SOCIB research contributes to optimise Search and Rescue operations in coastal areas

• A new study published in Frontiers in Marine Science led by SOCIB proposes a comprehensive approach to optimise SAR operations in coastal areas.
• This study helps emergency responders to minimize response time, thus contributing to maintaining our seas and coasts safe and clean and reducing the costs of operations.
• This work is built on previous work conducted during the IBISAR project, coordinated by SOCIB and funded by Mercator Ocean International in the frame of the Copernicus Marine Service User Uptake programme of the European Commission.

Researchers led by the Balearic Islands Coastal Observing and Forecasting System (SOCIB), involving the participation of the IMEDEA (CSIC-UIB), AZTI, Nologin Consulting, Puertos del Estado and the Spanish Maritime Safety and Rescue Society (SASEMAR), have published a new study on the sensitivity of the widely used Skill Score (SS) metric, an intuitive measure for evaluating the accuracy of ocean forecast models in predicting particle trajectories. Published in the journal Frontiers in Marine Science, this work contributes to addressing the overarching concern of maritime emergency responders: they want to know how good the data is and which data source they should select to run their applications, from an extensive (and growing) catalogue. By guiding them in the correct data source selection, they will be able to optimize their Search and Rescue (SAR) planning in coastal areas, thus increasing the safety and security at sea and reducing operating costs while contributing to maintaining our seas and coasts clean.

This study is built on previous work conducted during the IBISAR project, coordinated by SOCIB and funded by Mercator Ocean International in the frame of the Copernicus Marine Service User Uptake programme of the European Commission. The IBISAR project has been carried out in public-private partnership with AZTI and RPS Ocean Science, with the active collaboration of Puertos del Estado and SASEMAR. Overall, this work emphasises how scientific research can ensure the extension of science-based added-value products in societal relevant user-centered downstream services. 

Evaluating forecast accuracy

In case of an incident at sea, SAR operators run trajectory models to predict the drift of their target induced by the effect of ocean currents, waves, and winds and define a search area. The skill of drift prediction is highly dependent on the accuracy of the met-ocean forecast data used to advect the trajectory model. In order to cover diverse spatiotemporal scales and guarantee near real-time availability, different ocean forecast models are made available to SAR operators. When all predictions agree (i.e., their simulated trajectories are similar), there is a high level of confidence in the prediction, and the search area is reduced. But if different forecasts result in disparate trajectories, there are multiple viable outcomes, and the level of confidence decreases. Therefore, SAR operators need methods to assess, within the shortest possible time, which forecast model is likely to give the most accurate prediction at the moment and in the region of the incident. Drifter trajectories are generally used to evaluate the drift prediction accuracy, and the SS is one of the most widely used statistics for comparing observed and simulated trajectories. 

Map covering the four specific SAR areas of responsibility from the Spanish Maritime Search and Rescue Agency showing the location of the SAR incidents from 2019 colored based on their distance to the closest coastal point (as indicated in the color bar). The seven operational HF radar systems available within the area are marked with red boxes. Information source: Spanish Maritime Search and Rescue Agency.

In this study, the authors have assessed the accuracy of four forecast models, three of them from the Copernicus Marine Service, available in the Ibiza Channel (Western Mediterranean Sea) and evaluated the applicability of the SS in coastal areas, where most of the SAR incidents occur and drifter observations are scarce. They have analysed the sensitivity of the SS to the forecast time and showed that an overly long forecast time (i.e. 72 hours) leads to an overestimation of the SS. Thus, the researchers have concluded that, in coastal areas characterized by high variability of the surface currents, a short forecast time should be applied (e.g., 6 hours), although it leads to more irregular SS patterns. 

Furthermore, researchers have examined the best way to identify the model with the highest mean accuracy over an area and along a specified period. To avoid biased conclusions, they recommend the use of a slightly different definition of the SS, the novel SS*. Finally, addressing the sparseness of drifter observations in coastal areas, the authors have analysed the use of trajectories derived from High Frequency (HF) radar surface currents for assessing the accuracy of the models. HF radar data have the advantage to provide continuous measurements with high spatio-temporal resolution over wide coastal areas, improving not only the robustness of the SS statistics but also the spatial and temporal assessment of the model accuracy. The researchers have concluded that HF radar-derived trajectories provide very valuable inputs for assisting SAR efforts operationally.

• Reference article

Révelard A, Reyes E, Mourre B, Hernández-Carrasco I, Rubio A, Lorente P, Fernández CDL, Mader J, Álvarez-Fanjul E and Tintoré J (2021). Sensitivity of Skill Score Metric to Validate Lagrangian Simulations in Coastal Areas: Recommendations for Search and Rescue Applications. Frontiers in Marine Science. 8:630388.