gumbelMaxDistUncert.gumbel_multi_fit: Uncertainty in the Distribution of the Maximum Using the...

In usnistgov/potMax: Empirical Calculation of the Distribution of the Maximum from a Peaks Over Threshold Process

Description

Evaluate uncertainty in the mixture of distributions of maximums assuming the underlying data generating mechanism is the 2D extremal Poisson process with the Gumbel like intensity function (i.e. the tail length is zero), but without specifying a single threshold. The mixture is across multiple thresholds.

Usage

## S3 method for class 'gumbel_multi_fit'
gumbelMaxDistUncert(x, declust_obs, lt_gen,
  n_mc, n_boot, progress_tf = TRUE)

Arguments

x	An S3 object of class gumbel_multi_fit.
declust_obs	(numeric vector) The observed data used by gumbelMultiFit. This will very often be the $declustered_series element of an S3 object of class declustered_series
lt_gen	(numeric scalar) Length of each generated series. The units (seconds, minutes, hours, etc.) should be consistent with the value of lt provided to gumbelMLE.
n_mc	(numeric scalar) The number of samples to draw from the distribution of the maximum
n_boot	(numeric scalar) The number of bootstrap replicates of the distribution of the maximum to create.
progress_tf	Display a progress bar if TRUE, else not.

Details

The results of fitting a many Gumbel like 2D extremal Poisson process are fed into this function. The declustered data are repeatedly sampled with replacement, and for each resampled data set the distribution of the maximum is empirically constructed as described in gumbelMaxDist.gumbel_multi_fit. The bootstrap replicates of the mixture of distributions of the maximum may be used to quantify uncertainty and construct intervals.

Value

An S3 object of class gumbel_max_dist_uncert_multi_thresh, which is a list of length n_boot of S3 objects of class gumbel_max_dist_multi_thresh.

Examples

## Not run: 

complete_series <- -jp1tap1715wind270$value

declustered_obs <- decluster(complete_series)

gumbel_multi_fit <- gumbelMultiFit(x = declustered_obs, lt = 100,
                                   n_min = 10, n_max = 50,
                                   weight_scale = 5)

gumbel_multi_fit_uncert <- gumbelMaxDistUncert(x = gumbel_multi_fit,
                                               declust_obs = declustered_obs$declustered_series,
                                               lt_gen = 200,
                                               n_mc = 1000,
                                               n_boot = 200)

## End(Not run)

usnistgov/potMax documentation built on May 3, 2019, 2:38 p.m.