HIDRA -- sea-level dynamics and coastal flooding forecasts

Mon, 01 Jan 0001 00:00:00 +0000

The team: This research is result of a collaboration between the Visual Cognitive Systems Lab, University of Ljubljana, Faculty of Computer and Information Science, the Marine biology station at the National Institute of Biology (NIB) and the Slovenian Environment Agency (ARSO): Lojze Žust & Matej Kristan (FRI), Anja Fettich (ARSO), Matjaž Ličer (NIB)

HIDRA

Quick links: Live predictions

Interactions between atmospheric forcing, topographic constraints to air and water flow, and resonant character of the basin make sea level modelling in the Adriatic a challenging problem. We present HIDRA (HIigh-performance Deep tidal Residual estimation method using Atmospheric data) to address this challenge. HIDRA is a physics-informed deep model for sea-level forecasting, which makes predictions based on ensemble weather forecasts of the European Centre for Medium-Range Weather Forecasts (ECMWF) and past 24 hours of sea-level measurements at the Koper Mareographic Station. The model HIDRA1 was trained on a dataset of ECMWF weather forecasts for the years 2006-2016 and Koper sea-level measurements for the same time period. Later versions extended the dataset as well as prediction domains, with HIDRA-D now delivering a dense prediction for the entire Adriatic basin.

Results show that HIDRA matches (and surpasses in some cases) the accuracy of a numerical operational model, while being half million times faster, delivering predictions within less than a second on CPU in ~16 ms on GPU.

Sea-surface temperature reconstruction models

Mon, 01 Jan 0001 00:00:00 +0000

CRITER

Satellite observations of sea surface temperature (SST) are essential for accurate weather forecasting and climate modeling. However, these data often suffer from incomplete coverage due to cloud obstruction and limited satellite swath width, which requires development of dense reconstruction algorithms. The current state of the art struggles to accurately recover high-frequency variability, particularly in SST gradients in ocean fronts, eddies, and filaments, which are crucial for downstream processing and predictive tasks. To address this challenge, we propose a novel two-stage method CRITER (Coarse Reconstruction with ITerative Refinement Network), which consists of two stages. First, it reconstructs low-frequency SST components utilizing a Vision Transformer-based model, leveraging global spatio-temporal correlations in the available observations. Second, a UNet type of network iteratively refines the estimate by recovering high-frequency details. Extensive analysis on datasets from the Mediterranean, Adriatic, and Atlantic seas demonstrates CRITER’s superior performance over the current state of the art. Specifically, CRITER achieves up to 44 % lower reconstruction errors of the missing values and over 80 % lower reconstruction errors of the observed values compared to the state of the art.

Geophysics deep learning models on ViCoS Lab

HIDRA -- sea-level dynamics and coastal flooding forecasts

HIDRA

Sea-surface temperature reconstruction models

CRITER