The purpose of this presentation is to guide the user throughout the steps of using the Graphical User Interface (GUI) of the R package floodnetRfa
.
The GUI allows performing flood frequency analysis of hydrometric stations using local and regional methods.
Utilities for predicting flood quantiles at ungauged basins are only available from the R terminal.
library(floodnetRfa) FloodnetApp() # Create a shortcut on the desktop FloodnetApp(shortcut = "C:/Users/JohnDoe/Desktop/FloodnetApp.cmd")
The principal method in flood frequency analysis consists to fit a distribution on the sample of annual maximum streamflow.
Estimation is performed by the method of linear moments (L-moments).
Dispersion and confidence intervals are evaluated using parametric bootstrap.
A second approach is to fit a distribution of streamflow peaks using the Generalized Pareto distribution (GPA).
Peaks are exceedances extracted from a declustering method based on a threshold and the basin drainage area
Estimation is done using the method of maximum likelihood.
Dispersion and confidence intervals are evaluated using parametric bootstrap.
RFA used the index-flood model, which assumes that all distributions inside a homogenous region are proportional.
The sample mean is used as a scaling factor.
Estimation is done by averaging the L-moments of the scaled sample.
Dispersion and confidence intervals are evaluated using parametric bootstrap.
An homogenous region is neighborhoods (pooling group) formed around a target site.
It used a similarity measure based on the timing and regularity of the annual maximum discharge.
The homogeneity is assessed by a heterogeneity measure based on the Linear Coefficient of Variation (LCV).
Heterogeneous sites are sequentially removed from an initial pooling group.
For more details on the methodologies, please consult the Floodnet guideline or the documentation of the following R functions.
FloodnetAmax
FloodnetPot
FloodnetPool
'2020-07-15'
The model configuration panel is used to configure the desired model. Once configured, press the Fit
button.
Model ID: An identifier of the model. It must be unique and meaningful.
Method: Method used to perform frequency analysis (e.g. AMAX, POT, RFA).
Target Site: Station ID of the site of interest.
Return period: List of return periods used to evaluate the flood quantiles. Must be separated by a comma.
For POT, the distribution is known but the threshold and the target drainage area must be passed.
The automatic procedure selects the lowest threshold that respects the modeling assumptions.
For RFA, the user must select a Super Region
that identifies a class of stations similar to the target site.
Pooling groups include only stations inside the target super region.
The super region is associated with a column of the station data file.
RFA can be applied to both AMAX and POT approaches.
RFA POT requires that the threshold are known for all sites in the pooling group.
Fit
button, the graphics of the return levels (flood quantiles) versus the return period is produced for assessing the quality of the fit.The standard error (se) and the confidence intervals (lower, upper) are also available.
After pressing the Fit
button, the fitted model is added to a list, where each entry is identified by its Model ID
.
Remove
button.Show
button.Save
buttons..Rdata
.Open
button load previously saved models.Reset
button empties the current list of models.Show
button.If multiple models were selected, the display panel allows switching between them.
The <-Back
button allows going back the the model screen
Some graphics are specific to a model, some include mutiple models.
For example, the comparison of the confidence intervals allows visualizing the difference the estimated flood quantiles and it variability.
Some graphics provides information on the selected pooling groups.
For example, the L-ratio diagram shows the dispersion of 3rd and 4th L-moment among the pooling group.
Export
button saved the element of the screen in a PDF.