HIDRALERTA is a forecast and early warning system for coastal and port regions capable of predicting emergency situations, as well as carrying out risk assessment. The system concentrates on evaluating wave overtopping and flooding scenarios. It uses offshore sea-wave and wind forecast data, as well as tide data, as input to determine wave overtopping at specific locations. The calculation of the mean overtopping discharge over a structure is made through artificial neural network tools and/or empirical formulae, which are able to define the spatial distribution of the flow behind the structure. The HIDRALERTA system is composed of four modules developed in Open Source Web Technology, mainly using the Python Language. The Warning System can be configured and adapted for specific scenarios. The system has been successfully tested for the Port and Bay of Praia da Vitória, Terceira Island, Azores, Portugal, and is now under development and being implemented at the ports of Madalena do Pico and S. Roque do Pico. This paper describes the work carried out to date on the system and its application to these two ports in the Pico Island, Azores.

Introduction

Wave overtopping of maritime structures by sea waves is one of the main physical processes that threaten port and coastal areas around the world and may lead to flooding, damage to both property and the environment, and injuries or even death of people in exposed locations. Overtopping and flooding have recently been intensified due to climate change and have become more relevant issues for the public, coastal and port authorities, the civil protection and even politicians, Figure 1.

The anticipation of the occurrence of hazardous events related to overtopping and flooding is critical in the effective management of infrastructures and in the adoption of mitigation measures by stakeholders to prevent fatalities and to reduce the disaster risk in port and coastal areas.

Portugal and, in particular, the Azores islands, have a significant historical record of such serious incidents caused by extreme wave and water-level conditions. Despite the severity of the sea conditions and the importance of the socio-economic activities in those areas, Portugal does not have yet a fully operational forecast and early warning system for wave overtopping and flooding.

For these reasons, it is imperative to empower the national authorities with an adequate forecast and early warning system that allows the identification of emergency situations and enables the selection of measures by those entities to avoid loss of lives and minimize damage. Moreover, the system may also act as a long-term management tool since it can simulate the response to future scenarios related to climate change, such as the increase of the mean sea level and/or of the storm severity, which will increase coastal flooding probability.

With a view to build and implement an effective early warning system for the Portuguese coast, focused in port areas, the HIDRALERTA system (Poseiro, 2019) was developed and, in 2015, its first prototype was implemented in the port of Praia da Vitória, in the Terceira Island, Azores, and it has been operating since then. Under the framework of the ECOMARPORT project (MAC/1.1b/081), other prototypes of the system have been under development and implemented at the ports of Madalena do Pico and S. Roque do Pico, both at the Pico Island, Azores. Those prototypes are intended to predict and assess, on a regular basis, the risk of wave overtopping and consequent flooding, and the risk to sea navigation, in particular as regards the safety of moored and maneuvered vessels, both risks triggered by wave conditions.

This paper describes the work carried out to date on the system regarding wave overtopping/flooding and its application to the ports of S. Roque do Pico and Madalena do Pico.

The HIDRALERTA system

The HIDRALERTA system (Poseiro et al., 2013; Sabino et al., 2018; Poseiro, 2019) is a wave overtopping/flooding forecast system with early warning and risk assessment capabilities. As a forecast and early warning tool, HIDRALERTA enables the identification, in advance, of the occurrence of emergency situations, prompting the responsible entities to adopt measures to avoid loss of lives and minimize damage. As a long-term planning tool, the system uses datasets of several years of sea-wave characteristics and/or pre-defined scenarios, and evaluates wave overtopping and flooding risks for the protected areas, allowing the construction of risk maps based on the combined use of Geographic Information Systems (GIS) and the Analytic Hierarchy Process (AHP) (Poseiro, 2019).

The HIDRALERTA system was developed in a python framework. The development of the HIDRALERTA methodology used the basic conditions and requirements of the case study of Praia da Vitória. Nevertheless, the system was conceived with a generic architecture, which can be applied to other case studies.

The HIDRALERTA system encompasses four main modules (Figure 2), namely:

  • Sea-state Characterization, where the offshore wave conditions are propagated to inshore and inside ports using numerical models;

  • Wave Run-up and Overtopping determination, using empirical formulae or artificial neural network (ANN) tools;

  • Risk Assessment, which defines a risk level for the predicted overtopping values;

  • Warning System, which integrates all the information and disseminates warnings.

Using the Sea-state Characterization module (I), the system downloads hindcast or forecast data, predicting offshore wave conditions by the regional model WAM (WAMDI Group, 1988). Still-water-levels are predicted by adding astronomical tide levels (XTide model; Flater, 1998) and storm surges, which are obtained from the mean-sea-level (MSL) pressure (HRES model; Persson, 2001), using an empirical relationship. To transfer the offshore wave characteristics to the port entrance and then into the port, the SWAN (Booij et al., 1999) and the DREAMS (Fortes, 2002) models are applied (Figure 3a). Since there is no unique (and computational efficient) wave model that can simulate the whole propagation from offshore to inshore, including confined areas like harbours, marinas, etc., the use of several numerical models will take advantage of the potential of each model.

In the Wave Run-up and Overtopping module (II), these sea conditions are then used, in a port area scenario, as input to the ANN tool NN_OVERTOPPING2 (Coeveld et al., 2005), together with cross-section characteristics of the coastal structures, to obtain an estimate of the mean overtopping discharge, q, at each structure cross-section (Figure 3b). To determine the spatial distribution of wave overtopping behind the overtopped structure, the system applies an empirical method. In a beach scenario, the application of the DREAMS model and the NN_OVERTOPPING2 tool is unnecessary. Empirical formulae are used instead to estimate the flood levels in coastal areas obtained by using empirical formulae associated to the calculation of the maximum run-up, Rumax (or Ru2%, the run-up exceeded by only 2% of the run-ups) (Poseiro et al., 2014).

The Risk Assessment module (III) generates risk maps obtained from the application of risk assessment methodologies and using long-term time series (hindcast data) provided by modules I and II. This module was conceived for long-term planning and decision-making.

In the Warning System module (IV), several thresholds for the maximum run-up, Rumax (or Ru2%), and/or mean overtopping discharge, q, are defined for each stretch of the structures, bearing in mind the characteristics of the overtopped structures and of the protected area, and the activities developed there. Warnings are issued whenever pre-defined thresholds are exceeded, with warning levels and hazard signals shown for each stretch, to identify the affected elements/activities. The general architecture of the Warning System module and its procedure for warning are shown in Figure 4. The HIDRALERTA methodology was developed with the aim of obtaining a flexible framework application, where a specific configuration of the system is activated at run time or where several cases can be configured separately. Spatial data includes mandatory features, such as cartography, bathymetry and land use. According to the type of coastal area considered (with or without artificial structures), the DREAMS model may be used, but this is not compulsory (if no artificial structures exist).

The HIDRALERTA system runs in a computer cluster, called MEDUSA (Lopes et al., 2016), which makes it possible to run this system in a real-time mode (all numerical models are effectively running every day), using a parallelization technique, and considering a deterministic approach for all the forecast chain. Currently, HIDRALERTA needs approximately 1 hour to generate new forecast results (up to 72-hour forecast).

Main description

The port of Madalena do Pico is located on the northwest coast of the Pico Island and is protected by two breakwaters (Figure 5): the North and the West breakwaters.

North breakwater represents the main port protection structure against sea waves, mainly from W-N. With a total length of approximately 530 m, only about 250 m correspond to the berthing quay, the so-called “Madalena Dock”. At the breakwater root, the armor is protected by 150 kN tetrapods and rock of 90-120 kN. 150 m after the breakwater bend, the weight of the tetrapod units changes to 240 kN and 2 layers of rock of 1-3 ton at the slope of 3(V):4(H). On the inner side, the structure has two layers of rock of 1-3 ton and 5-8 ton.

The West breakwater protects the port from the southwest wave conditions. It is a rubble mound breakwater protected by cubes, placed on slopes of 2:3, which end on 10-30 kN rock mattresses located above the natural rocky bottom. The breakwater's crest is at +5.5 m (ZH) (with ZH the chart datum) and its deepest part is located at about -10.5 m (ZH).

Port of S. Roque do Pico is located at the north coast of the Pico Island (Figure 5). It is protected by a breakwater rooted at Ponta da Lage, with an approximate SE alignment. With a total development of about 400 m, it consists of two straight sections, each of them about 200 m long. The landward section protects the harbor embankment. The most seaward section is a berthing quay on the inner side, with a length of about 180 m, with service depths of -4.5 m (ZH) at about the first 40 m and of

-7.5 m (ZH) at the last 140 m. The breakwater service width is around 20 m and the breakwater head is founded at -50 m (ZH). The breakwater is protected by both concrete and rock units.

Application of modules I, II and IV of HIDRALERTA are described below, since they are part of the HIDRALERTA warning system prototype which is implemented in both ports.

Module I – Wave characteristics

The system starts by requesting from the data source (ECMWF - European Centre for Medium-Range Weather Forecasts) wave and wind fields from the WAM model and pressure parameters from HRES. Then, it creates different layouts to represent the offshore sea state (using matplotlib library from the python programming language): firstly, for the North Atlantic and then to the Azores archipelago (Figure 6).

Given the WAM model results, one selects the values closer to the Pico Island that can represent the boundary conditions for sea-wave propagation models to be used afterwards. These values are transferred into the ports of Madalena do Pico and S. Roque do Pico by using two models for sea-wave propagation and deformation: the SWAN and the DREAMS models.

Module II - Wave run-up and overtopping

The estimates of overtopping are obtained by using the ANN tool NN_OVERTOPPING2. With the results of the DREAMS model in front of each structure of the ports, and considering their main characteristics, it is possible to calculate mean overtopping discharges, q (l/s/m). From the overtopping predictions, a layout is generated for each instant (Figure 7). To provide an idea of the overtopping intensity, circles with different diameters are shown for the 65 points. The larger the circle, the highest the mean overtopping discharge. The layout also indicates the maximum mean overtopping discharge predicted for the case study for each instant.

Module IV - Warning system

The maximum mean overtopping discharge predicted for the case study for each instant at each point of the structure is compared with established pre-set thresholds. The pre-set thresholds have been defined in close collaboration with the local authorities of Madalena do Pico and S. Roque do Pico and based on existing EurOtop (2007, 2018) recommendations on tolerable wave overtopping. Note that not all overtopping events are unacceptable. Some mean overtopping discharges may be tolerable and may not trigger any warning. Figure 8 provides an example of the layout generated by the HIDRALERTA system for each instant, with warning levels shown for each stretch of the structures.

The user interaction component is materialized in a Web application, in which the whole warning system is parameterized. The application is designed for use in traditional and mobile Web browsers, by adapting the information in accordance with the characteristics of the user’s device.

Web application (Figure 9) displays all the forecasted data by the system, which include the wave conditions from offshore to the foreshore, the wave overtopping discharges and the warning levels (including signals) obtained for each stretch of the structures.

Based on this information, a daily forecast report is constructed (Figure 9) and it is sent by e-mail to institutions and authorities dealing with port/coastal management in Madalena do Pico, such as Portos dos Açores and the University of Azores. Data can be directly accessed at any time through the Web platform.

Conclusions

Currently, the HIDRALERTA system is already in operation for the ports of Madalena do Pico and S. Roque do Pico, with all the necessary elements to issue real-time warnings. However, to ensure the reliability of the system, there are still some aspects of the system validation that need to be improved with the collaboration of the local authorities and the use of historical data, including modules II and IV. Through their report of dangerous situations and their perception of existing risks, a better understanding of wave overtopping and flooding impacts in local communities is then achieved.

Acknowledgements

The works related with the participation of M.I. Santos, P. Poseiro, M. Salvador and E. B. de Azevedo on the simulation of the application of HIDRALERTA System to the Port of Madalena, have been done under the framework of the ECOMARPORT (MAC/1.1b/081), with the financial support of 85% by the EU (FEDER, MAC 2014-2020), and 15% from the Regional Government of the Azores. F. V. Reis is responsible for the wave data and wave database used by the system. The works of C.J.E.M. Fortes, M.T. Reis, L.V. Pinheiro, J. Barateiro,V. Serrazina, A. Mendonça and N. Smithers have been developed under the framework of To-SEAlert, Ref. PTDC/EAM-OCE/31207/2017 and BSafe4Sea, Ref. PTDC/ECI-EGC/31090/2017. The authors also acknowledge Portos dos Açores.

References

Booij, N.; Ris, R.C. and Holthuijsen, L.H.,
1999
.
A third-generation wave model for coastal regions, Part I, Model description and validation
.
J. Geographical Res., C4
, 104,
7649
7666
.
Coeveld, E.M.; Van Gent, M.R.A. and Pozueta, B.,
2005
. Neural Network: Manual NN_OVERTOPPING2. CLASH WP8 – Report BV.
EurOtop
,
2007
.
Wave Overtopping of Sea Defences and Related Structures: Assessment Manual
. Pullen, T.; Allsop, N.W.H.; Bruce, T.; Kortenhaus, A.; Schuttrumpf, H. and Van der Meer, J.W.
EurOtop
,
2018
.
Manual on Wave Overtopping of Sea Defences and Related Structures. An overtopping manual largely based on European research, but for worldwide application
. Van der Meer, J.W.; Allsop, N.W.H.; Bruce, T.; De Rouck, J.; Kortenhaus, A.; Pullen, T.; Schüttrumpf, H.; Troch, P. and Zanuttigh, B. www.overtopping-manual.com.
Flater, D.,
1998
. XTide Manual: Harmonic Tide Clock and Tide Predictor. E.U.A.
Fortes, C.J.E.M.,
2002
.
Transformações Não Lineares de Ondas em Zonas Portuárias. Análise pelo Método dos Elementos Finitos. PhD
Thesis, IST-UL.
Lopes, P.; Poseiro, P.; Fortes, C.J.E.M.; Reis, M.T.; Garcia, T.; Azevedo, E.; Rodrigues, M.C.; Rodrigues A. and Sabino, A.,
2016
. Otimização da arquitetura computacional do sistema HIDRALERTA. 13° Congresso da Água, 7-9 March, LNEC, Lisbon.
Persson, A.,
2001
. User Guide to ECMWF Forecast Products. ECMWF.
Poseiro, P.,
2019
.
Forecast and Early Warning System for Wave Overtopping and Flooding in Coastal and Port Areas: Development of a Model and Risk Assessment. PhD
Thesis. IST-UL.
Poseiro, P.; Fortes, C.J.E.M.; Reis, M.T.; Santos, J.A.; Simões, A.; Rodrigues, C. and Azevedo, E.,
2013
. A methodology for overtopping risk assessment in port areas: Application to the port of Praia da Vitória (Azores, Portugal).
Proc. SCACR 2013
, 4-7 June,
LNEC
,
Lisbon
.
Poseiro, P.; Reis, M.T.; Fortes, C.J.E.M.; Lopes, E.M.; Sabino, A. and Rodrigues, A.,
2014
.
Desenvolvimento do sistema de previsão e alerta de inundações HIDRALERTA: Caso de estudo da Costa da Caparica
. 3as
Jornadas de Engenharia Hidrográfica
, IH, 24-26 June, IH, Lisbon. ISBN: 978-989-705-073-2.
Sabino, A.; Poseiro, P.; Rodrigues, A.; Reis, M.T.; Fortes, C.J.E.M.; Reis, R. and Araújo, J.,
2018
.
Coastal risk forecast system
.
Journal of Geographical Syst.,
20
(
2
),
159
184
. doi:
WAMDI Group
,
1988
.
The WAM Model - A third generation ocean wave prediction model
.
Journal of Physical Oceanography
,
18
,
1775
1810
. doi: