Workshop: Build a Basic HEC-HMS

Model from Scratch

This workshop is designed to help new users of HEC-HMS learn how to
apply the software. Not all the capabilities in HEC-HMS are demonstrated
in the workshop as the focus is on creating a working model and
calibrating the model to an observed flood event. This workshop presents
HEC-HMS model development, calibration, and uncertainty assessment.
You will start with an existing shapefiles that can be imported and used to
create the HEC-HMS basin model network. You will parameterize the
Basin Model, create a Meteorologic Model, and simulate a historic event.
You will calibrate the model to the historic event by manually adjusting
model parameters to improve model performance. Then you will evaluate
the effects of parameter uncertainty in computed flow by running an HEC-
HMS Uncertainty Analysis.

Overview
In this workshop you will:

• Create a basin model from existing shapefiles

• Review determining precipitation gage weights with HEC-GeoHMS
• Parameterize an HEC-HMS Basin Model
• Add streamflow and precipitation gage information to the project
• Create an HEC-HMS Meteorologic Model
• Create and compute an HEC-HMS Simulation Run
• Calibrate the model and evaluate model performance
• Create an uncertainty analysis and evaluate results

1
Background
The Punxsutawney Watershed (400 km2) is part of the Allegheny River
Basin located in western Pennsylvania, USA. Primary conveyance
streams include: Stump Creek, East Branch Mahoning Creek, and
Mahoning Creek. The confluence of Stump Creek and East Branch
Mahoning Creek is located east of the enclave of Big Run. Mahoning
Creek is downstream of the confluence. A map of the watershed is shown
in Figure 1.

Figure 1. Study Area: Mahoning Creek upstream of Punxsutawney, PA

2
A storm event in April 1994 produced high-runoff in the town of
Punxsutawney. Several regional precipitation gages and a discharge
gage in Punxsutawney captured the event. These gages will be used to
create a calibrated model of the event. The watershed will be modeled as
3 subbasins with incremental precipitation from recording rainfall gages;
user-specified gage weighting will be used. Figure 2 shows the subdivided
basin and nearby gages.

Figure 2. Precipitation and streamflow gages near Punxsutawney, PA

3
Tasks
1. Create a Basin Model from Existing Shapefiles
Basin geometry is defined by delineating subbasins and river reaches.
HEC-GeoHMS, an add-in to ArcGIS, can be used to create the initial
Basin Model from terrain information. Another option is to import GIS
features from shapefiles and then link the elements using junctions. This
task shows how to import shapefiles, which can be used to define HEC-
HMS subbasin and reach elements.
The subbasin delineation is shown in Figure 3.

Figure 3. Final subbasin delineation of Mahoning Creek

1.1. Double click the HEC-HMS icon to start the program.

1.2. The main program window will appear; notice the menu bar across
the top of the window with menus beginning with File and ending
with Help. Also part of the window are the tool bars directly
beneath the menu bar, the Watershed Explorer, the Component
Editor, Message Log, and the Desktop. To open an existing project,
click File and select Open….

4
1.3. The Open an Existing Project window will open; click Browse to
open the Select Project File window. Navigate to the project
directory and select the “HMS_Example.hms” file. Click the
Select button and the existing project will open.
1.4. Create a new Basin Model using the Basin Model Manager (select
the ComponentsBasin Model Manager menu option). Name
the basin model “MahoningatPunx”. Open the basin model by
selecting it in the Watershed Explorer. The basin model map will be
empty.
1.5. Go to the GIS menu and choose the Coordinate System menu
option. Within the Coordinate System editor, choose the browse
button and navigate to the … \HMS_Example\maps directory on
your computer. Select the Subbasin.prj file. The coordinate
system information should populate as shown in Figure 4.

Figure 4. Coordinate system defined for the MahoningatPunx basin model

1.6. Import GIS element by selecting the GISImport Georeferenced

Elements menu options. Choose the Subbasins element type and
click the Next button. Navigate to the … \HMS_Example\maps
directory and select the Subbasins.shp file. Click next. Choose
the “Name” attribute field to set the subbasin element names. Click
the finish button. You will see three subbasin elements in the basin
model map. You can interact with the elements by clicking on the
subbasin polygons or element icons.
1.7. Follow the previous steps and import the reach feature in the
Reach.shp file. Figure 5 shows the three subbasin elements and
the one reach element that were imported from the two shapefiles.
These elements are not connected to one another.

5
Figure 5. Subbasin and reach elements imported into the basin model

1.8. Add a junction element to the basin model map. Click on the
Junction Creation Tool toolbar button and then click in the map
around the upstream end of the Mahoning Creek reach element.
Name the junction “Big Run”.
1.9. Connect both Stump Creek and the EB Mahoning Creek subbasin
elements to the Big Run junction element. Then, connect the Big
Run junction element to the Mahoning Creek reach element.
Elements can be connected by right clicking on the element icon in
the map, selecting the Connect Downstream menu option, and
then clicking on the downstream element. A simple connection
should be shown in the map when elements are connected to one
another. The exception is when elements are connected to a reach
created by GIS features. You will not see the simple connection
from the Big Run junction to the Mahoning Creek reach element,
but the Big Junction component editor will show the connection.
1.10. Add a sink element and name it Punxsutawney. Connect the
Mahoning Creek Local subbasin element and the Mahoning Creek
reach element to the new sink element. Now, the elements should
be linked to one another in the basin model map. Right click on the
MahoningatPunx basin model name in the Watershed Explorer and
choose the Resort Elements Hydrologically menu option. The
elements in the Watershed Explorer should be ordered from
upstream to downstream, as shown in Figure 6.

6
 Figure 6. Basin model with georeferenced elements that are connected and
displayed in hydrologic order

2. Review: Determine Precipitation Gage Weights with HEC-

GeoHMS
Note: This task has been already been performed for you and is presented
for your information.
A GIS was used to estimate precipitation gage weights for computing
basin average precipitation. Figure 7 shows three precipitation gages in
the region and their location with respect to the watershed. Thiessen

7
polygons were created and then area weights computed based on
intersecting the Thiessen polygons with the subbasin polygons.

DUJP Gage

MFFP Gage

PNXP Gage

Figure 7. Thiessen polygons for Mahoning Creek gages

Results from the gage weight analysis are shown in Table 1.

Table 1. Precipitation gage weights using Thiessen polygons
Subbasin DUJP MFFP PNXP
EB Mahoning Creek 0.00 0.93 0.07
Mahoning Creek Local 0.02 0.00 0.98
Stump Creek 0.64 0.33 0.03

8
3. Parameterize the Basin Model
3.1. Expand the Basin Models node and select the Basin Model
“MahoningatPunx”. A map of the Basin Model should open as
shown in Figure 8. Notice the subbasin area has been populated
for each subbasin element, and modeling methods have been
selected – Loss Method: Initial and Constant, Transform Method:
Clark Unit Hydrograph, Baseflow Method: Recession, and Routing
Method: Lag.

Figure 8. MahoningatPunx Basin Model map

3.2. The initial baseflow was determined by evaluating the observed

baseflow at the beginning of the simulation and dividing by the total
watershed area. The recession constant and ratio to peak values
were determined from other regional models.
From the Parameters menu, select Baseflow  Recession to
open the Recession Baseflow global editor. Set Initial Type to
Discharge Per Area. Recession baseflow parameters are shown in
Table 2. Click the Apply and Close buttons when finished entering
the baseflow parameters.

9
 Table 2. Recession baseflow parameters
Initial Q Recession Ratio to
Subbasin
(m3/s/km2) Constant Peak
E B Mahoning Creek 0.05 0.80 0.3
Stump Creek 0.05 0.80 0.3
Mahoning Creek Local 0.05 0.80 0.3

3.3. A table relating the Hydrologic Soil Group and landuse type to the
constant loss rate was used to estimate the constant loss rate for
each of the subbasins. The dominant soil types are Soil Group B
and C and the dominant landuse in the watershed is
Pasture/Dryland and Woodland/Grass. The initial loss was
estimated based on the number of dry days prior to the storm
event. The impervious area parameter was estimated based on
GIS datasets.
Enter initial and constant loss parameters for each subbasin. From
the Parameters menu select Loss  Initial Constant Loss to
open the Initial Constant Loss global editor. Initial and constant
loss rate parameters are shown in Table 3.
Table 3. Initial and constant loss rate parameters

Initial Loss Constant

Impervious
Subbasin Loss Rate
(mm) Area (%)
(mm/hr)
E B Mahoning Creek 8 2.5 3.0
Stump Creek 8 3 3.0
Mahoning Creek Local 8 2.8 5.0

3.4. The time of concentration was estimated using the TR-55 method
where the travel time along the longest flowpath was estimated
using slope, land use characteristics, and channel geometry. The
storage coefficient was estimated using regional regression
information between the time of concentration and the storage
coefficient. The simple relationship is storage coefficient = 1.6 X
time of concentration.
Enter Clark unit hydrograph parameters for each subbasin. From
the Parameters menu select Transform  Clark Unit

10
 Hydrograph to open the Clark Transform global editor. Clark
transform parameters are shown in Table 4.
Table 4. Clark unit hydrograph parameters
Time of Concentration Storage
Subbasin
(hr) Coefficient (hr)
E B Mahoning Creek 3.7 5.9
Stump Creek 5.3 8.5
Mahoning Creek Local 6.0 9.6

3.5. The routing reach, Mahoning Creek, is used to route flow from the
Big Run junction to the Punxsutawney junction. The Lag routing
method was selected by default. Change the routing method to the
Muskingum routing method. Select the Mahoning Creek reach
element in the basin model map to open the reach’s Component
Editor. Change the Routing Method from Lag to Muskingum.
The Muskingum K was estimated by computing a typical velocity for
bank full flow and then using the reach length. The Muskingum K
was determined to be 4 hours. The Muskingum X and number of
Subreaches are parameters best determined during model
calibration. Set the Muskingum X to 0.25 and number of
Subreaches to 4.
Within the reach element’s Component Editor, select the Routing
tab. Enter a Muskingum K of 4 hours, a Muskingum X of 0.25, and
number of Subreaches of 4.
You have now completed the basin model for this project. The next step is
to add observed flow and precipitation gage information.

4. Add Streamflow and Precipitation Gage Information

4.1. Create a new precipitation gage. From the Components menu
select Time-Series Data Manager. In the manager window, press
the New… button to create a gage (make sure the Precipitation
Gages data type is selected); the window for creating a gage will
open.
4.2. In the new gage window, change the default name to DUJP. Press
the Create button to create the new gage; it will automatically be
added to the Watershed Explorer. You can leave the manager
window open since it will be used again shortly.
4.3. In the Watershed Explorer, browse to the gage you just created. In
the Component Editor set the Data Source to Single Record
HEC-DSS.

11
4.4. Select the correct external data source. You can click on the select
button next to the filename field to navigate to the file. Browse
to …\HMS_Example\data\observe.dss. It is good practice to
store all external DSS data within the project directory. That way,
the information is contained in the project and will be included
whenever the model is handed off to others for application or
review.
4.5. To select the correct pathname, click the Select DSS Pathname
button, . You can use the Search by Parts filters near the top of
the screen to find pathnames. Select the pathname with B-Part
“DUJP” and C-Part “PRECIP-INC” (Figure 9). Click Set Pathname.

Figure 9. Selecting a the "DUJP" precipitation record

12
4.6. Change the default time window to inspect some of the data. In the
Watershed Explorer, click on the time window under the “DUJP”
gage icon. In the Component Editor, change start date to
10Apr1994, the start time to 00:00, the end date to 15Apr1994,
and the end time to 00:00. Click on the Table and Graph tabs in
the Component Editor to see the data, as shown in Figure 10.

Figure 10. Viewing data in the component editor

13
You have finished setting up the “DUJP” time-series precipitation gage.
We will use the gage later by referring to it by name.
Repeat steps 4.1 to 4.6 to create the “PNXP” and “MFFP”
precipitation gages.
All of the precipitation data is now ready to use.
4.7. Add observed flow gage data. In the Time-Series Data Manager,
change the data type to Discharge Gage and add a new gage.
Name the gage Punxsutawney Observed Flow and link it to the
flow record at Punxsutawney in the observe.dss file (B-Part =
PNXP, C-Part=FLOW). Change the time window for the discharge
gage and view the flood hydrograph for the April 1994 event, as
shown in Figure 11.

Figure 11. Flood hydrograph for April 1994 event

4.8. Reference the observed flow gage in the basin model to compare
computed and observed results for the April 1994 simulation. In the
Basin Model map, click on the junction element “Punxsutawney” to

14
 open the Component Editor. Select the Options tab, click the
dropdown for Observed Flow, and select “Punxsutawney Observed
Flow.”

5. Create a Meteorologic Model

5.1. From the Components menu, select the Meteorologic Model
Manager; the Meteorologic Model Manager will open. Press the
New… button to create a new meteorologic model.
5.2. Change the default name to GageWeights and press the Create
button to create the new meteorologic model.
5.3. In the Watershed Explorer, browse to the new meteorologic model.
Click on the model “GageWeights”. In the Component Editor, set
the Precipitation method to Gage Weights. The evapotranspiration
and snowmelt methods should be turned off, the unit system should
be set to metric, and the Replace Missing option should be Set to
Default.
5.4. Connect the meteorologic model to subbasins in the basin model.
In the Component Editor, open the Basins tab. For basin model
“MahoningatPunx,” set Include Subbasins to Yes.
The Gage Weights precipitation method requires parameters for each
subbasin element in the basin model. The gage weights presented in
Table 1 will be used in the Meteorologic Model for computing the total
subbasin average precipitation. Only one gage will be used to define the
time pattern.
5.5. In the Meteorologic Model, expand the node for “EB Mahoning
Creek.” A Gage Weights node will display (Figure 12).

15
 Figure 12. Gage Weights node selected

5.6. Select the Gage Weights node. The Component Editor will have
two tabs: Selections and Weights. On the Selections tab, for each
gage “DUJP,” “MFFP,” and “PNXP” change the Use Gage option to
Yes.
5.7. Select the Weights tab. Enter the Depth Weight for each
precipitation gage calculated using Thiessen polygons, shown in
Table 1. The depth weight can be entered as either percentage in
decimal format (Figure 13) or as an area.

16
 Figure 13. Depth and time weights specified for the EB Mahoning Creek
subbasin

5.8. For subbasin “EB Mahoning Creek,” enter a Time Weight of 1.0 for
the “MFFP” gage and 0.0 for the “PNXP” and “MFFP” gages. Here,
the temporal distribution of the resulting hyetograph is solely
governed by the temporal precipitation distribution of the “MFFP”
gage.
Repeat steps 5.5-5.8 for subbasins “Mahoning Creek Local” and “Stump
Creek.” For “Mahoning Creek Local,” set the Time Weight to 1 for the
“PNXP” gage and to 0.0 for the “MFFP” and “DUJP” gages. For “Stump
Creek,” set the Time Weight to 1 for the “DUJP” gage and 0.0 for the
“PNXP” and “MFFP” gages.
You have now completed the User-Specified Gage Weights Meteorologic
Model.

6. Create and Run a Simulation

In order to create a Simulation Run, a Basin Model, a Meteorologic Model,
and Control Specifications need to be defined in the project.
6.1. Create a new Control Specifications. From the Components menu
select Control Specifications Manager; the manager window will
open. Press the New… button to create a new control
specifications.

17
6.2. Change the default name to 1994 Event. Press the Create button
to create the new Control Specifications.
6.3. In the Watershed Explorer, browse to the new Control
Specifications. Enter a start date and time as 10Apr1994 at 00:00.
Enter end date and time as 15Apr1994 at 00:00. Select 1 Hour for
the time interval, as shown in Figure 14.

Figure 14. Control Specifications entered

The Basin Model, Meteorologic Model, and Control Specifications are

complete. Now you will create and compute a Simulation Run.
6.4. From the Compute menu, select Create Compute  Simulation
Run. A wizard window will open for creating a new simulation run.
In Step 1, enter Simulation Run name: Run 1994 Flood Event. In
Step 2, choose the “MahoningatPunx” Basin Model. In Step 3,
choose the “GageWeights” Meteorologic Model. In Step 4, choose
the “1994 Event” Control Specifications. Press the Finish button to
complete the process of creating a Simulation Run.
6.5. Compute the Simulation Run. A Simulation Run must be selected
before it can be computed. The Compute toolbar includes a
selection list that shows all of the simulation runs that have been
created in the project. Click on the selection list and choose “Run:
Run 1994 Flood Event.” Once a Simulation Run has been
selected, click the Compute button immediately to the right of
the selection list to perform the compute, as shown in Figure 15.
The Compute button is available on the tool bar when a Simulation
Run is selected. Alternately, from the Compute menu select the
Compute Run (the name of the selected run will be shown within
brackets next to the Compute Run menu option).

18
 Figure 15. Select simulation and compute from the HEC-HMS tool bar

6.6. A Compute Progress window will open to show the advancement of

the simulation. The simulation may abort if errors are encountered.
If this happens, read the messages and fix any problems; then
compute the Simulation Run again. Close the progress window
when the run computes successfully.

19
Results are now available from the completed Simulation Run.
6.7. From the Results menu select the Global Summary Table. This
Global Summary Table will display as shown in Figure 16.
Alternately you can click the Global Summary Table button on
the Compute tool bar.

Figure 16. Global summary results for Run 1994 Flood Event simulation

6.8. View results for each Basin Model element. Right-click on an

element in the Basin Map; a context menu is displayed with several
choices including View Results. From the View Results menu there
are options to view graph, summary table or time-series table. The
same results are available from the Results tab of the Watershed
Explorer.

20
6.9. View the observed vs. computed hydrograph by selecting the
Graph option for the junction “Punxsutawney” (Figure 17). The
computed hydrograph shape is similar to the observed hydrograph
shape; however, the peak flow and runoff volume are low and the
timing of the peak flows do not coincide. At this point, model
parameters have only been estimated using GIS datasets and
regional information. The model must be calibrated for computed
results to approximate observed flow.

Figure 17. Observed vs. computed hydrograph at the Punxsutawney

junction

21
7. Calibrate the Model and Evaluate Model Performance
Initial parameter values will be refined through several calibration runs.
7.1. First, loss parameters can be adjusted to increase the runoff
volume. The summary results for the junction “Punxsutawney”
show the observed runoff volume is 68.42 mm and the computed
runoff volume is 48.01 mm. Decreasing the initial loss rate might
improve the results on April 10 and 11. Select the Parameters 
Loss  Initial and Constant Loss menu options to open the Initial
and Constant Loss global parameter table. Reduce the Initial Loss
and re-run the simulation. You will notice the computed flow
hydrograph is not sensitive to the initial loss rate, even when setting
the initial loss to 0 mm. This is likely due to the precipitation
intensity and the constant loss rate values.
7.2. In addition to Initial Loss reduce the Constant Loss Rate to increase
the computed runoff volume and increase the computed peak
flows. The Initial Constant Loss global parameter table contains
special editors to adjust all selected cells. Highlight all three
Constant Rate values, right click and choose Fill…. In the Table
Fill Options editor, choose the Multiply by constant: option and
enter 0.75, as shown in Figure 18. Re-compute the simulation.

Figure 18. Editing initial and constant loss rate values

22
Notice the computed peak flow exceeds the observed, but the timing of
the computed peak flow is early when compared to the observed
hydrograph, as shown in Figure 19.

Figure 19. Comparison of simulated and observed hydrographs after

editing initial and constant loss rates

7.3. The Clark unit hydrograph parameters control the timing and
attenuation of runoff from subbasin elements. Notice in Figure 19
that a majority of the runoff at the junction “Punxsutawney” is due to
runoff from the subbasin “Mahoning Creek Local” (the plot at this
junctions shows all upstream contributions plus the total inflow).
Before adjusting the UH parameters for the subbasins “Stump
Creek” and “EB Mahoning Creek,” adjust the Clark unit hydrograph
parameters for the subbasin “Mahoning Creek Local” only. Select
this subbasin in the Basin Map to open the Component Editor. Go
to the Transform tab and change the Time of Concentration from 6
hours to 9 hours and the Storage Coefficient from 9.6 to 14.4
hours (maintaining the relationship R = 1.6 * Tc). Re-compute the
simulation and compare results at the junction “Punxsutawney.”

23
 Figure 20. Comparison of simulated and observed hydrographs after
editing the Clark unit hydrograph parameters for the subbasin “Mahoning
Creek Local”

7.4 As show in Figure 20, the timing of the runoff hydrograph is greatly
improved; however, the peak flow and volume need some
additional adjustment. Within the Initial Constant Loss Global
Parameter Table, reduce the constant loss parameters by
multiplying values by 0.4 and set the initial loss to 6 mm for all
subbasins. Re-compute the simulation and compare results at the
junction “Punxsutawney.” Figure 21 shows the computed results
and observed flow after the above changes were made to the
model. The computed volume, 62.13 mm, is much closer to the
observed volume, 68.42 mm, and the timing and magnitude of the
peak flow is much closer as well. The Nash-Sutcliffe metric is
0.949 for this simulation. The summary table for the junction
“Punxsutawney” is shown in Figure 22.

24
 Figure 21. Comparison of simulated and observed hydrographs after
editing the initial and constant loss parameters a second time

Figure 22. Summary table after model calibration is complete

The final loss and transform parameter values are presented in Table 5
and Table 6. Slight adjustments were required from the initial parameter
estimates. Additional modification could be made to the recession
baseflow parameters to improve the model fit to the baseflow portion of
the runoff hydrograph. For example, increasing the Recession Constant
to 0.9 increases the computed runoff volume to 65.50 mm and increases
the Nash-Sutcliffe metric to 0.971. The model should be calibrated to
additional flood events and then a final set of parameter values
determined. Additionally the model should be validated against multiple
flood events.

25
One thing to note is that boundary condition information should be
evaluated when attempting to calibrate a model. The loss rates required
for the April 1994 event that resulted in the best fit of modeled results are
extremely low. Comparison of precipitation gage information from the
DUJP, PNXP, and MFFP gages showed 64.77 mm, 83.82 mm, and 27.43
mm, respectively. This variability in measured precipitation shows the
precipitation was not homogeneously distributed over the gages and
watershed. It’s likely that the low constant loss rates are due to
inadequately modeling precipitation over the entire watershed. The
meteorologic model could be updated and more depth weight applied to
the PNXP precipitation gage. Applying a larger depth weight would result
in higher basin average precipitation, which would result in loss rates that
are more reasonable.
Table 5. Final initial and constant loss rate parameters
Initial Loss Constant Loss Impervious
Subbasin
(mm) Rate (mm/hr) Area (%)
E B Mahoning Creek 6 0.75 3.0
Stump Creek 6 0.90 3.0
Mahoning Creek Local 6 0.84 5.0

Table 6. Final Clark unit hydrograph parameters

Time of Concentration Storage Coefficient
Subbasin
(hr) (hr)
E B Mahoning Creek 3.7 5.9
Stump Creek 5.3 8.5
Mahoning Creek Local 9.0 14.4

26
8. Create an Uncertainty Analysis and Evaluate Results
The Uncertainty Analysis compute option can be used to evaluate
uncertainty in model parameters on the computed results. The uncertainty
analysis compute option allows the user to enter uncertainty distributions
for basin model parameters, and then define the number of samples to
perform. The uncertainty analysis does not vary the boundary conditions,
only the selected basin model parameters. In this workshop, we will only
explore the uncertainty in the loss rate parameters. Once the simulation is
complete, we will review the range of output results available from the
uncertainty analysis.
8.1. Create an uncertainty analysis. From the Compute menu select
Uncertainty Analysis Manager. Name the analysis “Analysis April
1994”. Select Basin Model “MahoningatPunx” and the
Meteorologic Model “GageWeights”.
8.2. On the Compute tab in the Watershed Explorer, browse to and
select the new Uncertainty Analysis. Finish entering the
configuration information with the following time specification data.
See Figure 23.
Start Date: 10Apr1994
Start Time: 00:00
End Date: 15Apr1994
End Time: 00:00
Time Interval: 1 Hour
Total Samples: 100

Figure 23. Completed Uncertainty Analysis Component Editor

27
8.3. Next you will add parameters to the Uncertainty Analysis. Right-
click on the Uncertainty Analysis and select Add Parameter. Next,
under the Uncertainty Analysis, select “Parameter 1” to edit the
parameter in the Component Editor. In the Component Editor
select Element: “EB Mahoning Creek” and Parameter: “Initial and
Constant – Constant Rate,” as shown in Figure 24. Select a
Triangular distribution, and enter a minimum value of 0.5 mm/hr,
and a maximum value of 4.5 mm/hr. Distribution parameters
should be 0.5 mm/hr for the lower, 2.5 mm/hr for the mode, and
4.5 mm/hr for the upper.

Figure 24. Configuring the first uncertainty parameter

8.4. Add two additional parameters to the Uncertainty Analysis (Right-

click “Analysis April 1994” and select Add Parameter). In the
Component Editor set the uncertainty parameters to Constant
Rate for the subbasins “Stump Creek” and “Mahoning Creek
Local.” These two locations will be linked to the subbasin “E B
Mahoning Creek” so that constant loss rates are adjusted in a
similar manner; all values increase or decrease within a sample. In
the Component Editor, set the sampling Method to Regression
With Additive Error. Then set the Regression Element to “EB
Mahoning Creek,” and set the Regression Parameter to Initial and
Constant – Constant Rate. A simple linear regression can be
defined where the slope is 1 and the intercept is 0; effectively
setting the parameter value to be equal to the related parameter
value. The error term is represented by an uncertainty distribution.
As shown in Figure 25, select the Normal uncertainty distribution
and enter a mean of 0 mm/hr, and a standard deviation of 0.2
mm/hr (do this for both subbasins Stump Creek and Mahoning
Creek Local). These parameter values should result in constant
loss rate parameters that are within 0.2 mm/hr for about 68 percent
of the sampled values.

28
 Figure 25. Configuring the Stump Creek constant loss rate parameter
distribution

When complete, there should be 3 parameters selected in the uncertainty

analysis.
8.5. The final step to preparing the Uncertainty Analysis for simulation is
to configure the output results. In the Watershed Explorer to right-
click on the Uncertainty Analysis “Analysis April 1994” and select
Results…. Next, press Select…. Select the element
“Punxsutawney” and time-series Outflow as shown in Figure 26.
Close all of the results configuration windows.

Figure 26. Choosing the Outflow time-series for the Punxsutawney

element

29
8.6. Compute the Uncertainty Analysis - in the Watershed Explorer
right-click “Analysis April 1994” and select Compute. You will
notice the compute dialog and message window shows progress
for the 100 sample simulation. After the simulation is completed,
switch to the Results tab of the Watershed Explorer and review the
available results. You should see three Parameter tables that show
the samples of constant loss rate values for each subbasin
element. Figure 27 shows the sampled constant loss rate values for
the “EB Mahoning Creek” subbasin. The sampled values can be
copied to another program to create a histogram of values as
shown in Figure 28.

Figure 27. Sampled constant loss rate values for the subbasin “EB
Mahoning Creek”

30
Figure 28. Histogram of constant loss rate values

There is a node for Outflow results at the junction “Punxsutawney.”

Results include hydrographs (Figure 29), and tables of the
Maximum Outflow and Outflow Volume. Open the tables to view
results.

31
Figure 29. Hydrograph results for the Punxsutawney junction element

32