Abstract's details
Improving mesoscale altimetric data in the Mediterranean Sea: convolutional retreatment of AVISO products with SSH/SST synergy
Event: 2016 Ocean Surface Topography Science Team Meeting
Session: Others (poster only)
Presentation type: Poster
Keywords: SSH fields, along-track data, convolutional models, OSSE, multi-tracer synergy
Multi-satellite measurements of altimeter-derived Sea Surface Height have provided a wealth of information about ocean circulation and atmosphere-ocean interactions. Numerous improvements have been brought to the optimal interpolation of daily SSH fields from along-track data. Yet, horizontal scales below 100km remain mainly unresolved. Especially, in the Mediterranean Sea, an important fraction of the mesoscale range, typically from 10km to 100km, is not revealed by altimeter-derived L4 products.
Here, we investigate a novel retreatment of AVISO products to resolve the horizontal scales sensed by current along-track altimeter data. The key feature of our framework is the use of linear convolutional operators to model the fine-scale SSH detail as a function of different sea surface fields, especially SST and SSH. Formally, it resorts to:
(1) SSH = SSHLR + H1*SSHLR + H2*SST + N
where SSHLR, SST refer respectively to the a low-resolution SSH field, typically the L4 gridded AVISO SSH product, and to a high-resolution L4 daily SST product. H1 and H2 refer respectively to linear convolutional operators and N to a residual noise, comprising unresolved scales. This model comprises the SQG model as a special case, where H1 is null and H2 is a fractional Laplacian operator. We consider a discrete representation of convolutional operators H1 and H2 to perform their estimation from along-track data according to a classical linear regression according to least-square criterion.
We carry out an observing system simulation experiment to quantitatively evaluate the proposed model. We use a four-year ROMS numerical simulation for a region south of Balearic Islands with rich mesoscale dynamics. We sample along-track altimeter data from the simulated SSH fields using a four-altimeter series of real along-track positions, and compute AVISO-like SSH fields using an optimal interpolation of the simulated along-track data. Model (1) can lead to relative mean improvements of 20% in the estimation of the SSH and of the kinetic energy, with a much greater contribution of SSHLR in model (1). Interestingly, we report potential significant improvements in the resolution of the spectral signatures for horizontal scales from 30km to 100km.
Figure. Evaluation of the proposed approach: distribution of the spectral slopes for the range [30km,100km] (left), distribution of the relative MSE for the SSH (center) and the kinetic energy (right). We compare the high-resolution SSH (magenta), the associated low-resolution optimally-interpolated SSH (black) and the reconstructed SSH field.
Multi-satellite measurements of altimeter-derived Sea Surface Height have provided a wealth of information about ocean circulation and atmosphere-ocean interactions. Numerous improvements have been brought to the optimal interpolation of daily SSH fields from along-track data. Yet, horizontal scales below 100km remain mainly unresolved. Especially, in the Mediterranean Sea, an important fraction of the mesoscale range, typically from 10km to 100km, is not revealed by altimeter-derived L4 products.
Here, we investigate a novel retreatment of AVISO products to resolve the horizontal scales sensed by current along-track altimeter data. The key feature of our framework is the use of linear convolutional operators to model the fine-scale SSH detail as a function of different sea surface fields, especially SST and SSH. Formally, it resorts to:
(1) SSH = SSHLR + H1*SSHLR + H2*SST + N
where SSHLR, SST refer respectively to the a low-resolution SSH field, typically the L4 gridded AVISO SSH product, and to a high-resolution L4 daily SST product. H1 and H2 refer respectively to linear convolutional operators and N to a residual noise, comprising unresolved scales. This model comprises the SQG model as a special case, where H1 is null and H2 is a fractional Laplacian operator. We consider a discrete representation of convolutional operators H1 and H2 to perform their estimation from along-track data according to a classical linear regression according to least-square criterion.
We carry out an observing system simulation experiment to quantitatively evaluate the proposed model. We use a four-year ROMS numerical simulation for a region south of Balearic Islands with rich mesoscale dynamics. We sample along-track altimeter data from the simulated SSH fields using a four-altimeter series of real along-track positions, and compute AVISO-like SSH fields using an optimal interpolation of the simulated along-track data. Model (1) can lead to relative mean improvements of 20% in the estimation of the SSH and of the kinetic energy, with a much greater contribution of SSHLR in model (1). Interestingly, we report potential significant improvements in the resolution of the spectral signatures for horizontal scales from 30km to 100km.
Figure. Evaluation of the proposed approach: distribution of the spectral slopes for the range [30km,100km] (left), distribution of the relative MSE for the SSH (center) and the kinetic energy (right). We compare the high-resolution SSH (magenta), the associated low-resolution optimally-interpolated SSH (black) and the reconstructed SSH field.
Contribution: poster_ostst2016_rfablet.pdf (pdf, 3662 ko)
Back to the list of abstract