“Enhancing Ocean Surface Gridded Products with VarDyn: A Physics-Informed Approach”
Gridded products of ocean physical surface variables, such as daily 2D maps of Sea Surface Height (SSH) and Sea Surface Temperature (SST), are used in various scientific and operational applications. These products generally rely on optimal interpolation algorithms applied to sparse satellite observations. The accuracy of the mapping can be affected by several factors, including limited space-time sampling for altimetry, cloud obstruction for infrared sensors, and low spatial resolution for microwave sensors. Although the recent SWOT mission provides SSH observations at an unprecedentedly high spatial resolution, its long revisit time prevents the tracking and differentiation of small-scale structures, particularly small-scale balanced motions and internal tides. For these reasons, recent studies have focused on enhancing algorithms to improve the accuracy of ocean surface gridded products, utilizing various strategies such as data assimilation and machine learning.
In this seminar, we will present VarDyn, a hybrid mapping methodology that leverages simple physics within advanced variational inversions. We will demonstrate how a simple physical constraint based on a Quasi-Geostrophic (QG) model enables VarDyn to outperform operational products in mapping SSH from both conventional altimetry and SWOT data. Interestingly, we will show how the QG model can be combined with a linearized Shallow Water model to map and separate balanced motions from internal tides. Finally, we will highlight VarDyn’s ability to leverage the dynamical coupling between SSH and SST, leading to improved mapping accuracy for both variables compared to operational products.
