UKFE (version 1.0.2)

Uncertainty: Uncertainty quantification for gauged and ungauged pooled estimates

Description

Quantification of aleatoric uncertainty for pooling results for the gauged and ungauged case

Usage

Uncertainty(
  x,
  Gauged = FALSE,
  qmed = NULL,
  Dist = "GenLog",
  Conf = 0.95,
  fseQMED = 1.55,
  UrbAdj = FALSE,
  URBEXT = NULL,
  Plot = TRUE,
  IncAMest = TRUE,
  Parametric = TRUE
)

Value

A dataframe with 10 rows and four columns. Return period in the first column, central estimate in the second, lower in the third, and upper in the fourth. If Plot = TRUE, a return level plot is also returned.

Arguments

x

the pooled group derived from the Pool() or PoolSmall() function.

Gauged

a logical argument with a default of FALSE. If FALSE the uncertainty intervals are calculated for the ungauged case. If TRUE they are calculated for the gauged case.

qmed

the QMED estimate for the ungauged case. It is derived from the observed AMAX if Gauged equals TRUE.

Dist

a choice of distribution to use for the estimates. Choices are "GEV", "GenLog", "Gumbel", or "Kappa3". The default is "GenLog".

Conf

the confidence level of the uncertainty intervals. Default is 0.95. Must be between 0 and 1.

fseQMED

the factorial standard error of the QMED estimate for an ungauged assessment. The default is 1.55.

UrbAdj

applies an urban adjustment to the growth curves.

URBEXT

URBEXT value for the site of interest. This is necessary if UrbAdj equals TRUE.

Plot

logical argument with a default of TRUE. If TRUE a return level plot with results and margin of error is plotted. If FALSE, it is not.

IncAMest

logical argument with a default of TRUE. Sometimes when doing gauged (enhanced single site analysis), the central estimate of the single site estimate is outside the intervals of the ESS estimate. When this argument is true the confidence interval is expanded to include the central estimate for the single site. If FALSE, it is not.

Parametric

logical argument with a default of TRUE. If TRUE, the bootstrapping is done by simulation with the distribution of choice. If FALSE the bootstrapping is done by resampling with replacement.

Author

Anthony Hammond

Details

Uncertainty for both the gauged (enhanced single site) and ungauged case are quantified according to the bootstrapping procedures, which account for weights in the pooling group, detailed in Hammond, A. (2021). Sampling uncertainty of UK design flood estimation. Hydrology Research. 1357-1371. 52 (6). Note that this function only quantifies sampling (aleatoric) uncertainty. It does not quantify uncertainty associated with models, model choices applied or hydrometric data. Lastly, the method assumes that AMAX samples within the pooling group are independent of each other and serially independent and identically distributed.

Examples

# Get an ungauged pooling group
pool_203018 <- Pool(GetCDs(203018), exclude = 203018)

# Quantify the central estimate and uncertainty
Uncertainty(pool_203018, qmed = QMED(GetCDs(203018)))

# Get a pooling group with subject site included
pool_203018 <- Pool(GetCDs(203018))

# Quantify the central estimate and uncertainty
Uncertainty(pool_203018, Gauged = TRUE)