bootstrapCI: Determine confidence intervals to rainfall intensities...

In dazamora/IDFtool: Estimation Intensity-Duration-Frequency Curves for Rainfall Series

Determine confidence intervals to rainfall intensities estimated by a PDF function.

Description

Conducts bootstrap to randomly sample of intensity values 'n' times for a specified distribution to estimate the confidence interval for each given non-exceedance probability.

Usage

bootstrapCI(Intensity, Parameters, Type = "Gumbel", Rsample = 1000,
  Return.P, Conf.Inter = 0.95)

Arguments

Intensity	a numeric vector with intensity [mm/h] values of different years for a specific time duration (e.g. 5, 15, 120 minutes, etc.)
Parameters	list with three elements: (i) type of distribution function (ii) fitted parameters, and (iii) source to call specfic function in the lmomco package.
Type	a character specifying a name of the probability distribution function fitted (see selecDIST) by fitDISTRI function.
Rsample	An integer representing number of resamples to conduct when confidence interval will be computed.
Return.P	a numeric vector with return periods like non-exceedance probabilities.
Conf.Inter	level of the confidence interval.

Value

A list of:

• nonexceed.prob a numeric vector with non-exceedance probabilities.

• lower.lim a numeric vector confidence bound lower for quantile estimates.

• upper.lim a numeric vector confidence bound upper for quantile estimates.

• Para.set a matrix containing estimated distribution parameters for each resample.

• quantiles a matrix of quantile estimates for each resample.

Author(s)

David Zamora <[e-mail]> Water Resources Engineering Research Group - GIREH

Examples

data(inten)
data(Pargumbel)
Tp <- c(2, 3, 5, 10, 25, 50, 100)
FR <- 1 - 1/Tp
CI.test <- bootstrapCI(Intensity = inten, Parameters = Pargumbel, Type = "gumbel",
Rsample = 50, Return.P =FR, Conf.Inter = 0.90)
(CI.test$Conf.Inter[,2])
# result = 125.9501, 139.6132, 154.6232, 173.6736, 197.5024, 215.0634, 232.4939
 

dazamora/IDFtool documentation built on Jan. 1, 2023, 3:29 p.m.