Flood Plain Mapping on the Yellowstone
River
By: Clay Woods
CEE 6440, GIS
in Water Resources, 2008
Introduction
The
Yellowstone River
is the longest un-dammed river in the United
States and flows through most of Montana
and parts of Wyoming. It
eventually joins the Missouri River and drains most of Eastern
Montana.
Eastern
Montana is mostly rural, farming communities and farmers near the Yellowstone
River use its water for
irrigation. Eastern Montana receives a lot of
precipitation in the form of snow during the winter, which causes large amounts
of runoff in the spring. The summer weather in Eastern Montana
provides a potential for local flash flooding due to intense
thunderstorms. Many of the farming communities in Eastern
Montana are located along the Yellowstone
River, in order to utilize the
water. Problems with flooding exist
anywhere people develop communities along a river. One way of managing this flooding is to
develop flood maps that will predict which areas will be flooded during a flood
of a given magnitude.
GIS
can be used with HEC-RAS
using the tools available in HEC-geoRAS to develop a model of a river and predict the
floodplain of that river. This project
will develop a floodplain map for the city of Glendive,
Montana. This is a small city near where I grew up and
is located right on the Yellowstone River
in Eastern Montana.
A google map image of Glendive is shown in
figure 1. The flood map will be
generated using GIS and HEC-RAS.
Figure 1: View of Glendive,
MT.
Objectives
The
objectives of this project are to:
- Gather data for the Glendive
area including DEM, NHDPlus data, information
about past floods, and real time flow data for the Yellowstone
River.
- Develop a base map of the
study area in GIS using the data
collected
- Develop a HEC-RAS
model of the Yellowstone River
at Glendive using data imported from my GIS
model
- Export the results of the HEC-RAS
model into GIS to develop a layer
showing the floodplain for different flood conditions
Data
The
first part of this project was gathering data to use in the GIS
model. NHDPlus
data was downloaded for the region including Glendive. This dataset included flowlines,
a digital elevation model, basin and catchment
boundaries, and information from the USGS stations on the Yellowstone
River. An aerial photograph was obtained from the Montana
government website. This photo was used
to delineate the normal river from the flood plain. Real time river flow data was obtained from
the USGS website. Data for the Yellowstone
River only goes back to 2002, so
the project will use a small set of flow data.
The
NHDPlus data sets that were downloaded were very
large and covered a large region. This
data was trimmed to contain only Dawson
County, where Glendive is
located. Figure 2 shows a picture Dawson
County with flowlines
and where Glendive is located in the county.
The DEM for the region was also trimmed using the mask feature in Arcmap. The trimmed
DEM is shown in figure 3. The flowline information was not actually used very much for
this project. It was useful verify that
the satellite imagery and DEM matched with what the NHDPlus
flowlines were showing, but was not needed for the
floodplain analysis.
Figure 2: Data
trimmed to only Dawson County
Figure 3: DEM for Dawson
County
Software
The
software required for this project caused a major problem. The interface between Arcmap
and HEC-RAS
is called HEC-geoRAS
and the current version of this software is not compatible with version 9.3 of Arcmap. The
interface was designed to work with version 9.1 or Arcmap,
but version 9.2 also works based on the experience of the author. This created a problem of finding the older
version of Arcmap in order to continue with the
project. The HEC-RAS
program can be downloaded for free from the U.S. Army Corps of Engineers along
with the HEC-geoRAS
interface files. Each of these programs
also has a users manual that can be obtained from the
same website.
Analysis
using HEC-geoRas
The
purpose of HEC-geoRAS
is to assist in preparing data for import into HEC-RAS. HEC-RAS
then uses the geospacial data to calculate water
surface elevations for specified flowrates. The HEC-geoRAS program starts with a digital terrain model that can
be either a grid or a TIN. The downloaded DEM was used for this model,
so a grid format was selected.
The
next step in preparing the export file is to create RAS
layers that show the stream centerline, flowpaths,
banks, levees, obstructions, bridges, and storage areas. Some of these layers are optional, so this
project utilized the centerline, banks, and flowpaths. The new layers are created in a personal geodatabase and the lines are added using the editing
tool. The lines for stream centerline
and banks were developed using the aerial picture for this project, however
they can also be downloaded from existing files.
The
other required layer is cross sections.
The cross sections are where the terrain is exported into HEC-RAS
for the channel information. Care must be
taken to use enough cross sections and to ensure the cross sections are long
enough to go across the floodplain.
These cross sections can be previewed and edited if needed. Figure 4 shows the study area with the
created cross sections, banks, and centerlines.
To improve the speed of the model for this project only six cross
sections were used. The HEC-RAS
model will be more accurate if more cross sections are used.
Figure 4: Study
area to be exported
Processing in HEC-RAS
The information developed in HEC-geoRAS is imported into the geometry portion of HEC-RAS. This portion is what creates the river
channel with its associated floodplain in HEC-RAS. This imported geometry is shown in figure 5
and a sample of the imported cross sections is shown in figure 6. The cross sections can be edited after
importing if the banks that were drawn in HEC-geoRAS do not match up with the banks based on the DEM..
Figure 5: Imported
geometry for the study area
Figure 6: Imported
cross section
Flow data must be provided for HEC-RAS
to run. The flows for this project were
obtained from the USGS gaging station in
Glendive. The maximum recorded flow was
118,000 cfs and the average maximum flow during
spring runoff is 54,000 cfs. A large flow of double the maximum recorded
was also modeled for this project.
Manning’s n values must be given for both the river channel and overbank areas. For
this project the channel’s n value was assumed at 0.04 which is based on a bed
of sand to large gravel with some vegetation.
The overbank areas were assigned a value of
.055 based on small shrubs and agricultural land. The flow in the river was assumed to be subcritical and a slope of 0.0027 was used for the HEC-RAS
calculation of normal depth. This slope
was determined by getting elevation and distance information from the DEM and
calculating the slope.
The next step in the project was to run the HEC-RAS
model for the various flows. Information
generated by running the model can be useful without exporting the data back
into Arcmap.
Each cross section shows the water surface in relation to the cross
section elevation, so it can be seen where the banks are overtopping. A cross section is shown in Figure 7
representing the normal flows seen during runoff. Figure 8 shows the same cross section with
the maximum recorded flow selected to mode.
Figure 7: Water
surface at a cross section
Figure 8: Max flow cross section
Back to Arcmap
After the HEC-RAS
model has been run, the information is exported back to Arcmap
in order to create an inundation map of the study area. This is done by incorporating the water
surface levels developed in HEC-RAS
with the elevations of the DEM in Arcmap. This process creates an inundation polygon
that shows which land areas will be flooded by the flow modeled in HEC-RAS. The inundation for the maximum flow is shown
in figure 9.
Figure 9: Inundation for max recorded flood
Conclusions
Because of the terrain around Glendive, an example of which
can be seen in figure 7, floods of different sizes are not likely. The ground is very flat along the banks of
the river and then the terrain rises very quickly once out of the river
valley. As seen in the profiles in HEC-RAS,
once the water overtops the banks of the Yellowstone,
the flood has to get much worse before any more land area is inundated. The inundated land around Glendive is all
agricultural land with little development, so the impacts of flooding would be
minimal.
The quality of this flood map could be improved with more
cross sections modeled in HEC-RAS
and an investigation of actual land cover and river channel conditions in order
to more accurately input values of Manning’s n.
The bridges and low flow areas of the river could also have been modeled
in HEC-RAS, however the final results would be similar to the ones
received from a simpler mode. A higher
quality DEM would also improve the flood map by giving more accurate elevation
data. In spite of all of the
improvements that could be made to the models, the terrain around Glendive
basically controls any flooding and limits the flood to land mostly used for
agriculture.
References
and Data Sources
Montana
Geographic Information Clearing House.
http://mris.mt.gov/gis/
National Hydrography Dataset. http://www.horizon-systems.com/nhdplus/
Google
Earth. www.google.com
HEC-RAS website. www.hec.usace.army.mil/software/hec-ras/.