Download PDF
Download page Task 6. Define and Fit Copulas in HEC-Neptune.
Task 6. Define and Fit Copulas in HEC-Neptune
Create a Copula Analysis
- Create a new Copula Analysis by selecting Analysis | Copula Analysis from the top ribbon.
General Information
- Enter Philippi as the analysis name.
- Enter a description such as Bivariate copula for Tygart River flow and Philippi precipitation.
- Click the Next button.
Scenarios
- We want to fit a copula to two scenarios:
- Annual peak 3-day precipitation with near coincident streamflow
- Annual peak 1-day streamflow with near coincident precipitation
- Under Copula Scenario A, enter the following:
- Name: Precipitation Dominant
- Conditioning Data: precipitation_dominant - Precipitation_AMS (Precipitation)
- Conditioning Data Label: Precipitation AMS
- Coincident Data: precipitation_dominant - Flow_Coincident (Discharge)
- Coincident Data Label: Flow Coincident
- Under Copula Scenario B, enter the following:
- Name: Flow Dominant
- Conditioning Data: flow_dominant - Flow_AMS (Discharge)
- Conditioning Data Label: Flow AMS
- Coincident Data: flow_dominant - Precipitation_Coincident (Precipitation)
- Coincident Data Label: Precipitation Coincident
- Under Envelope Curve, enter the following:
- X-axis Label: Precipitation (in)
- Y-axis Label: Flow (cfs)
- The Envelope Curve labels will be used to label the plot axes. Note the information at the top of the screen - the envelope curve axes follow Copula A. The Envelope Curve x-axis label should correspond to the dominant/conditioning variable under Copula Scenario A and the y-axis label should correspond to the coincident variable under Copula Scenario A.
- The Scenarios tab should appear as in the image below.

- Click Next.
Marginal Distributions
On this tab, we will tell HEC-Neptune which marginal distributions to fit based on Task 4. Fit Marginal Distributions in HEC-SSP. In the alpha release of Neptune, all distributions are fit using L-moments.
- Under the Precipitation Dominant panel, select Generalized Extreme Value under Conditioning Variable Distribution. This will fit a GEV distribution to the precipitation AMS dataset.
- Under Non-conditioning Variable Distribution, select Gamma to fit this distribution to the coincident flow dataset.
- Under the Flow Dominant panel, select Generalized Extreme Value under Conditioning Variable Distribution. This will fit a GEV distribution to the flow AMS dataset.
- Under Non-conditioning Variable Distribution, select Logistic to fit this distribution to the coincident precipitation dataset.
- Notice that Neptune displays the distribution parameters.
- Click Next.
Output Frequency Ordinates
- On this tab, the user can enter or edit the desired output frequency ordinates. Note that these ordinates are joint exceedance probabilities.
- Leave the defaults and click Next.
Copula Selection
On this tab, you must select a copula model (Gumbel, Gaussian, Student's t, Joe, Frank, or Gumbel) to fit to each scenario that you defined on the Scenarios tab. Neptune displays a table of the copula model name, dependence parameter, and Akaike Information Criterion (AIC). Lower AICs indicate better agreement between the data and the model. The rows of the table are sorted in ascending order of AIC. The copula model with the lowest AIC/best model fit is shown in the first row.
Neptune also displays contour plots of the data (shown in black points) and the probability isolines. You can plot the contour lines as a single color by toggling on the Single contour color option.
At the bottom of the tab, Neptune provides a Selected Copula dropdown menu. By default, the copula model with the lowest AIC is selected.
- Examine the tabular information and contour plots for each copula scenario, Precipitation Dominant and Flow Dominant. Select the Next and Previous buttons in the upper right to toggle between the copula fits for the Precipitation Dominant and Flow Dominant scenarios.


- Leave the default Selected Copula for both scenarios. The default copula should be the Gumbel copula in both scenarios since this copula model produdces the lowest AIC.
- Click the Next button in the lower right.
- The analysis will take a few seconds to compute.
Design Events
- On the Design Events tab, you will see an isoline plot and a table with the design points.

- Read through the explanation in the HEC-Neptune User's Manual.
Question 7. What is the 1/100 design event? In other words, what are the values of precipitation and river discharge that produce a joint 1/100 annual exceedance event?
A precipitation of 4.36 inches and a streamflow of 43,480 produce a joint 1/100 AEP event.
Question 8. Are the values of the design events certain? Why or why not?
The value of the design events is uncertain. The design point is determined from the following steps:
- One million randomly generated samples are draw from the copula(s).
- The kernel density of all points on a single isoline are computed.
- The density of points on a single isoline is normalized between 0 and 1.
- The event pair with the highest relative density is deemed the design event for a given joint probability.
Since the design event is computed from Monte Carlo simulation of samples, there is uncertainty in the value.
Return to Copulas for Interior Drainage for the final files.