floodnetRfa extra

Introduction

In this repository, I provide supplemental materials that add to my work on the R-package floodnetRfa. It includes codes and summaries resulting from the analysis of the 1114 Canadian stations. In particular, it includes part of the information I collected during the development of the R-package, which I summarize in two datasets: gauged_sites and descriptors.

The objective is to give further insights into the package functionalities and development. I derive all results from a version of the HYDAT database dating of August 11th, 2019. This specific version can be download here, and the user must add the local path of the database into the file config of the root folder.

The repository is divided into independent folders that I describe individually. Note that the results are provided "as is", and come without warranty of any kind.

Folders

Autopot

One method available to perform at-site flood frequency analysis is Peaks Over Threshold (POT). The objective of this section is to use the automatic selection method provided in floodnetRfa to obtain candidates thresholds for each station in gauged_sites.csv. The script autopot_script.R selects the thresholds, and the file autopot_readme.html briefly summarizes them.

Cache

In several sections, scripts run parallel tasks to speed up computation. These scripts save temporary results for each site in its cache folder. For instance, RFA/RFA_script.R uses the cache/RFA folder. The respective sections provide further details.

Dataset

The folder contains two CSV files (zipped) that summarize the information collected on the 1114 Canadian hydrometric stations.

Other

The section contains various small scripts.

• The file cmd_coordinates.R investigates the use of Classical Multidimensional Scaling (CMD) to obtain a coordinate system that preserves the great-circle distance among studied sites.

• The file GetSuperRegions.R contains a function that returns the super regions of a target site as suggested in the dataset gauged_sites.csv.

• The file intersite.R studies ways of estimating the coefficient of correlation of multivariate normal distribution that characterize the spatial dependence among annual maxima.

• The file LmomDiag_fitting.R find polynomial approximations of curves use in the L-moments ratio diagram.

RFA

At-site and regional frequency analysis are performed on each of the 1114 stations. The files atsite_script.R and RFA_script.R fit the extreme distributions and RFA_readme.Rmd compares the precision of the estimators based on the confidence intervals lengths.

Super regions

The recommended strategy for forming pooling groups first identifies a set of stations having similar hydrological properties called super regions. The file super_regions.html explains the way to obtain the super regions suggested in gauged_sites.

Trend

The package floodnetRfa focuses on the analysis of station where flood risk is constant in time (stationary). This section performs standard trend tests. The files trend_script.R evaluates the test statistics and the file trend_readme.html summaries them.