EGU2020-18033
https://doi.org/10.5194/egusphere-egu2020-18033
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

SOSeas: An assessment tool for predicting the dynamic risk of drowning on beaches

Javier F. Bárcena1, Javier García-Alba1, Felipe Fernández1, Javier Costas-Veiga2, Marcos Mecías2,3, María Luisa Sámano2,3,4, David Szpilman5,6, and Andrés García1
Javier F. Bárcena et al.
  • 1Universidad de Cantabria, Environmental Hydraulics Institute, Santander, Spain
  • 2Universidad Europea del Atlántico, Santander, Spain
  • 3Centro de Investigación y Tecnología Industrial de Cantabria (CITICAN), Santander, Spain.
  • 4Universidad Internacional Iberoamericana, Campeche, Mexico.
  • 5Brazilian Life Saving Society, Rio de Janeiro City, Brazil
  • 6Rio de Janeiro Fire Department (CBMERJ), Rio de Janeiro City, Brazil

This study focuses on the development of an operational service to prevent one of the major public health problems worldwide, drowning (https://soseas.ihcantabria.com/). Approximately, there are 360,000 annual deaths from drowning all around the world, although, global estimates may significantly underestimate the real values. In order to reduce the third leading cause of unintentional injury death worldwide, there is an urgent need to increase the understanding of drowning causes in highly, nonlinear and complex dynamic systems, such as beaches.

Usually, process-based models have been directly used to provide information for risk analysis or by means of hybrid downscaled systems in local areas. Nevertheless, this type of modelling is not operative to generate a worldwide system capable to offer a forecasting risk of drowning as a function of hydrodynamic information, suggesting that computational requirements will be an impediment to applications where a quick answer is required, e.g., managing temporary closures of bathing sites. Accordingly, different techniques have been proposed to overcome the large computational burden associated with process-based models, called dynamic emulation modelling. From a technical perspective, Artificial Neural Networks (ANN) models have a strong predictive ability for nonlinear systems allowing it to be synchronized with another system, and can enhance the overall reliability and applicability of process-based models simplifying the mathematical descriptions of the physical structure and mechanism of chaotic systems. From the operational perspective, the implementation of ANN models is highly efficient at a very low cost compared to the implementation of process-based models.

The catalogue of events included the information of metocean conditions provided by the global reanalysis and forecasts of Copernicus Marine Service (CMEMS) and the information provided by the Brazilian Life Saving Society (SOBRASA) about drowning in Santa Catarina beaches (Brazil). During the last 18 years, lifeguards have collected more than 132,000 observations from 139 coastal beaches, which have 346 lifeguard posts. This information has been provided in two databases: (1) a database of events (drownings or similar) and (2) a database of status flag at each lifeguard post. Events database contained 52,712 records from January 2001 to May 2019. The Flags database contained 79,487 records from November 2016 to July 2019.

From these databases, the chosen key-variables to predict the dynamic risk of drowning on beaches using electronic flags were geomorphological variables: morphological modal state, beach orientation, presence of estuary/river mouth, and metocean variables: Waves (significant total height, mean wave period, direction of waves), Wind (magnitude of wind velocity, direction of winds), Water Level (water level variation), and Currents (magnitude of marine currents).

The application to the Santa Catarina beaches demonstrated ANN models are viable surrogates of highly nonlinear process-based models and highly variable forcings to understand the synchronization between metocean conditions and drowning risks at beaches. The results showed that the neural networks conveniently reproduced the status flag of beaches. Finally, it is worthy to mention this service has created the availability of non-existent tools that enhance safety in these aquatic spaces, generating economic assets as a sign of high quality tourism.

How to cite: Bárcena, J. F., García-Alba, J., Fernández, F., Costas-Veiga, J., Mecías, M., Sámano, M. L., Szpilman, D., and García, A.: SOSeas: An assessment tool for predicting the dynamic risk of drowning on beaches, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-18033, https://doi.org/10.5194/egusphere-egu2020-18033, 2020

Displays

Display file