fitPot: Peak over threshold (POT)

In martindurocher/floodRFA: A R-package for regional frequency of floods

Description

Fit the parameters of a peak over threshold (POT) model.

Usage

fitPot(x, ...)

## S3 method for class 'data.frame'
fitPot(obj, ...)

## S3 method for class 'matrix'
fitPot(obj, ...)

## S3 method for class 'formula'
fitPot(form, x, ...)

## S3 method for class 'numeric'
fitPot(x, dt = NULL, u = 0, method = "mle",
  declust = NULL, r = 1, rlow = 0.75, nboot = 1000,
  declust.scale = FALSE)

## S3 method for class 'fpot'
coef(obj, rate = FALSE, ci = FALSE, alpha = 0.05)

## S3 method for class 'fpot'
vcov(obj, rate = FALSE)

## S3 method for class 'fpot'
print(obj)

## S3 method for class 'fpot'
predict(obj, rt = c(2, 5, 10, 20, 50, 100), se = FALSE,
  ci = FALSE, alpha = 0.05, nboot = 0, ...)

Arguments

x	Sample.
dt	Date or time of observation.
u	Threshold.
method	Estimation method. Either 'lmom', 'mle' or 'mle2'.
declust	Method for declustering. If necessary, Either 'run' or 'flood'.
r	Lag parameter for declustering. Either the run parameter or the minimum time between two flood Peaks. The scale must coincide with the observation date dt.
rlow	Declustering parameter. Percentage that define a minimal recession time between two flood peaks.
nboot	Number of bootstrap sample. Used to extimate the covariance matrix with the method of L-moments and predict return period. For prediction, if nboot = 0 confident intervals are obtained by profile likekihood.
se, ci	Should the Standard error or the confident interval be return. The standard error are obtained by the delta method and the confident interval by nonparametric boostrap.

Details

The access functions coef and vcov return respectively the parameters and the variance-covariance matrix of the POT model. The variance covariance matrix is available only with the method 'mle' The access function predict returns the return periods. If dt is a Date the return period is computed in year using the range of observation. The rate (i.e. number of event per year) can be manually adjusted in case. By default rate = 1.

The declustering can be a simple 'run' declustering, i.e. a cluster start when passing a threshold and stop when r steps are below the threshold.

References

Choulakian V, Stephens MA. (2001) Goodness-of-Fit Tests for the Generalized Pareto Distribution. Technometrics. 43(4):478–84.

Coles S. (2001) An introduction to statistical modeling of extreme values. Springer Verlag.

Davison AC, Smith RL. (1990) Models for Exceedances over High Thresholds. Journal of the Royal Statistical Society Series B (Methodological). 52(3):393–442.

See Also

which.floodPeaks, mrlPlot.

Examples

xd <- rgpa(100, 1, -.2)
fit <- fitPot(xd, u = 0)

print(fit)
vcov(fit)
predict(fit)

fit <- fitPot(flow~date, canadaFlood$daily, u = 1000,
               declust = 'flood', r = 14)

print(fit)
plot(flow~date,canadaFlood$daily, type = 'l', col = 4)
points(fit$time,fit$excess+fit$u, col = 2)

predict(fit, se = TRUE,ci = TRUE)

fit <- fitPot(flow~date, canadaFlood$daily, u = 1000,
               declust = 'flood', r = 14, method = 'lmom')

martindurocher/floodRFA documentation built on June 5, 2019, 8:44 p.m.