MIA: artificial intelligence system for sub-seasonal forecast of hydrometeorological variables in Brazil

• Grant number: 20/00566-8
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Duration: November 01, 2020 - July 31, 2021
• Field of knowledge: Physical Sciences and Mathematics - Geosciences - Meteorology
• Principal Investigator: Thomas Christian Marcel Martin
• Grantee: Thomas Christian Marcel Martin
• Host Company: Meteoia Datascience Ltda
• CNAE: Pesquisa e desenvolvimento experimental em ciências físicas e naturais
• City: São Paulo
• Associated researchers: Amanda Sabatini Dufek ; Edson Luiz Shoitchi Yatabe Barbosa ; Humberto Ribeiro da Rocha ; Paola Gimenes Bueno ; Pedro Leite da Silva Dias
• Associated grant(s): 21/14700-0 - MIA: Artificial intelligence system for sub-seasonal forecast of hydrometeorological variables in Brazil, AP.PIPE
• Associated scholarship(s): 21/02110-4 - MIA: artificial intelligence system for sub-sazonal hydrometeorological predictions, BP.PIPE

Abstract

The main goal of this proposal is the development of an automated artificial intelligence forecast system for hydrometeorological variables focused in the subseasonal scale, named MIA. The innovations associated with this project are two-fold, both as a process and a service. As a process, the project will connect state-of-the-art artificial intelligence techniques, little explored in the climate sciences, in an end-to-end automated system of model training, forecast and hyperparameter optimization. As a service, this product of this project will be among the few subseasonal scale specialized systems, scale in which the traditional climate modelling approaches are weakest. As an outcome, we expect to reduce the uncertainties and improve the planning of the Brazilian market regarding the water variability associated with climate. We also expect to place the company in the recent frameworks of industry 4.0 and AutoML (automated machine learning). (AU)