The ocean is a highly complex and evolving networked system characterized by a combination of interconnected processes, which plays an active role in regulating the climate. The changes in the ocean structure are influenced by both internal natural climate variability and persistent anthropogenic changes in the atmosphere, whose signals can be difficult to detect distinctively. Identification and prediction of the “fingerprints” of change caused by internal variability alone is also limited by existing observations. However, this internal variability contributes significantly to regional changes in ocean heat content and sea level rise. In face of such complexity, there is a need for new methods to help interpret these changes and feedback mechanisms. The MALOC project aims to model nonlinear relationships among ocean variables based on historical data that reflect climate variability using machine learning techniques and across different temporal scales. Then, using trained models on proxy-climate data or sea level records alone, assess past and future changes in regional sea levels. The proposed data exploitation techniques, which lead to physically consistent models with the regional underlying ocean-climate dynamics, can help obtain better-informed assessments, as well as the subsequent projection of regional sea level rise in response to climate warming and fluctuations.
The main objective of the MALOC project is to provide improved regional assessment in response to ocean warming by using data-driven machine learning methods. The scope of this project includes a machine learning-based analysis of the possible tendency towards an increase/decrease in near-future regional sea levels due to internal climate variations.