Banca de QUALIFICAÇÃO: LUCAS DE SIQUEIRA SANTOS

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DISCENTE: LUCAS DE SIQUEIRA SANTOS
DATA : 11/03/2025
LOCAL: PLATAFORMA DIGITAL
TÍTULO:

MACHINE LEARNING INTEGRATING CLIMATE DATA WITH GRACE MISSION DATA FOR THE RECONSTRUCTION OF TERRESTRIAL WATER STORAGE ANOMALIES IN BRAZIL

PALAVRAS-CHAVES:

terrestrial water storage anomalies, GRACE, machine learning, reconstruction, climate change

PÁGINAS: 43
RESUMO:

This research presents reconstructions of historical terrestrial water storage anomalies (TWSA) across Brazil’s 12 major river basins by integrating GRACE mission data with climate and anthropogenic variables using machine learning techniques. The study aims to extend the GRACE-derived TWSA time series back to 1985, overcoming the limited duration of satellite observations by employing Random Forest (RF) and Long Short-Term Memory (LSTM) models. While the RF model demonstrated better performance - with an overall R² of 0.84 and strong interpretability through variable importance analyses (highlighting the time of the year, soil moisture, and land use factors such as mosaic of uses and rice crops) - the LSTM model yielded satisfactory results in only a subset of basins, suggesting that model parameters must be tailored to specific hydrological conditions. The reconstructions capture seasonal and long-term trends and underscores the impact of both climatic elements and human activities on TWS dynamics. These findings can provide insights for water resource management in Brazil and enhance the understanding of hydrological variability in the context of climate change, by making it possible to analyze TWS variations in the country since 1985.

MEMBROS DA BANCA:
Interno - ***.677.059-** - HENRY DIVERTH MONTECINO CASTRO - UFPE
Externo à Instituição - PEDRO RODRIGUES MUTTI - UFRN
Presidente - 1769444 - RODRIGO MIKOSZ GONCALVES