Open AccessCOMPARISON OF L-MOMENT AND METHOD OF MOMENTS AS PARAMETER ESTIMATORS FOR IDENTIFICATION AND CHOICE OF THE MOST APPROPRIATE RAINFALL DISTRIBUTION MODELS FOR DESIGN OF HYDRAULIC STRUCTURES
Author(s) -
Augustine Iyekeoretin Agbonaye,
Ebierin Akpoebidimiyen Otuaro,
O.C. Izinyon
Publication year - 2022
Publication title -
journal of civil engineering, science and technology
Language(s) - English
Resource type - Journals
ISSN - 2462-1382
DOI - 10.33736/jcest.4207.2022
Subject(s) - gumbel distribution , estimator , moment (physics) , probability distribution , generalized extreme value distribution , mathematics , statistics , generalized pareto distribution , location parameter , method of moments (probability theory) , moment distribution method , distribution (mathematics) , extreme value theory , probability density function , mathematical analysis , physics , parametric statistics , classical mechanics
In rainfall frequency analysis, the choice of a suitable probability distribution and parameter estimation method is critical in forecasting design rainfall values for varying return periods at every location. Previously, some researchers in Nigeria used the method of moments (MoM) while others used the L-moment method (LMM) as parameter estimators. However, a more accurate result is obtainable if both estimators are used and their results are compared and ranked to obtain the most appropriate distribution models for each location This study compared the performance of two forms of parameter estimation, namely the method of moments (MoM) and the L-moment method (LMM). This was aimed at identifying and selecting the best fit probability distribution models among three distribution models for the design of hydraulic structures. These models are Generalized Pareto (GPA), Generalized Extreme Value (GEV), and Gumbel Extreme Value (EVI). Annual rainfall series of ten gauging stations with data from 33-50 years from ten southern States of Nigeria obtained from NIMET were used for Rainfall Frequency Analysis (RFA). At five locations, the best fit probability model was the GPA probability distribution model with L-Moment. EVI and GEV probability distribution models with the method of moments were the most appropriate probability models at two locations each. EVI probability distribution model with the L-moment was the most appropriate probability model at one place. The findings confirmed that no single distribution outperformed all others at all stations. Since no single model is regarded preferable for all practical purposes, the best-fit probability model with parameter estimator at any location is site-specific. Consequently, available models and parameter estimators are filtered based on the situation at hand and the type of data available. The identified best fit models with the most appropriate parameter estimator would be a tool to help decision-makers in sizing hydraulic structures in the area.