This is a Preprint and has not been peer reviewed. The published version of this Preprint is available: https://doi.org/10.13140/RG.2.2.34874.45768. This is version 1 of this Preprint.
Characterization of historical rainfall patterns: linear segmentation and quadratic bifurcation
Downloads
Authors
Jacques Blanchart 
Abstract
This article proposes a method for characterizing rainfall transitions based on historical series. Two synthetic parameters are analyzed: the annual or seasonal amount of rainfall (RR) and the median of daily rainfall (m). This median is estimated by means of a log-logistic adjustment (Fisk), ensuring a robust measurement of the daily rainfall distribution. Two complementary approaches are used to detect changes:
(i) a linear segmentation into a maximum of five schemes, selected according to the AIC/BIC information criteria ;
(ii) a quadratic bifurcation.
The results make it possible to highlight transition periods on the series studied. The Perpignan historical series is used as a case study, and avenues for extension to other regions and other detection tools are discussed.
DOI
https://doi.org/10.31223/X55T88
Subjects
Physical Sciences and Mathematics
Keywords
climate, rainfall, pluviometric regimes, perpignan, Weather Station
Dates
Published: 2025-11-28 06:50
Last Updated: 2025-11-28 06:50
There are no comments or no comments have been made public for this article.