Towards Sustainable Nowcasting: Exploring Lightweight Models for Local Precipitation Forecasting

This project focuses on investigating more efficient and simplified approaches for very short-term precipitation forecasting, known as nowcasting. The goal is to evaluate and adapt lightweight machine learning models that can deliver accurate predictions using fewer computational resources. By reviewing existing models, testing their performance, and exploring ways to simplify their architecture without losing accuracy, the project aims to create practical solutions that can be used even in regions with limited infrastructure. This work is particularly relevant in the face of increasing extreme weather events linked to climate change. It will support both institutions by promoting international collaboration, strengthening expertise in climate data modeling, and generating results that can contribute to disaster prevention and urban resilience. The project also aligns with global sustainability goals, such as climate action and innovation in infrastructure, and enhances Canada’s leadership in developing accessible, data-driven solutions for environmental challenges.

Faculty Supervisor:

Gabriel Spadon

Student:

Partner:

Universidade Estadual de Mato Grosso do Sul

Discipline:

Computer science

Sector:

Information and Communications Technology; Environmental Science and Technology

University:

Dalhousie University

Program: