Introduction

This workshop develops the Probable Maximum Flood (PMF) using a simple HEC-HMS model of the Foster Joseph Sayers Dam watershed in Centre County, Pennsylvania, along with information from the Sayers Dam reservoir regulation manual and historic event gage data. The reservoir regulation manual for Sayers dam includes the original Spillway Design Flood (SDF), but this design event did not use the most recent Hydrometeorological Report (HMR 52) to develop the Probable Maximum Precipitation (PMP) and it is not clear how the flood hydrograph was computed. Consequently, use of the SDF does not allow the modeler to complete a sensitivity analysis of the hydrologic processes and assumptions associated with the reservoir Inflow Design Flood (IDF). Therefore, reuse of the SDF contained in the reservoir regulation manual is not appropriate.

Ideally, the PMF event would be developed using a precipitation-runoff model calibrated to the historic event with the highest peak magnitude of record and driven with the most recent Hydrometeorological PMP depths. However, given the limited time required to complete a Periodic Assessment, only a simple precipitation-runoff model of the watershed can be completed for this study. Adequate time for model calibration and validation are not available and many conservative assumptions will need to be made. Conservative assumptions are appropriate during the Periodic Assessment given the high level of uncertainty regarding many of the components in the hydrologic analysis. If a dam was found to be hydrologically deficient during the Periodic Assessment, a more detailed study would follow where conservative assumptions are no longer necessary or required due to additional analysis and data. All assumptions made when estimating the PMF should be documented to aid in scoping any future analysis if warranted.    

The Sayers Dam reservoir regulation manual only provided a hydrograph for the 1972 flood event, and no unit hydrograph for the watershed is available. The 1972 flood event is the largest flood in recent memory and the starting point for this workshop. In this workshop, the watershed upstream of Sayers Dam will be treated as a single watershed. This approach should be adequate for a Periodic Assessment and could be applied to many dams with a relatively small upstream area (less than 1,000 square miles) that do not include upstream water control structures.

In addition to using the 1972 flood event from the reservoir regulation manual, several other historic events observed at a U.S. Geological Survey (USGS) gage will be used to develop unit hydrographs. The unit hydrographs developed for these historic events will be compared and guidance will be provided on how to determine the most conservative unit hydrograph to use in the PMF analysis.

Goals

  1. Use historic flows at a USGS gage and existing information in a reservoir regulation manual to develop the reservoir Inflow Design Flood (IDF). 
  2. Develop a simple precipitation-runoff HEC-HMS model where the watershed area above the dam is modeled as a single subbasin.
  3. Use the S-graph method to develop a basin unit hydrograph.
  4. Use a synthetic method (Clark Unit Hydrograph) to develop a basin unit hydrograph. 

Data

Foster Joseph Sayers Dam is located on Bald Eagle Creek in Centre County, PA. Sayers dam is about 14 miles upstream of the confluence of Bald Eagle Creek and the West Branch Susquehanna River at Lock Haven, PA. The drainage area upstream of Sayers dam is approximately 339 square miles, while the drainage area at the confluence with the West Branch Susquehanna is approximately 781 square miles. The primary purpose of Sayers Dam was to provide flood control for the downstream reach of Bald Eagle Creek and West Branch Susquehanna River. Sayers Dam was completed in 1969 and consists of a rolled earth-fill embankment about 6,835 feet long with a top width of 25 feet. The maximum height of the dam is 100 feet above the stream bed at elevation 683 feet.

Foster Joseph Sayers Dam - Location Map

Foster Joseph Sayers Dam - Location Map 2

Download the initial model files here - PMF_Workshop_A_Initial.zip

Steps

Note

Given the number of steps in this workshop, many have already been completed to save time. These steps are marked with a “Completed” designation at the beginning of the step. 

  1. Find Historic Events - Completed
    Find three historic events for the development of the watershed unit hydrograph. Based on information from the USGS website (https://waterdata.usgs.gov/nwis), there is one gage above Sayers Dam on Bald Eagle Creek. The USGS provides the latitude and longitude for the gage on the website and the below figure shows its location with the gray marker (access the National Water Information System Mapper at https://maps.waterdata.usgs.gov/mapper/index.html). The drainage area to the gage is 265 square miles and the drainage area to the dam is 339 square miles (28% larger) based on data on the Bald Eagle Creek at Blanchard, PA stream gage just below the dam. The USGS gage has 60 years of peak flows with the 1972, 2004, and 1996 ranking as the largest on record as seen in the second figure. These are the three events used in this workshop.
    Bald Eagle Creek at Blanchard - Location Map
    Bald Eagle Creek below Spring Creek Gage - Annual Peak Streamflow
  2. Find Hydrograph and Hyetograph Data - Completed
    1. Find hydrograph and hyetograph information for the three selected events. The events used to determine unit hydrographs will depend on the data (both streamflow and precipitation) available. Data for the 1972 event is provided in the reservoir regulation manual. Data for the 2004 and 1996 events will be gathered from the USGS and the USACE Corps Water Management System (CWMS) database.
    2. You should do a quick check of the quality of your data. Hourly streamflow data were available for the September 2004 and January 1996 events from the CWMS database. However, the peak of the 2004 event in CWMS (23,267 cfs) did not match the peak reported by the USGS (21,100 cfs) due to the use of an older rating curve in CWMS. To correct for this, all the flows for the 2004 event were scaled down so the peak flow equaled the USGS record. In reality, the problem with the rating curve is likely an issue with the higher flows only, but this example modifies all the flows for simplicity.
    3. As noted earlier, the gage is upstream of the dam so the flows at the gage need to be scaled to better represent flows at the dam.  For a PA level study and for this example, a scaling factor was developed from the 1972 peak flow at the upstream gage and at the dam and then was applied to all flows at the gage.
    4. Observed precipitation information is required to determine the duration of the unit hydrograph. Find precipitation data for the 2004 and 1996 events on the National Climatic Data Center (NCDC) website (access a map viewer with hourly precipitation gages at https://www.ncei.noaa.gov/maps/hourly/). In the case of the Sayer Dam site there are several precipitation gages upstream of the dam with hourly precipitation data for the dates needed. Given the size of this watershed and the amount of data available, a detailed analysis would incorporate precipitation from several gages. For this example and a Periodic Assessment level of study, sample rainfall depths from a handful of gages. Three gages that appear to fall within the watershed boundary are Tyrone, PA; Philipsburg 8 E, PA; and Hollidaysburg 2 NW, PA.
    5. Finding precipitation gages that recorded adequate precipitation depths and a representative time pattern can be challenging. Refer to EM 1110-2-1417 for a detailed discussion about developing a calibrated model. For a Periodic Assessment level of study, verification that the precipitation gage recorded more than 80% of the runoff volume and that the time pattern supports the timing of the runoff hydrograph should be adequate.
    6. Determination of initial loss and constant loss values should not be a time-consuming exercise. Initial losses can be based qualitatively on the duration since the previous precipitation event, and time of the onset of precipitation and the onset of the rising limb of the runoff hydrograph. Constant loss rates can be based on historic studies.
    7. If acceptable flow and precipitation data had not been available for both the 2004 and 1996 events of this workshop, then another historic event could be selected with guidance from the ranking of the peak flows at the gage.
  3. Remove Hydrograph Baseflow - Completed
    Remove the baseflow from the historic event hydrographs. Baseflow can be removed using several different methods and the choice of method is generally negligible if the hydrograph is a historic event. One simple graphical method for separating baseflow from an observed hydrograph is called the straight-line method. With this method, a line is drawn from the point on the rising limb where direct runoff begins and the point on the receding limb where normal baseflow resumes. In the case of the 1972 event, assume that the surface runoff begins on 6/22/1972 at 6:00 AM and that the normal baseflow resumes (surface runoff ends) at 6/25/1972 at 6:00 AM. A straight line is drawn between these two points to separate baseflow (area underneath the line) from direct runoff (area above the line). 
  4. Scale the Observed Flow to Produce a Unit Hydrograph - Completed
    1. Scale the observed flood hydrographs (without baseflow) so the volume under each of the hydrographs is equivalent to 1 inch. Do this by determining the total runoff volume under the flood hydrograph in inches and dividing all of its ordinates by this volume. In the case of the 1972 event, the volume of direct runoff under the hydrograph is 2.84 inches. This means all the flows of the 1972 flood are 2.84 times larger than those of the unit hydrograph. Dividing all the ordinates by 2.84 reduces the flows to those that have 1 inch of volume under the resulting hydrograph. The volume under the resulting unit hydrograph should be checked to ensure it is 1 inch.
    2. The three figures below show the 1972, 2004, and 1996 flood hydrographs for Sayers Dam. Included in the figures are the precipitation hyetographs, baseflow hydrographs, direct runoff hydrographs, and the resulting unit hydrographs for the flood events. The spreadsheet used to compute the unit hydrographs is named Unit Hydrographs.xlsx and can be found in the data folder located within the HEC-HMS model project directory. You can open this spreadsheet to get an idea of how to manually compute the unit hydrograph for a flood event and check its volume.
      Hyetograph and Hydrographs - 1972 Event
      Hyetograph and Hydrographs - 2004 Event
      Hyetograph and Hydrographs - 1996 Event
  5. Determine Unit Hydrograph Durations - Completed
    The unit hydrograph duration is the duration of the effective rainfall that produced the hydrograph. Determining the effective rainfall is subjective. In this example, the effective rainfall was assumed to begin just before when the baseflow and observed flow lines diverge and end when the precipitation drops sharply back to lower values (the total depth of excess precipitation should be equal to the direct runoff volume). This results in a 20-hour unit hydrograph for the 1972 flood, a 10-hour unit hydrograph for the 2004 flood, and a 3-hour unit hydrograph for the 1996 flood.
  6. Standardize Unit Hydrographs to a 1-Hour Duration
    Use HEC-HMS to convert the unit hydrographs for the three events to a 1-hour duration. Duration conversions of unit hydrographs are calculated using the S-hydrograph method. This is the methodology used by HEC-HMS. Calculations for the 1996 event using the S-hydrograph method are shown in the Unit Hydrograph Duration Conversions.xlsx file located in the data folder of the HEC-HMS project directory. Conversions are not shown for the 1972 and 2004 events because the S-hydrograph method is not effective for these events due to the large difference between their observed durations (e.g. 20 and 10 hours) and the 1-hour duration. Do not use the S-graph method to convert unit hydrograph duration if going from a longer duration to a shorter duration with a reduction in duration more than 3 times (a 10-hour duration could be reduced to a 5-hour duration with some possible problems, but reducing the 10-hour duration unit hydrograph to a 1-hour duration could result in issues that make the 1-hour duration unit hydrograph unusable). The following sub-steps will illustrate this point. 

    Note

    Many of the preliminary sub-steps below were completed for you, such as creating a basin model, adding a subbasin element (for the drainage area at the dam), and loading a background map.


    Open the HEC-HMS model of the Bald Eagle Creek basin, the HEC-HMS project is named PMF Workshop A.
    1. Completed. Create a basin model named 1972 Bald Eagle Basin, make sure the unit system is "U.S. Customary" and the replace missing option is set to "Yes".
    2. Completed. Load a background shapefile of the watershed. This is for visual purposes only. 
    3. Completed. Add one subbasin element to represent the area upstream of Sayers Dam. Name the subbasin Bald Eagle Basin. Enter a drainage area of 338.6 square miles, set the loss method to “None”, sent the transform method to "User-Specified Unit Hydrograph", and the baseflow method to "None". The below figure shows the parameterized Component Editor for the subbasin.
       Bald Eagle Basin Component Editor
    4. Completed. Add the 1972 20-hour unit hydrograph to the model as a unit hydrograph curve by selecting Components | Paired Data Manager | Unit Hydrograph Curves. Name the curve 1972 20-Hr Unit Hydrograph.  Go to the Watershed Explorer on the left of the screen and expand Paired Data | Unit Hydrograph Curves and select 1972 20-Hr Unit Hydrograph to open the unit hydrograph curve input. Set the units to "CFS", the interval to "1 Hour", and the duration to "20 Hours". Enter the unit hydrograph developed earlier into the Table tab. Repeat this procedure for the 1996 unit hydrograph. 
    5. Completed. Assign the 1972 unit hydrograph to the Bald Eagle Basin element by navigating to the Watershed Explorer, expanding Basin Models, expanding 1972 Bald Eagle Basin, and clicking on the Bald Eagle Basin element. Select the Transform tab and choose 1972 20-Hr Unit Hydrograph from the dropdown. Choose to use "10" passes. This controls the smoothing of the falling limb of the unit hydrograph used if the duration of the unit hydrograph needed by the model is different than that entered as the User Specified unit hydrograph. In this case, the 1972 unit hydrograph is 20-hours and the model time step is 1-hour so the 20-hour unit hydrograph will be converted to a 1-hour unit hydrograph for use by the model. As shown below, the result will be poor when converting a longer duration unit hydrograph to a shorter during unit hydrograph when using the s-graph approach (the following steps are illustrating this point). 
      Setting the Transform Method for the Bald Eagle Basin Element
    6. Completed. Navigate to Components | Create Component | Time-Series Data. Create a precipitation gage in the Time-Series Data named 1-inch 1 hour; the units should be set to "Incremental Inches" and the time interval set to "1 Hour". The time window should be equivalent to the flow gage; set the start date and time to 01Jan3000 00:00 and the end time/date to 07Jan3000 03:00. Enter 1 inch of rainfall at the first hour as shown below.
      1-inch 1 hour Precipitation Gage Settings
    7. Completed. Navigate to Components | Create Component | Meteorologic Model. Name the meteorologic model 1-inch Precip 1 hour. Be sure the replace missing option is "Set to Default" and select "Specified Hyetograph" as the precipitation method. In the Watershed Explorer, select the 1-inch Precip 1 hour meteorologic model and click on the Basins tab. Link the meteorologic model to the 1972 Bald Eagle Basin basin model. Make sure the precipitation gage is chosen on the Specified Hyetograph tab, as shown in the below figure.
      Specified Hyetograph Component Editor
    8. Completed. Navigate to Components | Create Component | Control Specifications. Name the control specification Time Window and enter a start date and time of 01Jan3000 00:00 and end date and time of 10Jan3000 00:00. Select a 1 Hour time interval.
    9. Completed. Navigate to Compute | Create Compute | Simulation Run. Name the simulation Convert UH Duration (1972) and the select the 1972 Bald Eagle Basin basin model, 1-inch Precip 1 hour meteorologic model, and the Time Window control specifications. Be sure the output DSS File is saved to Convert_UH_Duration__1972_.dss (default DSS file for the simulation run). This will allow comparison of results to input values as demonstrated in the next two steps.
    10. Run the Convert UH Duration (1972) simulation by navigating to the Watershed Explorer, selecting the Compute tab, right-clicking on the Convert UH Duration (1972) simulation, and selecting Compute
    11. Open the output DSS file by navigating to the folder the model is saved in (...\PMF_Workshop_A_Initial\) and double clicking on the DSS file Convert_UH_Duration__1972_.dss (DSSVue will proceed to open the output DSS file). Use the pathname filter and set C to "FLOW-UNIT GRAPH". Select both the 1972 20-Hr Unit Hydrograph and the Bald Eagle Basin Unit Graph produced by the model and open a plot. Although a 20-hour unit hydrograph was input into the model, the program converted it to a 1-hour unit hydrograph because a 1-hour time step was used in the simulation. In the top figure below, you can see that the 1-hour unit hydrograph created is not realistic. When the same procedure is repeated for the 1996 3-hour unit hydrograph as seen in the second figure, the resulting 1-hour unit hydrograph is of better quality but still jagged. This illustrates the problem encountered when a unit hydrograph of a longer duration is converted to one of a smaller duration. The farther apart the durations (e.g. 20 hour to 1 hour), the less realistic the resulting unit hydrograph could become. If you do use a unit hydrograph that is a result of a longer duration precipitation excess event, then verify the unit hydrograph is reasonable when it is converted to a shorter duration (make sure the shape is reasonable and the volume of runoff is close to 1-inch). If found unreasonable, then use the method described below in step 7 where a synthetic unit hydrograph is fit to the historic unit hydrograph. There is no issue when changing the duration of synthetic unit hydrographs.
      Unit Hydrograph Comparison 1
      Unit Hydrograph Comparison 2
  7. Use Synthetic Methods to Model the PMF
    Now you will use synthetic unit hydrograph methods to reproduce the 1972 unit hydrograph.

    Note

    This synthetic method is preferred to using the user specified unit hydrograph method illustrated above in step 6 because the unit hydrograph produced by synthetic methods can be used with different model time steps and it avoids having to spend time smoothing a jagged unit hydrograph and checking its volume. Also, synthetic unit hydrographs allow you to peak the unit response at the dam.

    1. Continue working with the HEC-HMS model of the Bald Eagle Creek basin.
    2. Completed. Input the unit hydrograph computed in steps 1 through 4 as a time-series discharge gage in the HEC-HMS model. Navigate to Components | Create Component | Time-Series Data. Use a name of 1972 20-hour Unit Hydrograph and set the units to "Cubic Feet per Second" and time interval to "1 Hour". Define a start date of 01Jan3000 and start time of 00:00. The end date and time should be 10Jan3000 at 00:00. Copy and paste the unit hydrograph computed in steps 1 through 4 into the Table tab. The figure below shows the Graph tab for the unit hydrograph.
      1972 20-hour Unit Hydrograph Plot
    3. Completed. Create a basin model named 1972 Bald Eagle Calibration and make sure the unit system is "U.S. Customary" and the replace missing option is set to "Yes". 
    4. Completed. Load a background shapefile of the watershed. This is for visual purposes only. 
    5. Completed. Add one subbasin element to represent the area upstream of Sayers Dam. Name the subbasin Bald Eagle Basin. Enter a drainage area of 338.6 square miles, set the loss method to "None", set the transform method to "Clark Unit Hydrograph", and the baseflow method to "None". Under the Transform tab, set both the Time of Concentration (Tc) and Storage Coefficient (R) to 5 hours. The below figure shows the completed basin model. You can use general rules of thumb when setting the initial Tc and R parameter values. The time of concentration can be estimated using the TR-55 methodology and then the R value can arbitrarily be set equal to the Tc value. As discussed in later steps, the Tc and R values will be adjusted so that the synthetic unit hydrograph matches the historic unit hydrograph. 
      Bald Eagle Basin Component Editor and Map
    6. Completed. Open the Options tab for the Bald Eagle Creek subbasin element and link the Observed Flow to the 1972 20-Hour Unit Hydrograph discharge gage. The computed unit hydrograph will be treated as observed flow. Tc and R parameters will be adjusted manually to reproduce the unit hydrograph for the 1972 flood event. Using a synthetic unit hydrograph method, like the Clark Unit hydrograph model, is preferable to using a user-specified unit hydrograph since synthetic unit hydrographs are independent of the simulation time step.
    7. Completed. Create a Time-Series Data precipitation gage and name it 1-inch 20 hours. The Units should be set to "Incremental Inches" and the time interval set to "1 Hour". The time window should be equivalent to the flow gage; set the start date and time to 01Jan3000 00:00 and the end date and time to 07Jan3000 03:00. To recreate the unit hydrograph, one inch of precipitation must be distributed over the duration of the unit hydrograph. If 20 hours of excess precipitation generated the unit hydrograph for the 1972 event, then 1 inch of precipitation must be distributed over 20 hours. Therefore, enter 0.05 inches for each hour starting at 01Jan3000, 01:00 and ending at hour 20:00 on 01Jan3000. The below figure shows the precipitation hyetograph with 0.05 inches distributed over 20 hours.
       Precipitation Hyetograph
    8. Completed. Create a meteorologic model named 1-inch Precip 20 hours, make sure the replace missing option is set to "Set to Default" and select "Specified Hyetograph" as the precipitation method.  Click on the Basins tab, and link the meteorologic model to the basin model 1972 Bald Eagle Calibration. Make sure the precipitation gage is chosen on the Specified Hyetograph.
    9. Completed. Create a control specifications named Time Window and enter a start date and time of 01Jan3000 00:00 and end date and time of 10Jan3000 00:00. Make sure the time interval is set to "1 Hour".
    10. Completed. Create a simulation run named Calibrated 1972 Clark Model that combines the 1972 Bald Eagle Calibration basin model, 1-inch Precip 20 hours meteorologic model, and Time Window control specifications. 
    11. Run the Calibrated 1972 Clark Model simulation and view results for the Bald Eagle Creek subbasin element. As shown below, the initial Tc and R values (5 hours for both Tc and R) do a poor job recreating the unit hydrograph for the 1972 event.
      Initial Results at Bald Eagle Basin 
    12. Adjust the Tc and R parameters to reproduce the unit hydrograph for the 1972 flood event. 
      Question 1: What values of Tc and R did you find that best reproduced the 1972 unit hydrograph?

      After several iterations, a Tc value of 12.5 hrs and an R value of 11.5 hrs reproduced the 1972 unit hydrograph well. The element result graph for Bald Eagle Basin is shown below. The computed unit hydrograph represents the unit response from 1 inch of precipitation spread over 20 hours.

      Calibrated Results at Bald Eagle Basin

      Question 2: What Nash-Sutcliffe value did your calibrated model produce? Hint: HEC-HMS will compute several calibration statistics, including the Nash-Sutcliffe metric, and display them in the results "Summary Table" for an element.

      Ideally, your Nash-Sutcliffe value should be as close as possible to 1. A Tc value of 12.5 hrs and an R value of 11.5 hrs produced a Nash-Sutcliffe value of 0.967 as seen in the below summary table. 

      Bald Eagle Basin Results Summary Table

  8. Compare the 20-Hour and 1-Hour Unit Hydrographs
    1. Completed. Create a new precipitation gage named 1-inch 1 hour in the Watershed Explorer, Time-Series Data section (by creating a copy of the existing one) and enter a precipitation depth of 1 inch over 1 hour.
    2. Completed. Create a new meteorologic model named 1-inch Precip 1 hour. Go to the Basins tab and link the meteorologic model to the basin model if it is not already. Make sure the replace missing option is "Set to Default" and select "Specified Hyetograph" as the precipitation method. Select the Specified Hyetograph node to open the editor and link the 1-inch 1-hour precipitation gage to the Bald Eagle Basin subbasin element.
    3. Create a simulation named 1-inch 1 hour UH (1972) that combines the 1972 Bald Eagle Calibration basin model, 1-inch Precip 1 hour meteorologic model, and Time Window control specifications. Be sure you add the 1972 Bald Eagle Calibration Basin Model to the 1-inch Precip 1 hour Meteorologic Model by navigating to the Watershed Explorer, clicking the 1-inch Precip 1 hour meteorological model, and clicking the Basins tab. 
    4. Run the simulation and compare the results from the 1-hour precipitation duration to the 20-hour precipitation duration to see how the unit hydrograph changes with respect to precipitation duration. When using the Clark unit hydrograph model, unlike a user-specified unit hydrograph, the unit hydrograph duration is automatically updated given the simulation time-step.
      20-hour UH vs 1-hour UH Comparison
      The above figure compares the original 20-hour unit hydrograph (solid line) and the 1-hour unit hydrograph (dotted line). Both unit hydrographs have 1 unit of volume as dictated by unit hydrograph theory. Converting the unit hydrograph from 20 hours to 1 hour increased the peak flow and shifted it forward in time. The graph was developed by holding down the Ctrl key and selecting the Outflow graph in the results of both simulations.

      This workshop illustrates how to transform the unit hydrograph from one duration to another. Another method is the summation method (S-hydrograph) method. However, this method is not flexible when “peaking” the unit hydrograph. Therefore, if existing unit hydrographs are available from historic reports, it is recommended to use a synthetic unit hydrograph, preferably the Clark unit hydrograph model, as opposed to the historical unit hydrograph.  
  9. Determine the Most Conservative Unit Hydrograph
    The below figure compares the 1-hour unit hydrographs determined for all three events (1972, 2004, and 1996). A separate HEC-HMS basin model was developed for each event and calibrated to the unit hydrographs using the Clark UH method. For a PA level PMF analysis, it is best to select the most conservative unit hydrograph for modeling. The most conservative unit hydrograph is the one with the fastest timing and highest peak. Of the events considered, the 2004 unit hydrograph is the most conservative and will be used for the rest of this example. The Clark UH parameters for the 2004 event were Tc = 10 hours and R = 8 hours.

    Question 3: Why do you think the 2004 event resulted in the most conservative unit hydrograph?

    The 2004 event had the most concentrated and intense precipitation of the three events which resulted in a narrow unit hydrograph with a high peak. Thus, while it did not have the largest peak flow of the three events, it produced the 1-hour unit hydrograph with the highest peak. This comparison illustrates that it cannot be assumed that the historic event with the highest peak will produce the most conservative PMF results.

  10. Apply a Peaking Factor to the Unit Hydrograph
    Based on Engineering Regulation 1110-8-2(FR), Inflow Design Floods for Dams and Reservoirs, unit hydrograph inflows must be peaked to account for the non-linearity observed in real flood events. We will now peak the 2004 unit hydrograph for a PMF analysis.
    1. Make a copy of the 1972 Bald Eagle Calibration basin model and call it 2004 Bald Eagle Basin. Set the Clark Unit Hydrograph parameters in the Bald Eagle Basin element to Tc = 10 hours and R = 8 hours (these parameters were set based on calibration to the 2004 unit hydrograph). Go to the Options tab and set the Observed Flow to "None" since we are no longer calibrating. Create a new simulation named 1-inch 1 hour (2004). Select the correct basin model, meteorologic model, and control specification for the new simulation. Under Meteorologic Models in the Components tab include the 2004 Bald Eagle Basin basin model on the Basins tab of the 1-inch Precip 1 hour meteorologic model. Run the 1-inch 1 hour UH (2004) simulation.
    2. Create a copy of the 2004 Bald Eagle Basin basin model and name it 2004 Bald Eagle Basin 25P. This copy will be used to peak the unit response at the dam by 25 percent.
    3. Create a new simulation named 1-inch 1 hour UH (2004) 25P using the 2004 Bald Eagle Basin 25P basin model, 1-inch Precip 1 hour meteorologic model, and Time Window control specifications. Make sure to update the 1-inch Precip 1 hour meteorologic model so that it is linked to the Bald Eagle Basin 25P basin model (go to the Basins tab within the meteorologic model editor).
    4.  Run the 2004 Bald Eagle Basin 25P simulation and adjust the Tc and R parameters using the same scaling factor so that the unit hydrograph peak is scaled by 25 percent (i.e. unit hydrograph peaking). The Tc parameter controls the hydrograph timing and the R parameter controls the hydrograph shape (but also timing to some degree). Both parameters are influenced by similar watershed characteristics, slope, flow path, roughness, and storage to name a few. The peak flow of the 2004 event unit hydrograph is 16,003 cfs. You want the 25-percent peaked unit hydrograph to have a peak flow of about 20,004 cfs.

      Note

      You can easily compare hydrographs from two simulations by clicking on results time-series on the Results tab of the Watershed Explorer (hold down the control key to add multiple time series to the same plot). You can also compare peak flows by opening the Summary Table for both simulations.  

      Question 4: What are your final Tc and R parameters that peak the unit hydrograph by approximately 25 percent?

      7.92 and 6.34 hours respectively. The below figure shows the 2004 1-hour unit hydrograph (solid line) peaked by 25 percent (dashed line). Set your Clark Parameters in the 2004 Bald Eagle Basin 25P basin to Tc = 7.92 and R = 6.34 so your results match. During a PA, you would also peak the unit hydrograph by 50 percent. Multiple PMF simulations are required for the PA that evaluate scenarios of no peaking, 25 percent peaking, and 50 percent peaking. 

    5. Add losses and baseflow to both 2004 basin models (select the Initial and Constant loss method and the Recession baseflow method). Using information in the current Sayers Dam reservoir regulation manual, losses should be very low. The spillway design flood assumed 2.75 inches of losses (for a 48-hour event), which is approximated using a relatively small initial loss of 0.2 inches and a constant loss around 0.05 inches per hour. In the Loss tab, set the Initial Loss to 0.2 inches and the Constant Loss Rate to 0.05 inches per hours within the loss editor. The Impervious Area should be 1.6 percent (the reservoir area is 5.4 square miles when the reservoir elevation is at the spillway invert). If the drainage area of the lake is significant in relationship to the watershed area, then a higher percent impervious area scenario could be included (to reflect the surface area in between the starting pool and full pool). In the Baseflow tab, set the Initial Type to Discharge, Initial Discharge to 300 cfs, Recession Constant to 0.6, and the Ratio to Peak to 0.05.

      Note

      It is important that baseflow parameters are defined so that baseflow is a very minor portion of the total runoff hydrograph. Since losses are low, most runoff will be direct runoff (overland flow). It is always important to check the total runoff volume and make sure the total runoff volume is less than the precipitation volume. 

    6.  Completed. Add a new Precipitation Gage named PMP Gage to the project and add the PMP time series. The HMR 52 PMP was computed in another workshop and was entered manually as an hourly time series record. 
    7. Create a new meteorologic model named PMP Event, and set the Replace Missing option to Set to Default and select the Specified Hyetograph precipitation method. On the Basins tab, make sure the meteorologic model is linked to both PMF basin models. Then go to the Specified Hyetograph editor and select the PMP Gage for the Bald Eagle Basin subbasin element.
    8. Create two simulations. Use descriptive names when creating the simulation runs, PMF No Peaking and PMF Peak 25P, for example. Both simulations will use the PMP Event meteorologic model and Time Window control specifications. One simulation will use the 2004 Bald Eagle Basin basin model, and the other will use the 2004 Bald Eagle Basin 25P basin model. During a PA, a 50 percent unit hydrograph peaking scenario should also be included.  
    9. Run the simulations and compare the PMF hydrographs. 
      Question 5: What is the difference in Peak Flow and Runoff Volume between the non-peaking and 25% peaking simulation runs?

      No Peaking: Qp = 269,407 cfs, Vol = 24.18 in.
      25% Peaking: Qp = 320,387 cfs, Vol = 24.70 in. 

    10. The below figure shows the original Spillway Design Flood for Sayers Dam. The Spillway Design Flood contained 21.8 inches of precipitation in 48 hours. The peak flow is 251,000 cfs and the runoff volume is 19.05 inches.
      Spillway Design Flood from the Sayers Dam Reservoir Regulation Manual 
      Question 6: How are the current PMF simulations different than the orginal Spillway Design Flood?

      The updated PMF increased in both peak flow and volume in comparison with the Spillway Design Flood (SPF). The peak flow increased from 251,000 cfs to 269,000 and 320,000 cfs using updated procedures. The direct runoff value also increased from 19.05 to 21.79 inches. This illustrates that it is very important to update unit hydrographs and PMPs for Periodic Assessments.

    11.  This workshop walked through a quick method for developing a simple hydrology model to simulate the PMF. The approach presented is a good option when no existing detailed models are available and when only a short amount of time is available to determine the PMF using the most recent HMR.
      Question 7: What is the benefit of fitting a model to the basin unit hydrograph?

      Fitting a model to the basin unit hydrograph allows you to run the simulation with time steps that do not equal the duration of the unit hydrograph. While HEC-HMS will change the duration of the Specified Unit Hydrograph when the time step of simulation is changed, the conversion does not always produce a realistic unit hydrograph. This was illustrated in the first part of this workshop where converting the 20-hour UH to a 1-hour duration produced unrealistic results. Using a synthetic method like the Clark Unit hydrograph avoids these issues.

       Download the final model files here - PMF_Workshop_A_Final.zip