Rafael Rocha
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Rafael Rocha has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Ewerton Oliveira
Instituto Tecnológico Vale / Universidade Federal do Pará
4 shared publications - Eduardo Carvalho
Affiliation Environmental Intelligence Lab, Vale Institute of Technology, Belém, Pará, Brazil
3 shared publications - Ronnie Alves
Instituto Tecnologico Vale Belém PA Brazil
3 shared publications - Douglas Batista da Silva Ferreira
Instituto Tecnológico Vale
2 shared publications - Nikolas Carneiro
Instituto Tecnologico Vale Belém PA Brazil
2 shared publications - Renata Goncalves Tedeschi
Instituto Tecnológico Vale
2 shared publications
Latest Publications
- 29 September 2025
- Published by Sociedade Brasileira de Computação
- p. 261-272
Towards statistical-based prediction models for seasonal precipitation in the southeast of Brazil
Este estudo avalia a previsão de precipitação em 12 pontos de monitoramento no sudeste do Brasil, utilizando cinco modelos estatísticos (MLR, ARIMA, SARIMA, SARIMAX e VARMAX). A acurácia das previsões foi avaliada em dois momentos (M1 e M2), utilizando o RMSE. O modelo MLR apresentou o melhor desempenho no curto prazo (M1) em metade das localidades (50%), enquanto o SARIMAX foi superior em quatro delas (33%). No longo prazo (M2), o VARMAX superou os demais modelos em sete localidades (58%), evidenciando sua capacidade de capturar dinâmicas multivariadas. Os resultados destacam a importância dos modelos estatísticos para a previsão meteorológica local e o planejamento de infraestrutura.
- 29 September 2025
- Published by Sociedade Brasileira de Computação
- p. 261-272
Towards statistical-based prediction models for seasonal precipitation in the southeast of Brazil
Este estudo avalia a previsão de precipitação em 12 pontos de monitoramento no sudeste do Brasil, utilizando cinco modelos estatísticos (MLR, ARIMA, SARIMA, SARIMAX e VARMAX). A acurácia das previsões foi avaliada em dois momentos (M1 e M2), utilizando o RMSE. O modelo MLR apresentou o melhor desempenho no curto prazo (M1) em metade das localidades (50%), enquanto o SARIMAX foi superior em quatro delas (33%). No longo prazo (M2), o VARMAX superou os demais modelos em sete localidades (58%), evidenciando sua capacidade de capturar dinâmicas multivariadas. Os resultados destacam a importância dos modelos estatísticos para a previsão meteorológica local e o planejamento de infraestrutura.
- Open Access
- 17 November 2024
- Published by Sociedade Brasileira de Computação
- p. 49-56
Evaluating Past Horizons for U-Net-Based Precipitation Nowcasting with Radar Data in Southeastern Pará, Brazil
Severe weather events significantly impact daily life, especially during emergencies, affecting human lives and the economy. Decision-making in such events, like heavy rainfall, strong winds, and flash floods, is challenging due to the rapid changes and strong interconnections between the variables involved. Nowcasting models use real-time data, commonly from weather radars, to forecast short-term rain up to 6 hours ahead, supporting decision-making in severe weather situations. These models provide early warnings and precise information about the location, intensity, and duration of these events. Recently, machine learning models for precipitation forecasting have gained prominence due to their ability to learn from data and offer reliable and fast predictions. This study explores precipitation nowcasting using weather radar data in the southeast of Pará, Brazil, focusing on a one-hour forecast horizon utilizing the U-Net architecture. Four models based on U-Net algorithm, investigating past horizons of 30, 60, 90, and 120 minutes, are evaluated using categorical and continuous metrics, and a visual comparison of the 60-minute forecast horizon. The results demonstrate that the model with a past horizon of 120 minutes outperforms the other models in all evaluated metrics, achieving 37.96 and 0.5476 scores in continuous and categorical metrics, respectively, improving the forecast of severe events and decision making up to a 60-minute forecast horizon.
- 17 November 2024
- Published by Sociedade Brasileira de Computação
- p. 49-56
Open AccessEvaluating Past Horizons for U-Net-Based Precipitation Nowcasting with Radar Data in Southeastern Pará, Brazil
Severe weather events significantly impact daily life, especially during emergencies, affecting human lives and the economy. Decision-making in such events, like heavy rainfall, strong winds, and flash floods, is challenging due to the rapid changes and strong interconnections between the variables involved. Nowcasting models use real-time data, commonly from weather radars, to forecast short-term rain up to 6 hours ahead, supporting decision-making in severe weather situations. These models provide early warnings and precise information about the location, intensity, and duration of these events. Recently, machine learning models for precipitation forecasting have gained prominence due to their ability to learn from data and offer reliable and fast predictions. This study explores precipitation nowcasting using weather radar data in the southeast of Pará, Brazil, focusing on a one-hour forecast horizon utilizing the U-Net architecture. Four models based on U-Net algorithm, investigating past horizons of 30, 60, 90, and 120 minutes, are evaluated using categorical and continuous metrics, and a visual comparison of the 60-minute forecast horizon. The results demonstrate that the model with a past horizon of 120 minutes outperforms the other models in all evaluated metrics, achieving 37.96 and 0.5476 scores in continuous and categorical metrics, respectively, improving the forecast of severe events and decision making up to a 60-minute forecast horizon.
- Open Access
- 25 September 2023
- Published by Sociedade Brasileira de Computação
- p. 726-740
A machine learning and statistical learning-based pipeline to perform multipoint rainfall forecasting
Analisar e prever a precipitação é crucial para a sociedade, principalmente quando ocorrem chuvas extremas e inundações. Tais eventos impactam as estruturas socioeconômicas e podem levar a fatalidades. O entendimento da formação das chuvas e das variáveis associadas auxilia no desenvolvimento de modelos preditivos dos níveis de precipitação, auxiliando na tomada de decisões nas cadeias produtivas e na mobilidade urbana. Avanços recentes neste campo resultam do aumento do poder de processamento do computador e da disponibilidade de dados meteorológicos em todo o mundo. Este estudo se concentra na avaliação dos modelos SARIMA, RNN e XGBoost para prever a precipitação mensal ao longo de uma ferrovia no norte do Brasil. Usando valores de precipitação defasados, SARIMA teve melhor desempenho em 11 dos 13 pontos (com R² variando de 0,704 a 0,817), enquanto RNN superou nos pontos restantes (15,39% dos pontos avaliados).
- 25 September 2023
- Published by Sociedade Brasileira de Computação
- p. 726-740
Open AccessA machine learning and statistical learning-based pipeline to perform multipoint rainfall forecasting
Analisar e prever a precipitação é crucial para a sociedade, principalmente quando ocorrem chuvas extremas e inundações. Tais eventos impactam as estruturas socioeconômicas e podem levar a fatalidades. O entendimento da formação das chuvas e das variáveis associadas auxilia no desenvolvimento de modelos preditivos dos níveis de precipitação, auxiliando na tomada de decisões nas cadeias produtivas e na mobilidade urbana. Avanços recentes neste campo resultam do aumento do poder de processamento do computador e da disponibilidade de dados meteorológicos em todo o mundo. Este estudo se concentra na avaliação dos modelos SARIMA, RNN e XGBoost para prever a precipitação mensal ao longo de uma ferrovia no norte do Brasil. Usando valores de precipitação defasados, SARIMA teve melhor desempenho em 11 dos 13 pontos (com R² variando de 0,704 a 0,817), enquanto RNN superou nos pontos restantes (15,39% dos pontos avaliados).
- Open Access
- 25 September 2023
- Published by Sociedade Brasileira de Computação
- p. 782-793
Assessing U-Net Model Performance in Weather Radar-Based Precipitation Nowcasting: A Reflectivity Threshold Analysis
Severe weather events pose a global threat, causing property damage and endangering lives. Weather-related disasters account for 50% of all natural and technological disasters. The development of accurate prediction systems are crucial for early warnings and mitigation. The present study evaluates the effectiveness of the U-Net model in weather radar-based precipitation nowcasting, considering reflectivity thresholds. Visual comparison and evaluation metrics are used to assess observed and predicted reflectivity. The 10 dBZ threshold achieved a prominent result, accurately predicting over 75% of values above the reflectivity threshold. Results contribute to improving severe weather prediction and decision-making.
- 25 September 2023
- Published by Sociedade Brasileira de Computação
- p. 782-793
Open AccessAssessing U-Net Model Performance in Weather Radar-Based Precipitation Nowcasting: A Reflectivity Threshold Analysis
Severe weather events pose a global threat, causing property damage and endangering lives. Weather-related disasters account for 50% of all natural and technological disasters. The development of accurate prediction systems are crucial for early warnings and mitigation. The present study evaluates the effectiveness of the U-Net model in weather radar-based precipitation nowcasting, considering reflectivity thresholds. Visual comparison and evaluation metrics are used to assess observed and predicted reflectivity. The 10 dBZ threshold achieved a prominent result, accurately predicting over 75% of values above the reflectivity threshold. Results contribute to improving severe weather prediction and decision-making.