Skip to contents

Compute Return Levels with Credible Limits

RL.TVGEVBayes.Rd

Compute return levels with credible limits. The result can be used to produce the classical return level plot by using. autoplot.TVGEVBayes.

Usage

# S3 method for TVGEVBayes
RL(
  object,
  newTimeRange = NULL,
  period = NULL,
  level = 0.7,
  credintType = c("HPD", "eqtail"),
  smooth = missing(period),
  ...
)

Arguments

object

A TVGEVBayes object.

newTimeRange

A time range to be passed to timeRange.

period

A vector of return periods.

level

Credible level to be used for the intervals.

credintType

The type of credible interval. "HPD" corresponds to the Highest Posterior Density interval and "eqtail" corresponds to the "equal-tail" choice, where both tails are given the same probability \((1 - level) / 2\).

smooth

Logical. If TRUE, the lower and upper credible bounds will be smoothed by using smooth.spline.

...

Not used yet.

Value

An object with class "RL.GEVBayes" inheriting from "data.frame".

Details

For a given period \(T>1\) e.g. \(T = 100\), the return level \(\rho(T)\) is defined as the quantile with exceedance probability \(1 / T\). In the time-varying framework, the quantile is for a marginal GEV distribution hence relates to a specific block in time which has to be given by using newTimeRange. Hence the return level \(\rho\{T; \boldsymbol{\theta}(t)\}\) where \(\boldsymbol{\theta}(t)\) is the vector of the thee GEV parameters for the block \(t\). This is deterministic function of the GEV parameter and it can or should come along with a credible interval.

Note

Since the class "TVGEVBayes" is devoted to time-varying models it would be tedious (and potentially misleading) to derive plotting positions for the observations used in the fit. So we do not provide on this plot empirical points as usually shown in the non time-varying (stationary) framework.

See also

credInt.