This repository contains the code used for creating all the results in the paper Quantile based modeling of diurnal temperature range with the five-parameter lambda distribution, which performs modelling of diurnal temperature range in Norway using the FPL distribution.

Installation

You can install the package directly from github:

devtools::install_github("siliusmv/FPLD")

Scripts

There are a number of scripts available in the exec/ folder. These are described below.

Master-script

The script master.R runs all the code in its intended order.

Download the distance to sea

Distance to the open sea is used as an explanatory variable in our regression model. The script download_data.R downloads a DEM and computes the distance to sea using that.

Simulation studies

The script univariate_simulation_study.R performs a simulation study where we test the method of quantiles against the starship method and the maximum likelihood method.

Data preparation

Data preparation is performed in the prepare_data.R script.

Data exploration

In the script explore_data.R, we examine the diurnal temperature range data and how it is linked to the available explanatory variables.

Univariate modelling with the FPL distribution

The script local_estimation.R performs univariate modelling of diurnal temperature range using the FPL distribution. The script display_local_estimation_results.R evaluates the model fit.

Regression-modelling with the FPL distribution

The script regional_estimation.R performs regression-modelling of diurnal temperature range using a combination of quantile regression and FPL marginal distributions. The script display_regional_estimation_results.R evaluates the model fit.

Data

Data from all our chosen weather stations is located in the directory inst/extdata