CONSTRUCTION OF A PACKAGE IN R PROGRAMMING LANGUAGE FOR DETERMINING EXTREME DATA ABOVE A THRESHOLD: AN APPLICATION WITH RAINFALL IN THE STATE OF SÃO PAULO, BRAZIL

• Grant number: 25/17416-2
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: November 01, 2025
• End date: October 31, 2026
• Field of knowledge: Physical Sciences and Mathematics - Probability and Statistics - Applied Probability and Statistics
• Principal Investigator: Gilberto Rodrigues Liska
• Grantee: Arthur Fellipy Pereira Silva
• Host Institution: Centro de Ciências Agrárias (CCA). Universidade Federal de São Carlos (UFSCAR). Araras , SP, Brazil

Abstract

Analyses of extreme events have received significant attention in television media, and research is no exception. There are computational resources that enable analysis of such events, and those based on the R programming language are prominent in academia due to their ease of use compared to other languages, as well as their open and free use. However, algorithms that organize original data series into extreme value series in R and are available in official sources do not exist. Furthermore, the global increase in the frequency of extreme weather events, particularly heavy rainfall, has significantly impacted Brazil, requiring more sophisticated statistical approaches for risk analysis and management. Therefore, the project aims to develop an R package to help organize extreme value series and, thus, facilitate and optimize the modeling of extreme precipitation series in the state of São Paulo using the Peaks over Threshold (POT) methodology of Extreme Value Theory (EVT). Aligned with SDG 13, the study will provide tools for public policies on climate adaptation. The expected results include: (1) an open-source R package, documented and made available to the scientific community, that organizes extreme series, selects optimized thresholds, fits Generalized Pareto (GP) distributions using maximum likelihood, calculates return levels with confidence intervals, and validates models with statistical tests and performance metrics; (2) accurate predictions of extreme precipitation for strategic municipalities in São Paulo; (3) technical support for resilient urban planning, disaster prevention, and scientific publications. The initiative integrates computational innovation and socio-environmental relevance, standing out for its practical applicability in a climate change context and for democratizing access to extreme analysis via open-source tools.