simulation, with the much longer duration (25’ cell size, 10 second time step) to the left, and the 50’ cell size
simulation using a 30 second interval to the right (each utilizing the same ‘average’ computing power). Bumping up
to a 200’ cell size allows for a very short simulation, yet maintains relative accuracy compared to the smaller cell
sizes.
Results
For the purposes of this investigation, the relative differences were more interesting than the actual elevations. The
1D steady-state water surface elevations were about a foot higher than the 1D unsteady elevations, though the depth
computed at the bridge along the split reach to the west was a couple feet lower for the unsteady analysis. Greater
differences were found between the 1D and combined 1D and 2D methods, in general, and specifically between 1D
unsteady and combined 1D and 2D unsteady methods. The difference in elevations among all of the combined 1D
and 2D unsteady simulations was less than 0.8’ suggesting a limited, however evident, relationship between a
decrease in 2D mesh cell size and a decrease in simulated elevation for this study.
Figure 18 displays the more conservative floodplain delineation determined using a combined 1D and 2D approach,
particularly along the split reach to the west. The 1D split flow depth grid is shown in blue in both images, and on
top of orange and red 1D/2D results (image to the right).
Figure 18. Comparison of Inundations for 1D Steady (left) and 1D Steady Results (blue) Overlaid on Combined 1D
and 2D Unsteady Results (right, orange)
RAS Mapper provides a number of ways to view and display model geometry and computation results. Not only can
the hydraulic properties of each cell of a 2D mesh be plotted (or tabulated) by right-clicking on the cell, but water
surface, depth, velocity, and shear stress time-series results can also be viewed for a cell, post-simulation. Figure 19
shows a water surface elevation plot for a particular cell in the 25’ average cell size 2D mesh.
-
16
Figure 19. Example 2D Flow Area Cell Hydraulic Property Plot (Water Surface Elevation Hydrograph)
Velocity mapping is a useful feature available within RAS Mapper. The Hydraulic Tables editor (accessed from the
Geometric Data Viewer) allows the user to adjust the resolution of reported velocities at a given cross section.
Similar to water surface elevation and depth rasters, a velocity grid is generated for a simulation, by default.
Supplemented with directional vectors, which can be set via the Layer Properties of the velocity surface, the
magnitude of velocity and the direction of flow through a system can be well represented. Figure 20 shows velocity
results overlaid on the depth surface for the loop network.
Figure 20. Velocity Magnitude and Direction Vector Mapping on Depth Surface Mapping
-
17
An exciting extension of the velocity mapping utility in RAS Mapper is the Particle Tracing feature. This feature
provides an engaging way to interpret and present results of a system response to runoff. Figure 21 shows a still shot
of a Particle Tracing animation generated in RAS Mapper.
Figure 21. RAS Mapper Particle Tracing over Depth Surface
As previously mentioned, the terrain modeling capabilities of HEC-RAS 5.0 Beta are generally limited to compiling
a terrain model (.e.g. a .tif raster) based on prioritized sources. The right image of Figure 22 shows the results of the
terrain model compilation, with the roadway obstructions removed, ‘burned’ from terrain model generated from the
1D model cross section geometry.
Figure 22. Terrain Model Before and After Incorporating Channel Geometry from 1D model
-
18
Figure 23 shows the backwater effects caused by the road crossing along the split reach to the west using an
‘unburned’ terrain model (left image of Figures 22 and 23).
Figure 23. Comparison of Inundations for Combined 1D/2D Model – blue colors indicate floodplain boundary (left)
and depth (right) for simulation that does not incorporate cross section geometry from the 1D model into the terrain
model; red, orange, and green colors indicate results for a simulation using the compiled terrain model.
This concept is discussed in detail in Combining 1D and 2D Modeling with HEC-RAS, and demonstrated early in
that document in “Figure 2. Example showing the benefits of using the detailed sub terrain for the cell and face
hydraulic properties.” The figure shows how the movement of water is governed within a 2D mesh, able to move
freely within channels until reaching flood elevations and spilling into floodplains, and this relatively fine definition
of a 2D computational mesh is dependent much more on a relatively high resolution terrain and sub terrain (sub-
LiDAR) than on a fine cell size. It may go without saying, though should be noted anyhow—the steeper and more
variable the terrain being modeled, the finer the cell size that will be required. The example, in fact, demonstrates
how even a 500’ cell size is sufficient for representing the flow of water through 100’ canals, given say a 2’
resolution terrain surface, even though only portions of those 500’ cells are wet as water remains within the canal
banks. The same was demonstrated in this analysis, as a 200’ cell size provided an extremely fast computation time,
and very similar results, however slightly more conservative in these relatively flat areas.
Discussion
The RAS Mapper utility, relatively new to HEC-RAS software, and the new 2D and combined 1D and 2D
capabilities available in Version 5.0 Beta, demonstrate the commitment of the HEC to provide proven and state-of-
the-art tools to the hydraulic engineering community. The exciting features available in the latest releases prompted
this investigation, which only scratches the surface of the upgraded (and existing) utility. When it comes to 2D
hydraulic modeling, as previously discussed, perhaps the coolest upgrade available within HEC-RAS 5.0 Beta is the
methodology for handling flow between cells within a 2D computational mesh, enabling the use of relatively large
-
19
cell sizes for achieving detailed representations of 2D flow areas, as well as great flexibility in defining the geometry
of the 2D mesh.
The figure below was adapted from Combined 1D and 2D Modeling with HEC-RAS, demonstrating how the cells of
the 2D mesh have a virtually unrestricted geometry, “a mixture of cell shapes and sizes”, and particularly that
boundary cells can have detailed multipoint edges (Brunner, 2014). The figure also shows the Delaunay
Triangulation technique used by the program, and the resulting Voronoi diagram by which the cells are created
(obviously similar to the Thiessen Polygon method), as detailed by Brunner. A levee following a meandering river
bank can be represented and connected to 2D Flow Areas, with the boundary cells of the 2D mesh following the
irregularly shaped levee, and thus maintaining detailed hydraulic properties for irregular boundary cells and interior
cells alike, despite the size of these cells exceeding the smaller hydraulic features within a 2D area. The resolution of
the resulting delineation, however, is entirely dependent on the size and shape of the 2D computational mesh.
Figure 24 (adapted from Combined 1D and 2D Modeling with HEC-RAS, Wikimedia Commons). (Left) Figure 18.
Delaunay – Voronoi diagram example. And (Right) Figure 17. Description of HEC-RAS 2D modeling
computational mesh terminology.
The desired resolution of output (cell size) can often drive the cell size used in 2D hydraulic modeling. HEC-RAS
allows a user to define a multitude of 2D areas with specifiable average cell sizes, and allows for ‘peppering’ a
particular area, such as along a levee or other obstruction, with additional cell computation points. Therefore, it
seems reasonable that a relatively large cell size could be specified for a large 2D mesh, say a sizeable portion of the
floodplain of the Santa Clara River north of Los Angeles, Figure 25, and numerous additional mesh points could be
added near the tributaries, such as Bear Creek shown in Figure 26, as necessary to refine computational and
mapping resolution to a desirable (or acceptable) level for each source.
-
20
Figure 25. Santa Clara River Watershed Combined 1D and 2D RAS Mapper Geometry View
Figure 26. Bear Creek Combined 1D and 2D Depth Mapping, a Tributary of the Santa Clara River
-
21
Using a fine cell size for the entire area is prohibitive, perhaps even with expensive computing power, though this
peppering approach could be used in order to capture the finer behavior of the tributaries, and their interactions with
the main stem under particular runoff loadings. Perhaps ‘stitching’ together several 2D areas could be a viable
approach. It is evident even these approaches require engineering imagination and judgment to be entirely viable.
The author hopes for more automated tools for developing computational meshes with future releases; much more
so, for more pre-processing capabilities for developing models, such as quickly cutting georeferenced cross sections,
to go along with the tremendous post-processing tools now available within RAS Mapper. HEC-RAS has long had
capabilities for running various Monte Carlo simulations, via relatively light programming (Goodell, 2014), and
many users have surely noticed the Monte Carlo Analysis tool under the Run menu of RAS 5.0 Beta. This feature
has the potential to make extending a simple floodplain delineation resulting in a lumped polygon of flooding, to an
expressive probabilistic flooding delineation, almost elementary. Thanks to the mapping features of RAS Mapper,
more than ever the program can be used as a provocative teaching tool for clients and young students, alike. Not to
mention, entry-level civil engineers. In any case, from the beginning, the simplicity of HEC-RAS remains its
greatest strength as it continues to evolve and empower all hydraulic modelers.
-
22
References
Brunner, G. W., Hydrologic Engineering Center, Combined 1D and 2D Modeling with HEC-RAS,
October, 2014.
(http://www.hec.usace.army.mil/misc/files/ras/Combined_1D_and_2D_Modeling_with_HEC-RAS.pdf )
USACE, Hydrologic Engineering Center, River Analysis System HEC-RAS 5.0.0 Beta, Davis, CA,
October 2014.
USACE, Engineer Manual 1110-2-1416, River Hydraulics, 1993.
USACE, Hydrologic Engineering Center, HEC-RAS River Analysis System User’s Manual Version 4.1,
2010.
USGS. Water-Supply Paper 2207. Flood Characteristics of Urban Watersheds in the United States,
Alexandria, VA, 1984.
Goodell, C., Breaking the HEC-RAS Code: A User’s Guide to Automating HEC-RAS, 2014.