Determining the best fitting distribution of annual precipitation data in Serbia using L-moments method

TL;DR

This study identifies the best statistical distribution for annual precipitation data in Serbia using L-moments, finding that the GEV distribution fits best and providing various return levels for risk assessment.

Contribution

It applies L-moments to compare multiple distributions for precipitation data and determines the most suitable model specifically for Serbia's climate data.

Findings

GEV distribution is the best fit for Serbian annual precipitation data.

Provides return levels for 2, 5, 10, 20, 50, 100, and 1000 years.

Uses multiple goodness-of-fit tests to validate the model.

Abstract

The monthly precipitation data from 29 meteorological stations for the period 1946--2012 from Serbia were used. To describe the behaviour of precipitation data at a specific location, it is necessary to identify the distribution that best fits the data. For this purpose, three distributions i.e. generalized extreme value (GEV), generalized Pareto (GPD) and generalized logistic (GLO) distribution were fitted using the method of L-moment. The goodness-of-fit for the selected three distributions was confirmed using L-moment ratio diagram and three tests namely relative root mean square error, relative mean absolute error and probability plot correlation coefficient. From the result of this analysis, the GEV distribution was selected as the best fitting distribution of annual precipitation data in Serbia. The 2, 5, 10, 20, 50, 100 and 1000-years return levels are provided.