Background

Multiple stream gages are operated by the Baltimore District (NAB) and United States Geologic Survey (USGS) in and around Foster Joseph Sayers Dam in Pennsylvania.  The locations of the gages in relation to Bald Eagle Creek and Sayers Dam are shown in the following figure.

Stream Gage Locations

Annual peak flows at the Spring Creek at Milesburg, PA (Spring Creek), Bald Eagle Creek Bl Spring Creek at Milesburg, PA (Milesburg), and Bald Eagle Creek at Blanchard, PA (Blanchard) gages are shown in the following figure.  This data can be retrieved using the USGS plugin within HEC-SSP.

Normally, stream gage records are relatively short (i.e. less than 100 years).  However, within dam and levee safety studies, estimates of the magnitude of infrequent large floods are needed to adequately assess the risk associated with the structure(s) in question.  Together with the observed record, a probability model (e.g. Log-Pearson Type III, Generalized Extreme Value, Log-Normal, etc.) is used to make inferences about these extremely rare floods.

In many watersheds, multiple types of flood-causing mechanisms are capable of generating large flows in a given year.  The annual maximum flood for a given year is the largest of these flows.  Generally, the annual maximum series (AMS) for a gage location is analyzed without regard for the type of flood that caused each individual annual maximum; however, in many cases the AMS may be made of a mixture of floods of varying cause.  A foundational assumption to estimating a population probability distribution from a limited sample is that the sample is representative of the population and made of independent and identically-distributed (IID) observations.

When multiple flood types are contained within the AMS, the assumption of an identical distribution can become tenuous, affecting the ability to make inferences about rare floods.  Additionally, the behavior of the right-hand tail (which is especially important within dam and levee safety studies) may be dominated by a type of flood mechanism that is less dominant in the more-frequent part of the annual maximum flood frequency curve.  Treating multiple flood types as separate IID samples of floods of a particular cause allows for more accurate examination of the frequency of very rare floods.  The Mixed Population analysis within HEC-SSP can be used to combine flood mechanism-specific frequency curves into a frequency curve that reflects the occurrence of more than one type of flood in a year.

Bald Eagle Creek is affected by different flooding mechanisms occurring on a seasonal basis, which means that the annual maximum series may contain floods caused by more than one mechanism.  By splitting the record of daily flows at Milesburg into multiple seasons and analyzing them separately, the IID assumption made in analyzing the AMS can be more closely examined.

Objectives

In this workshop you will gain experience assessing flood frequency when more than one flood mechanism is at play.  First, you will import multiple datasets for use within all subsequent analyses.  Then, you will fit an LPIII distribution to the all-season AMS which doesn't explicitly separate floods by their causal mechanism.  Afterwards, you will then fit multiple probability distributions to different AMS, each representing a specific flood mechanism.  Following that task, you will then combine these multiple flood mechanism-specific flow-frequency curves within a Mixed Population analysis to create a combined flood frequency curve.  Finally, you will compare the all-season flood frequency curve against that which was produced by the Mixed Population analysis and note key differences.  The following major tasks will serve as an outline for the workshop:

  1. Create a new HEC-SSP project and import flow data for the Milesburg gage location.
  2. Fit an LPIII distribution using Bulletin 17C (B17C) techniques to the all-season AMS.
  3. Fit probability distributions to the flood mechanism-specific AMS.
  4. Create a Mixed Population Analysis and compute a combined flood frequency curve with uncertainty.
  5. Compare the all-season B17C and Mixed Population analysis results.

Spend approximately 10 minutes per task to accomplish Tasks 1 - 5.

Continue to Task 1. Create a New HEC-SSP Project and Import Flow Data for the Milesburg Gage Location.