The Copula Analysis allows users to pair datasets for up to two copulas within the user interface. A copula scenario is defined by conditioning (extremal) events and near coincident events.

Neptune allows users to specify one or two copula scenarios. Defining one copula scenario is appropriate when (1) it is obvious which input variable's extremes are causing extremes at the response variable or (2) when extremes at both input variables are always coincident. 

In some situations, it is obvious which variable's extreme events are causing extreme events at the response variable. Let's consider a tributary and mainstem junction. The two variables are mainstem flow and tributary flow and we are interested in describing the frequency of tributary stages. In this example, the mainstem gage's drainage area is significantly larger than the tributary gage's drainage area so we are confident that mainstem river flows are causing backwater, which is the driver of tributary stages. In these situation, a single copula scenario can be defined that describes the joint probability of extreme mainstem flows and near coincident tributary flows. 

An example of a "one-sided" copula is available here: Copula Example: Joint Probability Model for Stock Indices

In other situations, it is not obvious which input variable is responsible for extreme events at the response variable. In this case, we are still considering mainstem and tributary flows but now the drainage areas are similar so it's less obvious which waterway is driving stages on the tributary. In this case, a "two-sided" copula can be computed: the first copula describes the joint probability between extreme mainstem flows and near coincident tributary flows while the second copula describes the joint probability between extreme tributary flows and near mainstem flows.

An example of a "two-sided" copula is available here: Copulas for Interior Drainage

Extremal events can be defined using either an annual maximum series (AMS) or a partial duration series (PDS). Coincident events are identified based on ordinal association, as measured by Kendall’s tau, within a user-specified time window surrounding the extremal events.

An example of copula scenarios for a “two-sided” copula analysis of an interior drainage study is shown below. More generally, Copula Scenario A represents extreme events in Variable A coupled with near-coincident events in Variable B, while Copula Scenario B represents extreme events in Variable B coupled with near-coincident events in Variable A. The designation of the variables as A or B is arbitrary and does not affect the results.

The Envelope Curve labels are used to label the x- and y-axes of the isoline plot on the Results tab. They should describe the variable type and units, rather than the sample type (extreme/coincident), as shown below.