Self-Organizing Maps-based ocean currents forecasting system

Abstract

An ocean surface currents forecasting system, based on a Self-Organizing Maps (SOM) neural network algorithm, high-frequency (HF) ocean radar measurements and numerical weather prediction (NWP) products, has been developed for a coastal area of the northern Adriatic and compared with operational ROMS-derived surface currents. The two systems differ significantly in architecture and algorithms, being based on either unsupervised learning techniques or ocean physics. To compare performance of the two methods, their forecasting skills were tested on independent datasets. The SOM-based forecasting system has a slightly better forecasting skill, especially during strong wind conditions, with potential for further improvement when data sets of higher quality and longer duration are used for training.

Effect of small scale transport processes on phytoplankton distribution in coastal seas

“…As ocean surface currents forecasting systems are being developed 5 , 59 , our methodology and the diagnostics exploited here could be applied on current predictions and thus serve to project future spatial distribution and magnitude of Chl-a in the coastal ocean. More generally, our study proves that the combination of both LCSs and FDLD computed from HFR is a powerful framework to study the effect of transport on biological quantities in coastal seas, as well as to localize convergence/divergence zones which are relevant for the tracking of debris accumulation or jellyfish aggregations, and for other coastal management activities, such as search and rescue missions and oil spill management.…”

Ocean Circulation in the Western Gulf of Mexico Using Self‐Organizing Maps

“…Later, Liu and Weisberg (, ) used SOMs to extract patterns of ocean currents from moored velocity measurements on the West Florida Shelf and to relate the coastal upwelling/downwelling patterns and the velocity structures of a coastal upwelling jet to local winds. More recently, Vilibić et al () used SOMs for an ocean surface current forecasting system. The above examples show the versatility and capability of SOMs as a powerful and robust technique for pattern recognition and feature extraction, where nonlinearity plays a major role, such as the ocean physics in the WGoM.…”

HF Radar Activity in European Coastal Seas: Next Steps toward a Pan-European HF Radar Network

“…A neural network based approach to obtain short term forecasts for currents and water levels from HFR data was presented in Wahle and Stanev (2011). Recently, Vilibić et al (2016) presented an ocean surface currents forecasting system, based on a Self-Organizing Maps and applied to HFR measurements and wind numerical forecasts.…”

“…In Europe, the number of systems is growing with over 50 HFRs currently deployed and a number in the planning stage. Nowadays, these systems are integrated in many European coastal observatories with proven potential for monitoring (e.g., Wyatt et al, 2006;Molcard et al, 2009;Berta et al, 2014b) and even providing short-term prediction of coastal currents (e.g., Orfila et al, 2015;Solabarrieta et al, 2016;Vilibić et al, 2016), and inputs for data assimilation and the validation and calibration of numerical ocean forecasting models, especially near the coast (e.g., Barth et al, 2008Barth et al, , 2011Marmain et al, 2014;Stanev et al, 2015;Iermano et al, 2016). The growing number of HFRs, the optimization of HFR operation against technical hitches and the need for complex data processing and analysis, highlight the urgent requirement to increase the coordination in the HFR community.…”