Introduction

After the devastating tsunami in 2004 in the Indian Ocean and in 2011 offshore Japan, it became evident that even though large tsunami do not happen very frequently, the implementation of effective early warning systems is crucial due to the severe impact of such events. TOAST is a software package designed precisely for this purpose. Specific tsunami warning bulletins targeted to administrative districts are generated automatically based on templates. They can be disseminated through various channels like text message, e-mail or video stream immediately after an event.

These warning bulletins are based on tsunami simulation results which are either extracted from precalculated databases or computed using a so-called on-the-fly approach. This is a unique feature of TOAST and enables it to flexibly react to unexpected events, such as earthquakes in unconsidered areas or with atypical rupture mechanism. Generating simulations adequate to the situation is done in a fully automatic way and can be complemented by manual design. Simulation results can be compared to real-time data from buoys or tide gauges even during the propagation of a tsunami. TOAST has good scalability and integrates perfectly with SeisComP to form a fully functional tsunami early warning system.

Features

arrival times

Arrival isolines on maps visualize the tsunami propagation.

forecast zones

Maps with forecast zones show the local threat levels.

mariograms

Real-time waveforms provide instantaneous sea-level information and comparison to simulations.

SSHmax

Maps show the maximum wave height across the ocean.

History

TOAST has been developed by gempa GmbH, a spin-off from the German Research Centre for Geosciences (GFZ) in Potsdam. gempa is part of the SeisComP development group. Several staff members were involved in the the German Indonesian Tsunami Early Warning System project (GITEWS) as scientists, software- or field engineers. After the implementation of the GITEWS decision support system initially used by BMKG in Jakarta it soon became obvious that the system was complex and specialized to an extent that made it hard to adapt it to other environments. This is why gempa started the development of TOAST.

In 2022 TOAST, new requirements were proposed for TOAST:

  • Allow several users concurrently working on the same incident.

  • Share simulations across workstations.

  • Enable user authorization.

  • Shield the database from direct user access.

As a consequence, TOAST was redesigned: the former standalone application (hereafter referred to as TOAST legacy) was split in three components: the TOAST server, the TOAST client and the GSS.

A description of the components is given in the next section.

TOAST architecture

TOAST consists of three components:

  • the TOAST server, technically the plugin toastd for scmaster,

  • the TOAST client (also referred to as TOAST GUI)

  • and the GSS (gempa Simulation Server).

../_images/architecture-multi.png

TOAST components.

TOAST server

The TOAST server consists of scmaster and its plugin toastd (TOAST daemon). scmaster is the central process in any SeisComP installation. It provides messaging and database write access.

The toastd plugin extends scmaster by the ability to execute TOAST specific tasks like the automatic triggering of incidents and simulations. It includes the TOAST data model which is necessary to store TOAST objects in the database.

It acts as a processor for all messages sent to its queues. The main purpose is to respond to journal entries sent from clients in order to queue and serialize concurrent operations from multiple clients.

The toastd plugin can furthermore connect to a SeisComP processing queue and create and update TOAST incidents based on configured rules automatically whenever a new SeisComP event is created or updated.

The plugin configuration also holds the configuration of the templates (bulletins), the automatic simulation profiles, the trigger criteria and others. The configuration of scmaster and also its plugins is done in the file scmaster.cfg (look for toastd in the queues).

TOAST client

The TOAST client is the graphical user interface (GUI) for the user. But is not just a GUI, as for instance the computation of threat levels based on selected simulations is done by the TOAST client. The creation of bulletins based on the templates, the visualization in the live tabs and execution of external scripts, for instance to send the bulletins to the gempa Dissemination Server GSS is also done by the TOAST client.

Configuration which is done client-side includes among others: Live Tabs, Threat level mapping configuration or Forecast zones configuration.

In general, the user workflow for the TOAST client is very similar to the one of the TOAST legacy version.

GSS

The GSS, the gempa Simulation Server is a separate process running server-side which provides pre-computed or on-the-fly simulations. It accepts requests by the TOAST server and the TOAST client. It has a REST API and can be accessed via HTTPS and telnet. In principle, it can be used by other applications than TOAST.

The simulation plugins like EasyWave have to be registered in the GSS configuration and additionally have their own configuration files. The installation and configuration of the GSS is described in its documentation.

The GSS has its own documentation.

Summary

  • Automatic incidents and simulations are triggered by the TOAST server

  • Only the TOAST server writes to toast database

  • Simulations are provided by the Simulation server GSS

  • Results are retrieved by the TOAST client from the TOAST server and the GSS

  • Connection between clients and servers can employ authorization

  • Templates are configured on the TOAST server and evaluated by the TOAST client

Application worldwide

Nowadays TOAST is operational worldwide in many institutions where fast and reliable tsunami early warning is key. Some reference installations are:

Simulation concepts

Conventional tsunami early warning systems are exclusively based on prototypes developed within the context of more or less scientific projects. Based on the actually determined earthquake parameters, a lookup in a database of precalculated earthquake scenarios and tsunami simulations is performed, enabling prediction of arrival times and amplitudes of the tsunami wave for relevant coastlines. This approach takes a limited number of standard situations into account. In case of an atypical or unexpected earthquake, these systems fail because an on the fly simulation capability is not provided and the system is not able to adapt to the situation. TOAST supports fast access to precalculated simulations by plugins.

../_images/conventional_tews.png

Conventional Tsunami Early Warning System: no simulations for unexpected scenarios are available.

In order to achieve higher flexibility, TOAST provides the approach of on-the-fly simulation of tsunami propagation. This enables TOAST to react to atypical situations, like earthquakes in supposedly inactive regions or earthquakes exceeding the maximum magnitude assumed for a region. For instance, before the 2004 tsunami, most experts wouldn’t have thought that a magnitude 9.3 earthquake in the Sumatra region would be realistic, and similarly for the 2011 Tohoku earthquake in Japan. Also the Fiordland earthquake in New Zealand in 2009 with a magnitude of 7.8 is an example for an unexpected event.

Spanning a wide range of possibilities, on-the-fly simulations are triggered automatically according to configuration options, and manually designed simulations can be added with a few clicks. Updates of earthquake source parameters can be immediately considered.

On the other hand, if a database with precalculated simulations is already available, it can easily be integrated into TOAST using customer-specific plugins.

../_images/toast_tews.png

TOAST: precalculated as well as “on-the-fly” simulations are available.

On top of that, TOAST is also able to compute worst-case-aggregations over a set of simulations. This is useful in order to get a more conservative assessment, for instance directly after a large earthquake, while the direction of its rupture propagation is still uncertain.

Tsunami generation and propagation

A tsunami is an oceanic gravity wave generated by submarine earthquakes or other geological processes such as landslides or volcanic eruptions. 90 percent of world-wide tsunami result from earthquakes. In deep ocean a tsunami may show low amplitude and be barely noticed, but it increases when propagating into shallower water and can cause devastating damage upon reaching the coast or harbors.

The largest tsunami are generated by subduction earthquakes at the boundaries of tectonic plates. These submarine events lift the seafloor along tenths to hundreds of kilometers by several meters within seconds to minutes. As the sea surface is raised along with it, waves are caused that travel outward in all directions away from the source area, much like ripples caused by throwing a rock into a pond.

A tsunami is made up of a series of very long waves. The wavelength and the period depend on the generating mechanism and the extension of the causing event. If a tsunami is caused by a large earthquake affecting a large area, its initial wavelength and period are larger. If on the other hand a tsunami is caused by a local landslide, both its initial wavelength and period are shorter. The period of a tsunami wave typically ranges from 5 to 90 minutes. The wave crests of a tsunami can be more than a thousand km long, and from a few to a hundred kilometers or more apart as they travel across the ocean. While the wavelength diminishes when moving into shallower water, the period remains constant.

The height of a tsunami from trough to crest in the open ocean may be only a few centimeters to a meter or more - again depending on the generating source. Tsunami waves in the deep water travel at high speeds over distances of thousands of kilometers and lose little energy in the process (except for geometrical spreading). The deeper the water, the higher the speed of the tsunami. When the tsunami approaches the coast and enters shallower water, the front is slowed down while the rear still has higher speed and thus the amplitude of the tsunami increases [1].

The so-called long wave equation describes the phase speed at which a tsunami travels as follows: c = \sqrt{g*H}. Here g is the gravity of Earth (g\approx 9.8 m/s^2) and H is the local water depth. The following table shows the effect of changing water depth. The first three columns apply in general to any tsunami, the last two correspond to an example.

Water depth [m]

Phase speed [m/s]

Phase speed [km/h]

Period [min]

Wave length [km]

10

10

36

10

6

50

22

80

10

13

100

31

113

10

19

500

70

252

10

42

1000

99

356

10

59

2000

140

504

10

84

3000

171

617

10

103

4000

198

713

10

119

5000

221

797

10

133

6000

242

873

10

146

The wavelength \lambda and the phase speed c determine the period T of a tsunami through the dispersion relation: \lambda=c*T. Note that both speed and wave length depend on water depth, while the period remains constant.