Banca de QUALIFICAÇÃO: ALEXANDRE ESTRELA DE LACERDA NOBREGA

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : ALEXANDRE ESTRELA DE LACERDA NOBREGA
DATE: 30/06/2025
TIME: 09:00
LOCAL: Video conferência (Google Meet)
TITLE:

Proposed machine learning-based tool for precipitation time series forecasting under climate change scenarios

 

KEY WORDS:

machine learning, climatic changes, downscaling

PAGES: 90
BIG AREA: Ciências Exatas e da Terra
AREA: Ciência da Computação
SUMMARY:

Climate change, largely driven by human activity, has led to severe global impacts, particularly
the increase in extreme weather events. The projection of these changes relies on
global climate models, which simulate climate behavior based on different emission scenarios.
However, these models have low spatial resolution, limiting their direct application to
regional and local contexts. To address this limitation, statistical downscaling techniques
have been employed to generate time series of climate variables, such as precipitation, with
greater regional detail. With technological advancements, artificial intelligence and machine
learning techniques have shown promising results in the development of such models,
often outperforming traditional methods. Nevertheless, their application remains limited,
partly due to the lack of standardized procedures and the complexity involved. Furthermore,
after applying statistical downscaling, the generated data are restricted to specific
adjusted points, requiring the use of spatialization models to estimate values in surrounding
areas. This entire process, from downscaling to spatialization, is often inaccessible
to professionals who need these data but lack technical expertise in the methodologies
used. In this context, the present study investigated the main techniques employed in
statistical downscaling and proposed a standardized workflow for their application. The
research was conducted in a specific region as a case study, integrating data from global
climate models with local information. Based on this, a scalable computational tool was
developed, featuring an intuitive interface capable of applying the constructed models
and generating time series data in a practical and accessible manner, based on the user’s
specified location.

 

COMMITTEE MEMBERS:
Presidente - 2180207 - ITAMIR DE MORAIS BARROCA FILHO
Interno - 2668551 - ANDRE MORAIS GURGEL
Interno - 1669545 - DANIEL SABINO AMORIM DE ARAUJO