ESA signature

VarDyn: Dynamical joint-reconstructions of Sea Surface Height and Temperature from multi-sensor satellite observations

VarDyn: Dynamical joint-reconstructions of Sea Surface Height and Temperature from multi-sensor satellite observations

Tagged:

Objectives

The use of satellite data has revolutionized the study of upper ocean dynamics, offering global, continuous, and real-time information about various physical processes at the ocean’s surface. These data provide key insights into ocean circulation, temperature distribution, wave patterns, and other dynamic factors. For instance, Sea Surface Height (SSH), observed by altimetry, is a key variable for monitoring ocean circulation. Similarly, Sea Surface Temperature (SST) serves as a crucial indicator of heat exchange between the ocean and atmosphere.

Typically, scientific and operational applications require gap-free, gridded satellite data (L4 products). However, operational algorithms often depend on basic interpolation techniques applied to L3 satellite data, which tend to filter out fine-scale processes. In this project, an hybrid methodology, called VarDyn, which combines minimal physically-based constraints with a variational scheme is designed to enhance the mapping capabilities of upper ocean dynamics. By synthesizing multi-modal satellite observations, VarDyn aims at improving the accuracy of SSH and SST maps in comparison to operational products, both in terms of RMSE and effective spatial resolution. 

Figure 1: the VarDyn method

Data

Near-real time along-track SSH products are considered (DOI https://doi.org/10.48670/moi-00147). The 1Hz product of all available missions (Sentinel-6A, Jason-3, Sentinel-3A, Sentinel-3B, Saral/AltiKa, Cryosat-2, HY-2B) are used. DUACS multimission altimeter data processing system ensures all the mission measurements to be homogenized with respect to Sentinel-6A ones (reference missions).  

For SST observations, estimates from the Advanced Microwave Scanning Radiometer (AMSR-2) and the Global Precipitation Measurement Microwave Imager (GMI) are considered (version 8.2). To benefit from their time sampling coverage, both ascending and descending passes are selected. Diurnal cycle corrections are not applied. The data are distributed by the Remote Sensing Systems (REMSS, https://data.remss.com).

 
Figure 2: Relative Vorticity derived from SSH for the VarDyn product (top) and for the operational DUACS product (bottom)

Method

The VarDyn (for Variational Dynamical interpolation) method is illustrated on Figure 1.  SSH and SST observations are jointly assimilated in simple dynamical models: (i) SSH dynamics is simulated by a 1.5-layer Quasi-Geostrophic (QG) model; (ii) SST dynamics is simulated by the passive tracer conservation law. The two models are coupled by the geostrophic assumption, assuming that SST is advected by the SSH-derived geostrophic currents. The SSH and SST observations are assimilated within a 4D Variational scheme, optimizing the forcing fluxes of the reduced models to fit the simulated trajectory to the observations. 

Figure 3: SST spatial gradients from infrared sensors, VarDyn and REMSS-MW products on September 15th, 2023, over the whole study domain (left) and zoomed in three regions along the main current (right). 

Results

As a proof of concept, the VarDyn method has been tested in a study domain spanning the Gulf Stream and its extension. The reconstructed SSH and SST maps have been compared with high resolution independent satellite data: (i) from the recently launched SWOT satellite for SSH and (ii) from infrared sensors (e.g., onboard Sentinel-3) for SST. Systematic comparison with state-of-the-art operational products (DUACS L4 for SSH, REMSS-MW L4 for SST) have been performed. 

Qualitatively, the VarDyn products reveal smaller scale structures than in the operational products, with better contrasts and improved continuity of frontal structures and eddies (Figure 2 and Figure 3). 

Quantitatively performances show a clear gain in RMSE, reaching 30% (10%) in high (low) energetic areas for SSH (Figure 4). The same conclusion applies for effective spatial resolutions: most of the gain occurs in high energetic areas while matching operational resolutions in low energetic areas. 

Figure 4: Diagnostics for SSH reconstruction over the study domain. Top: Standard Deviation (STD) of SSH observed by SWOT. Left: RMSE evaluated with SWOT for VarDyn and DUACS products. Right: effective spatial resolution λeff evaluated with SWOT for VarDyn and DUACS products. The two bottom panels depict the gain/reduction ratio (blue means better performance for VarDyn product relative to the considered metric, RMSE or λeff ). The green line refers to the 0.15m STD contour.

Perspectives for Future Research

For future investigations, an original asset of the VarDyn method is an estimation of the SST forcing terms.

These effective fluxes contain many hidden contributions: atmospheric fluxes, vertical and ageostrophic advection, mixed layer depth variations. Figure 5 illustrates the resulting signature of a hurricane marine cold wake and the associated SST fluxes. More dedicated investigations are needed, but VarDyn opens numerous perspectives, such as the study of amplification mechanisms and/or restratification rates for extreme events, like cyclones and marine heatwaves.

Figure 5: Signature of extra-tropical hurricane Franklin (September 2023) on VarDyn SST (left) and the associated estimated fluxes (right).

Publications

Code and data

This research was carried out by Florian Le Guillou.

European Space Agency