BaySMEV BAYesian Simplified Metastatistical Extreme Value method v1.0.0 Author: Matteo Darienzo (University of Padua, Padua, Italy) Contributions from: Francesco Marra (University of Padua, Padua, Italy). Year: 2025 Funders: NextEU PNRR (PRIN2022 "INTENSE" project, grant 2022ZC2522, https://intenseproject.uniud.it/en/) Last update: 10/01/2026 Purpose: Software BaySMEV (BAYesian Simplified Metastatistical Extreme Value) provides the Bayesian version of SMEV method (Marra et al., 2019, Marani and Ignaccolo, 2015). It contains tools for rainfall extreme analysis with both stationary and non-stationary models within a Bayesian framework. MLE method is also available (as per the original SMEV method of Marra et al., 2019, https://zenodo.org/records/15047817). Method: It identifies all ordinary events within the given precipitation time series, by identifying the maximum value within the wet periods. Wet periods are defined by instants with precipitation > minimum threshold (e.g., 0.1 mm) separated by a user-defined minimum dry period (e.g., 24h). It computes the cdf of extremes as F^n, where F is the cdf of the ordinary events and n is the occurrence frequency of ordinary events (number of events per year). F is assumed as a 2-parameters Weibull distribution with a left-censoring approach (https://zenodo.org/records/11934843). Scale and shape parameters of the Weibull distribution are estimated by means of (choice of the user): MLE with "fminsearch()" minimization function, or Bayesian inference with STAN package ("cmdstanr"). BaySMEV source codes are written in R. An example of main launcher ("smev_main_EXAMPLE.R") and associated ".rds" R data file with precipitation timeseries are provided in order to run the script and test its functionalities and different settings. To apply BaySMEV to gridded datasets (satellite, reanalysis data) please refer to main launcher "smev_main_sat_IMERG.R" which performs SMEV analysis (stationary or nonstationary) for each pixel of the dataset. For creating maps with results of the SMEV analysis computed on gridded datasets, please use postprocessing module "/BaySMEV/postprocessing/smev_results_maps_IMERG.R". To apply BaySMEV to rain gauges datasets, you can use main launcher "smev_main_gauges.R" as an example. single storm type is here implemented. Multi-type SMEV is under development and will be available in future releases. Acknowledegments: This research has been supported by the INTENSE project (raINfall exTremEs and their impacts: from the local to the National ScalE) funded by the European Union - Next Generation EU in the framework of PRIN (Progetti di ricerca di Rilevante Interesse Nazionale) programme (grant 2022ZC2522). BaySMEV codes have been developed and used for paper Darienzo et al. ("Contrasting changes in extreme hourly precipitation across Italy revealed by satellite and re-analysis products, in preparation). BaySMEV takes inspiration from matlab codes of Marra et al., 2019, 2020 downloadable from https://zenodo.org/records/11934843 (SMEV methodology, including storm separation) and https://zenodo.org/records/15047817 (for the non-stationary model implementation). BaySMEV for the Bayesian and MCMC algoryhtm takes ideas from methods: RsTooDS of Benjamin Renard, INRAE (https://zenodo.org/records/5075760); BaM of Benjamin Renard, INRAE (https://github.com/BaM-tools/RBaM); BayDERS of Matteo Darienzo, INRAE (https://github.com/MatteoDarienzo/BayDERS); packages "rstan" and "cmdstanr" for stan Bayesian inference and diagnostics, package "adaptMCMC" for adaptive metropolis algorithm. BaySMEV uses R package 'trend' of Thorsten Pohlert (https://cran.r-project.org/web/packages/trend/index.html) for the trend analysis (MK and Sen's slope). BaySMEV makes use of the following R packages: "rstudioapi", "methods", "lattice", "gridExtra", "reshape","reshape2", "ggplot2", "extrafont", "grid", "gtable", "chron", "coda", "RColorBrewer", "cowplot", "viridis", "psych", "mosaicData", "tidyr", "lubridate", "Kendall", "trend", "tidyverse", "latex2exp", "dplyr", "png","pracma", "CFtime", "ggpubr", "raster", "maptools","sf", "terra", "exactextractr", "ncdf4", "rnaturalearth", "parallel", "scales", "mcmc", "evd", "stats", "R.matlab", "extRemes", "imager", "mmand", "adaptMCMC", "PEIP", "posterior", "loo", "bayesplot", "rstan" Disclaimer: Please, notice that BaySMEV is an experimental software. Further testing and investigation are required for validating the proposed analysis and codes. We are not responsible for any loss (of data, profits, business, customers, orders, other costs or disturbances) derived by their use in the research or operational practice. The authorized user accepts the risks associated with using the software given its nature of free software. It is reserved to expert Users (developers or professionals) having prior knowledge in computer science, hydrology and statistics.