A Methodology Based on Random Forest to Estimate Precipitation Return Periods: A Comparative Analysis with Probability Density Functions in Arequipa, Peru DOI Open Access

Johan Anco-Valdivia,

Sebastián Valencia-Félix, Alain Jorge Espinoza Vigil

et al.

Water, Journal Year: 2025, Volume and Issue: 17(1), P. 128 - 128

Published: Jan. 6, 2025

Precipitation within specific return periods plays a crucial role in the design of hydraulic infrastructure for water management. Traditional analytical approaches involve collecting annual maximum precipitation data from station followed by application statistical probability distributions and selection best-fit distribution based on goodness-of-fit tests (e.g., Kolmogorov-Smirnov). However, this methodology relies current data, raising concerns about its suitability outdated data. This study aims to compare Probability Density Functions (PDFs) with Random Forest (RF) machine learning algorithm estimating at different periods. Using twenty-six stations located various parts Arequipa department Peru, performance both methods was evaluated using MSE, RMSE, R2 MAE. The results show that RF outperforms PDFs most cases, having more precision metrics mentioned estimates 2, 5, 10, 20, 50, 100 years studied stations.

Language: Английский

A Methodology Based on Random Forest to Estimate Precipitation Return Periods: A Comparative Analysis with Probability Density Functions in Arequipa, Peru DOI Open Access

Johan Anco-Valdivia,

Sebastián Valencia-Félix, Alain Jorge Espinoza Vigil

et al.

Water, Journal Year: 2025, Volume and Issue: 17(1), P. 128 - 128

Published: Jan. 6, 2025

Precipitation within specific return periods plays a crucial role in the design of hydraulic infrastructure for water management. Traditional analytical approaches involve collecting annual maximum precipitation data from station followed by application statistical probability distributions and selection best-fit distribution based on goodness-of-fit tests (e.g., Kolmogorov-Smirnov). However, this methodology relies current data, raising concerns about its suitability outdated data. This study aims to compare Probability Density Functions (PDFs) with Random Forest (RF) machine learning algorithm estimating at different periods. Using twenty-six stations located various parts Arequipa department Peru, performance both methods was evaluated using MSE, RMSE, R2 MAE. The results show that RF outperforms PDFs most cases, having more precision metrics mentioned estimates 2, 5, 10, 20, 50, 100 years studied stations.

Language: Английский

Citations

0