Watershed
Comparison Using
ArcGIS
for Flood Plain Mapping
Prepared by
Kameron
Ballentine
Jeremy Jensen
Prepared for
Dr. Neale
CEE 6440
Utah State University
12/7/2007
TABLE
OF CONTENTS
Page #
INTRODUCTION
.
.1
WATERSHED
COMPARISON OBJECTIVES
...
1
WATERSHED
COMPARISON METHODOLOGY
2
WATERSHED
COMPARISON DATA
..
.6
WATERSHED
COMPARISON RESULTS
.
6
FLOOD
PLAIN DELINEATION OBJECTIVES
.7
FLOOD
PLAIN DELINEATION METHODOLOGY
.
7
FLOOD
PLAIN DELINEATION RESULTS
.
14
SUMMARY
AND CONCLUSIONS
..15
RECOMMENDATIONS
....
.16
INTRODUCTION
On
June 3, 2005 a debris flow started in Cedar Canyon and traveled down the canyon
towards Cedar City. This debris flow
damaged state highway SR-14 (Giraud
& Lund, 2005) which comes into Cedar City from the east.
There was a state of emergency declared in Iron County due to this
debris flow and other flooding taking place from spring storms and other rapid
snow melt. From this, concerns about
the flood capacity of local rivers and the accuracy of flood maps arose. With the national flood rate insurance map
(FIRM) project underway, the state of Utah decided to verify the capacities of
the rivers and remap the floodplains in Iron County. Two of the rivers that had their floodplains
remapped were Coal Creek and Parowan Creek.
Coal
Creek and Parowan Creek are located in neighboring watersheds named Coal Creek
Watershed and Parowan Watershed, respectively.
About 75 years of historical flow data is available at the mouth of the
Coal Creek where it enters Cedar City, but less than ten years of data were
available for the Parowan Watershed. From the historical Coal Creek flow data,
the flow rate for a 100 year flood can be determined and used in the flood
plain analysis, but there is not sufficient gauge data to produce an accurate
flood frequency curve for Parowan Creek. Without this information, the flow
rate for the 100 year flood could not be determined with the same
methodology. However, since the two
watersheds neighbor each other, their characteristics should be similar. A relationship based on watershed drainage
area was developed to determine the 100 year flood for Parowan Creek.
WATERSHED
COMPARISON OBJECTIVES
The
original scope of this project included comparing rainfall data, slope, and
mean elevations of the two watersheds. However, it was found that these were
previously compared in a study performed by Bowen, Collins, and Associates. As a result of this finding, a new scope was
developed to further analyze the hydrologic similarity between the Coal Creek
and Parowan Watershed. The new scope includes the following comparisons:
·
Watershed
Size
·
Drainage
Density
·
Available
Flow Data
·
Spring
Density
·
Soils
Data
These
comparisons were chosen on the basis of their availability and relationship to
the hydrology of the watersheds. Snow pack was also going to be used to compare
the two watersheds, but no Snotel data was found to
exist within the project boundaries or close enough to the project boundaries
to be deemed accurate for the Coal Creek and Parowan Watersheds.
WATERSHED
COMPARISON METHODOLOGY
In
order to perform an accurate analysis of the watershed, a 10-meter Digital Elevation
Model (DEM) for Iron County was obtained from the AGRC website. A hill shade of
the area was produced using the spatial analyst toolbar so that the terrain
could be more easily visualized as shown in Figure 1. The two watershed
boundaries are shown also shown in the figure. From this figure, it can be seen
that the two watersheds are touching each other and that they are located in
very mountainous terrain.
.
Figure 1: Hill
Shade for Iron County
As
mentioned previously, the slope and aspect of these two watersheds were
compared in a previous study. In order to perform a full comparison of any two
watersheds, however, these can be found using the spatial analysis tools built
into ArcMap. The Elevations of the watersheds can
also be compared just using the original DEMs.
The
entire Iron County DEM was around 500 Megabytes (MB), which was difficult to
work with and slowed down Arc-Map considerably.
The function Extract by mask in the ArcToolbox
was used to cut the DEM to the size and shape of the watersheds. This made them
smaller and much more manageable since the clipped DEMs were only about 20 MB
each. The clipped DEMs can be seen below
in Figure 2.
Figure 2: DEM
Cut to Watershed Sizes
The
NHDPlus and USGS gage data were also much larger than
the area of the Coal Creek and Parowan Watersheds. The NHDPlus
data was obtained for the entire area of Region 16 and the USGS gage data was
obtained for the whole of the United States. These were clipped down to size
using the Clip function in the ArcToolbox. The hill
shade for the clipped DEMs along with the clipped NHDPlus
data can be seen in Figure 3.
Figure 3: Hill Shades and NHDflow Lines for Each Watershed
After
the DEMs and other data were clipped to a manageable size, they were prepared
to be pre-processed. The pre-processing was done using the ArcHydro
toolbar which can be downloaded from the ESRI website. First, the DEMs had to
be reconditioned using the NHD Flowlines for the two
watersheds. This process burns the streams into the DEM so that their
locations area shown properly throughout the rest of the process.
The
rest of the preprocessing prepared the data for the comparisons that were the
goal of this section of the report. The commands used to preprocess the data
are listed under the Terrain Preprocessing tab on the ArcHydro
toolbar. The order and description of each step are as follows:
1.
Fill
Sinks This removes low points in the DEM that would prevent water from
flowing downhill
2.
Flow
Direction This step computes the flow direction for each cell in the DEM.
This is why it is so important to fill the sinks first. Without filling the
sinks, there could be cells with no flow direction right in the middle of the
river.
3.
Flow
Accumulation This step calculates how many other cells will flow into any
given stream cell by counting all the stream cells flowing into the given cell.
In this way, it shows how large the river will be at any point.
4.
Stream
Definition This digitizes the streams in the watershed by using the flow
accumulation grid.
5.
Stream
Segmentation This groups the pixels from the stream definition into sections
with the same cell count.
6.
Catchment
Grid Delineation This delineates the catchments for each stream segment and
calculates the area of each catchment.
7.
Catchment
Polygon Processing This converts the catchment grids into polygons and
assigns a HydroID, GridID,
and NextDownID to each catchment.
8.
Drainage
Line Processing This converts the stream segments from raster format into
lines and assigns a HydroID and GridID.
It also calculates the points upstream and downstream of each stream segment.
The drainage lines were used to find the total stream length for each
watershed.
9.
Adjoint
Catchment Processing This calculates the adjoint
catchments from the catchments and drainage lines and assigns a HydroID and GridID. The
difference between the catchment and adjoint
catchment is that each adjoint catchment has its own
drainage point whereas catchments can share a drainage point.
10. Drainage Point Processing This
calculates the drainage points for each catchment and assigns a HydroID, GridID, and DrainID.
The
results of the above named steps can be seen in Figure 4. Notice that the thin
strip directed towards Cedar City was lost in the process. It was determined
that this lost area was negligible and made little difference in the results of
the comparison.
After
completing the preprocessing for each watershed, the watersheds were delineated
using the Batch Watershed Delineation tool in the Watershed Processing tab in
the ArcHydro toolbar. The watersheds shown along with
the USGS gages, NHD Points, and NHD Flowlines
delineated with the MAFLOWU (mean annual flow) attribute in Figure 5. Notice
that one of the catchments in the Parowan Watershed was actually outside the
watershed and changed the preliminary assumption of the watershed size.
Figure 4:
Preprocessing Results
Figure 5:
Watershed Delineation with Gages and NHD Flowlines
Also as part of the
comparison, the NHD Points shown in Figure 5 were also analyzed. It was found
in the attribute table that the points shown were spring seeps, which are the
probable sources of baseflow in each watershed. The
soils data obtained from the USBR Soil Data Mart can be seen in Figure 6. Due
to the fact that the data only covered about half of each watershed, it was
decided that the soils data could not be used for accurate comparison due to
the lack of data.
Figure
6: USBR Soils Data
Table 1 shows the
results of the preprocessing and data collected from .
Table 1
|
|
Coal Creek Watershed |
Parowan Watershed |
|
Area (from delineated watershed) |
83.0 mi2 |
58.9 mi2 |
|
Stream Length |
61.5 mi |
56.12 mi |
|
MAFLOWU |
2.56 cfs |
1.89 cfs |
|
Spring Count |
30 |
12 |
WATERSHED
COMPARISON DATA
Several
sources were used to gather the data used in comparing the two watersheds. Data
sources included the AGRC website to obtain the DEM, USGS NWIS website for streamflow data, NHDPlus website
for stream and watershed data, and USBR Soil Data Mart for soils data. The Snotel website was also searched, but no data was found for
the Coal Creek or Parowan Watersheds.
WATERSHED
COMPARISON RESULTS
After
all the data was gathered and manipulated, the two Coal Creek and Parowan
Watersheds could be compared. The results of these comparisons are shown in
Table 2 where the percentage was found by dividing the Parowan Watershed value
by the Coal Creek Watershed value. As stated earlier, the soils data was not
included in this comparison due to lack of data.
Table
2: Watershed Comparisons
|
Comparison |
Coal Creek Watershed |
Parowan Watershed |
Percentage |
|
Size (from delineated watershed) |
83.0 mi2 |
58.9 mi2 |
71.0 % |
|
Drainage Density |
0.741 mi/mi2 |
0.950 mi/mi2 |
128 % |
|
Mean Annual Flow |
2.56 cfs |
1.89 cfs |
73.8 % |
|
Spring Density |
0.361 springs/mi2 |
0.204 springs/mi2 |
56.5 % |
It should
also be noted that the Parowan Watershed is much denser than the Coal Creek
Watershed, but has a much lower spring density. This illustrates the fact that
springs most likely provide the baseflow for both of
these watersheds. By multiplying the drainage density and spring density
percentages, a 72.3 percent difference is obtained which is very close to the
mean annual flow. As a result of these comparisons, it was found that the two
watersheds are hydrologically similar. It is
suggested that a scaling factor of 75 percent be used to apply the Coal Creek
flood frequency curve to Parowan Creek. The study previously done by Bowen,
Collins, and Associates found that the 100 year flood for Coal Creek was 5,500 cfs. Using this flow and the scaling factor, the 100 year
flood for Parowan Creek can be projected at 4,125 cfs.
Flood Plain Delineation Objectives
With
the 100 year flood-rate obtained for Parowan Creek, the process of creating a 100
year flood plain can begin. The
delineation of the 100 year flood plain will be carried out for Coal Creek. The objective of this aspect of the project
will be to create a floodplain from the USGS gauging site then downstream of
the USGS gauging site for approximately 4.5 miles. The floodplain will be determined using the
following three programs: Arc-GIS to create a river profile and to draw the
floodplain, HEC-RAS will be used to model the river, and Hec-Geo
Ras will be used to interface between the two
programs.
Flood Plain Delineation Methodology
Before
beginning the process of creating a floodplain, an aerial photo of the area is
needed. The photo that was downloaded
was obtained from the AGRC website is referred to as a Mr. Sid county mosaic. It is an aerial photo in the form of a sid file that covers most of Iron
County. From this county mosaic Coal
Creek, its riparian area, and the Coal Creek watershed can be seen. The DEM also needs to be altered before
proceeding to create the floodplain. Using
the raster calculator, the DEM for Iron County was converted to measure
elevation in feet. This is because the
projected coordinate system that was used in this project measures distances in
feet.
The
next step is to create the layers that will be exported to Hec-Ras. The Hec-Geo Ras toolbar is used for this process. A personal database is created by the Hec-Geo Ras interface in which
layers for the river, river cross sections, flowlines,
and structures are created. Once these
layers are created and measured by Arc-Map, they can be exported to Hec-Ras. Hec-Ras is used to model the river and the
100 year flood.
Figure 7: Iron County Mosaic
The
river representing Coal Creek is drawn by hand, using the edit toolbar. The line must be drawn from upstream to
downstream, and must be one continuous line.
The river can be easily followed and navigated with the county mosaic in
place, as shown in Figures 8 and 9. The
figures 8 through 11 only show a small portion of the river, to make the layers
easier to see.
Figure 8: Area Without
Coal Creek Drawn Figure
9: Area With Coal Creek Drawn
With
the river layer drawn and stored in the database, the flowpaths layer can be
cut into the database. The flowpaths are
created by taking the river layer and using the edit command copy parallel. They are the layer which
represent the center line of the flow that will travel outside the river
banks.
Figure 10: Flowpaths Delineated
The
layer that is used to represent bridges and culverts is referred to as the bridge
layer. They are drawn by hand, like the
other layers. They are used in Hec-Ras to
show the exact location of the structures so they can be modeled correctly.
Figure 11: Bridges and Culvert Cross
Sections
The
final cross layer that is drawn in Arc-Map is the river cross sections, which
are labeled as XScutlines. These are the cross sections that represent
the river in Hec-Ras. They define the
shape and area for the river and the floodplain. They need to be created so they are large enough
to contain the 100 year flow rete. Once the cross sections are drawn some of
them are surveyed. This is to ensure
that the information that is input into the model will be as accurate as possible. However, not all the cross sections are
surveyed, only a portion of them.
Figure 12: River Cross Sections
With
all the layers drawn, the Hec-Geo Ras
tool can prepare the database to be sent to Hec-Ras with the needed
information. With the Iron County DEM
used for measuring elevations, and a projected coordinate system used for
measuring distance in the X and Y directions, stationing can be assigned for
each of the layers from a reference point.
The reference point, or zero station, in this situation is Rush Lake, where
Coal Creek ends. The river and flowpaths
are assigned stationing at every point of inflection along their path. The elevation associated with every one of
those stations is also assigned and populated in their respective attribute
tables. The bridge cross sections and
river cross sections have stationing assigned from left to right, relative to
the downstream direction of the river, at every point of inflection. The left side of the cross section is station zero,
and the station at the right side of the cross section is the length, in feet,
of the cross section. Then elevations
are assigned for each one of those stations.
These cross sections are also given a station in relation to the river.
This station is assigned based on the point where the cross section intersects
the river layer. Figure 13 shows the
toolbar command that assigns stationing and elevations for the cross sections,
and Figure 14 the station that is assigned with each cross section that is related
to the river.
Figure 13: Screenshot of the Hec-Geo Ras Toolbar
.
Figure 14: Cross Sections
Stationing
With
all the information assigned to each of the layers the information can be exported
out of Arc-GIS and then imported into Hec-Ras.
When the information is imported into Hec-Ras it comes in with all the stationing
and elevations that were assigned in by Arc-Map. Figure 15 shows the river and cross sections
as they appear in Hec-Ras.
Figure 15: Overview of the
Hec-Ras Model
Once
the information is imported into Hec-Ras the cross sections that have survey
data available are updated. The survey
data for the cross sections needs to be input by hand. The survey data for the structures is also
input by hand. Mannings friction values
are also input into the model, as are the boundary conditions. The
flow that was calculated as the 100 year flood rate earlier in this report is
also input into Hec-Ras. Once all the
values are input, the model is ready to run and create a water surface
profile. From the water surface profile,
water surface elevations and top widths are assigned for each cross
section. These top widths and water
surface elevations are then exported from Hec-Ras in a spatial data (sdf) format file.
Figure 16: Cross Section Data Within Hec-Ras
Once
the information is imported back into Arc-GIS, a new layer set is created with
the imported bridge and river cross sections, and their water surface
elevations. The water surface elevations
are read in on the DEM and to make sure that they area accurately placed. The top widths are also used to create the
flood plain that is imported from Hec-Ras.
The floodplain from Hec-Ras is called Bounding Polygons, as shown in Figure
17.
Figure 17: Hec-Ras Flood Plain
From
this Hec-Ras flood plain the actual flood plain can be drawn. Contours are created from the DEM and then
used to examine the topography of the surrounding area. Using the topography, cross sections top
widths, water surface elevations and engineering judgment, the final floodplain
is delineated.
Flood
Plain Delineation Results
The final
flood plain looked significantly different then the flood plain exported from
Hec-Ras. This is due to the topography
surrounding the river. The west side of
the river slopes downhill, so the cross sections that were cut were not able to
hold the flood. Therefore the flow was
assumed to leave the river and flow away from it. Hec-Ras does not have the capability to model
this scenario with any accuracy. Also,
due to this loss of water, the flow in the river needed to be reduced at the
cross section where it is lost. Hec-Ras
also has difficultly with this situation.
Both of these limitations of Hec-Ras account for the major differences
in the flood plain that Hec-Ras mapped and the final flood plain.
Interstate 15
Figure 18: Final Flood Plain
Interstate
15 is elevated approximately 8 feet from the surrounding area. This causes the Interstate to act as a dam for
the water that leaves the river. The flood water will not reach depths greater
than 3 or 4 feet so it cannot over top the Interstate. Most of the water flows back into the river
at the culvert that travels under Interstate 15, increasing the flow in the
river back to a value close to its original flow rate. Then later downstream, the river doesnt have
capacity to contain the flow, and the water spills out of the river until it
reaches the end of the floodplain study.
SUMMARY
AND CONCLUSIONS
As a
result of the watershed comparisons, it was determined that the Coal Creek and
Parowan Watersheds are hydrologically similar. From
this knowledge, a scaling factor of 75 percent is suggested to obtain a
conservative estimate for the 100 year flood for Parowan Creek.
The
floodplain was created from the 100 year flood rate determined from the watershed
comparison. The river did not have
capacity to hold this flow rate at a few of its cross sections, and because of
the topography, the water did not flow back into the river until it reached
Interstate 15. The Interstate functioned
as a dam for this scenario, and brought all the water back into the Coal
Creek. But downstream of Interstate 15,
the river did not have capacity to hold the 100 year flood rate, which causes
more flooding.
RECOMMENDATIONS
There
are only a few cross sections in Coal Creek that did not have capacity to hold
the 100 year flow rate. These cross
sections that cannot hold the flow are causing most of the flooding. One of the recommendations that would improve
the capacity of the river is to determine exactly how many cross sections are
causing the flooding problem and increase their capacities. If the few cross sections that do not have
capacity have their capacities increased the floodplain could be reduced, or
even eliminated for the 100 year flow rate.
References
Army
Corps of Engineers. http://www.hec.usace.army.mil/software/hec-ras/hecras-hecras.html
Automated
Geographic Reference Center (AGRC).
Utah GIS Portal. http://agrc.utah.gov/
National
Ocean and Atmospheric Administration (NOAA).
http://www.noaa.gov/
Natural
Resources Conservation Science (NRCS). http://www.wcc.nrcs.usda.gov/snow/
United
States Bureau of Reclamation (USBR). http://www.usbr.gov/
United
States Geologic Survey (USGS). http://www.usgs.gov/