EXPERT CONSULTATION MEETING – TSUNAMI RISK IN EUROPE
STATUS, GAPS AND NEEDS RELATED TO EARLY WARNING SYSTEMS
GREEK– DRAFT PAPER
G.A. Papadopoulos1 and Ch. Koutitas2
1 – Institute of Geodynamics, National Observatory of Athens
2 – Department of Civil Engineering, University of Thessaloniki
1. Is there a tsunami risk in Europe?
European coastal segments were hit in the past by large, destructive tsunamis. Hundreds of both local and regional tsunami waves have been reported in documentary sources from the antiquity up to now. Other historic or pre-historic events were revealed from palaeotsunami methods based on geological observations. Tsunamis are mainly generated by submarine earthquakes while some events were reported to have been caused by volcanic eruptions and locally by landslides. Therefore, tsunami waves should not be neglected as a potential source of risk that threaten coastal communities of the European and particularly of the Mediterranean Sea. However, the frequency of tsunami occurrence is quite different in different coastal zones. The highest rate of tsunami occurrence is observed in the east Mediterranean Sea where the seismicity is also the highest in Europe. Medium rate is observed in the central and western sides of the Mediterranean Sea, while in the Marmara Sea, in the North Sea, in the Black Sea as well as in the Atlantic coasts of Europe the rate of tsunami generation is low.
2. What do we know from the past?
Data contained in The New European Tsunami Catalogue, one of the main achievements of the EU-funded GITEC (1992-95) and GITEC-TWO (1996-98) projects, indicate that more than 250 tsunami events were reported in the Mediterranean Sea in about the last 3500 years. However, the data basis is statistically complete only for the last few centuries. The largest known destructive events, assigning intensity 5 or 6 in 6-grade intensity scale, took place (1) in the Thera (Santorini) volcanic island complex, Cyclades, South Aegean Sea, Greece, in association with volcanic eruptions on c.1630 B.C., 1650A.D. and tectonic earthquakes on 09.07.1956, (2) in the Hellenic Arc on 21.07.365, 03.08.1303, ?.04.1609, and 31.01.1741, (3) in the Maliakos and Corinth Gulfs, central Greece on 426B.C. and 373B.C., (4) in the Levantine Sea on 09.07.551 and 25.11.1759 and (5) in the Messina straits-Calabria region on 06.02.1783 and 28.12.1908. The statistical completence in the last few centuries does not implies that all the historical tsunamis are known. Therefore, more research in both the historical archives and the geological field is needed in order to reveal tsunami events that remained unidentified so far and to enrich the existing data basis.
3. How to assess and predict tsunami risk?
In natural disaster science, it is generally accepted that the risk (R), that is the expected
impact of a potentially damaging or disastrous natural event on a particular region, is a convolution of three parameters: hazard (H), vulnerability (VU) and value (VA):
R = H * VU * VA (1)
where hazard (H) is a measure of the probability for the natural event to occur in a given time window, vulnerability (VU) is a measure of the degree of resistance of the anthropogenic environment (structures, population etc.) to the natural event, and value (VA) is the economic value exposed to the natural hazard. Hazard is a description of only the natural process and does not include components of the impact of it. In particular, the assessment of the tsunami hazard involves several parameters like the frequency of tsunami occurrence, the different mechanism of tsunami generation and the identification of the potential tsunamigenic sources, the propagation of the tsunami waves from the source to the threatened coastal zones as well as the inundation of the coastal zones. As for the vulnerability, it is increasing with the decrease of the degree of resistance of the anthropogenic environment. In this sense, the assessment of the tsunami risk in a particular coastal segment or region depends on the assessment of hazard, vulnerability and value: risk is an increasing function of H, VU and VA.
a) Tsunami Occurrence
In the Mediterranean Sea, the recurrence of damaging or destructive waves, that is of tsunami intensity equal to or larger than 4, 5 and 6, in 6-grade intensity scale, is on the order of 10, 30 and 90 years, respectively. The most frequent tsunami generation is noted in the east Mediterranean Sea and particularly in the Western and Eastern segments of the Hellenic arc as well as in the Corinth Gulf, Greece. Infrequent but large or strong events were described in Cyclades, South Aegean Sea, the Messina straits-Calabria in South Italy, as well as in the coasts of Algeria (Alboran Sea), Cyprus, Israel, Lebanon and Palestine (Levantine Sea), and of Marmara Sea. In the area of Greece, the recurrence of tsunami intensity equal to or larger than 4, 5 and 6 is on the order of 15, 60 and 250 years, respectively.
b) Tsunami Generation Mechanisms
In Greece and in general in the Mediterranean Sea the earthquake activity is the main cause of tsunami waves. The earthquakes capable to generate tsunamis, however, have some particular properties: they are strong (magnitude around 6 or larger), shallow (focal depth less than about 40km) events associated with submarine or near-coast dip-slip fault motions. These properties make a very important knowledge background for the formulation of warning algorithms in real time conditions. However, submarine volcanic eruptions and coastal or submarine landslides are additional mechanisms that may generate tsunamis from time to time. The Thera (Santorini) active volcano in the South Aegean Sea is a typical source of large volcanigenic tsunamis. On the other hand, the Corinth Gulf, Central Greece, is a typical area where the high rate of earthquake occurrence is combined with the frequent occurrence of seismic or aseismic landslides and result in a high rate of local but powerfull tsunamis. Since Corinth Gulf is a closed sea area the design of instrumental tsunami warning systems should take into account the local nature of the phenomena and the very short travel times of the waves.
c) Identification of Potential Tsunamigenic Sources
The identification of potential tsunamigenic areas constitutes a corner-stone in the effort to mitigate tsunami risk on the basis of instrumental warning systems and other actions. The past tsunamicity of the Mediterranean Sea is rich enough to provide data for the construction of a reliable tsunami zonation map. On the basis of the past tsunami history sixteen main tsunamigenic zones have been identified (Fig. 1 and Table 1). Most of them are located in the east Mediterranean Sea and particularly in Greece. This is due to that Greece, being located in the front of convergence of the African and Eurasian lithospheric plates, is characterized by the highest seismicity in the western Eurasia, that is from Caucasus to the Atlantic Ocean and from Africa to the north pole. However, additional sources that are potentially tsunamigenic may have not been identified and, therefore, more research effort is needed towards this aim. Submarine active fault segments are of particular interest as for their possible association with future strong, regional tsunamis. Coastal and submarine landsliding masses also bear some interest as possible agents of local tsunami generation.
d) Tsunami Propagation Properties
The tsunami travel timesin the Mediterranean Sea are relatively short due to that most of the tsunamigenic sources are lying close to the coasts. Therefore, the expected travel times may range from a few minutes up to about two hours at maximum. In addition, the tsunami wave attenuation in the Mediterranean Sea is much stronger than that in the Pacific Ocean. These two properties imply that only local and regional tsunamis are expected to be observed in the Mediterranean Sea. On the contrary, no distant or transoceanic tsunamis are expected and never such tsunamis were observed in the past although several megatsunamis were produced in historical times as well as during the 20th century. As an instance, reliable documentary sources leave no doubt that the large 365 and 1303 tsunamis, that were caused by large earthquakes in the Hellenic Arc, propagated towards remote places of the east Mediterranean Sea, like Alexandria, North Egypt on 365 and Akko, Israel, on 1303. Similarly, the 9 July 1956 tsunami, generated by a magnitude 7.5 tectonic earthquake in the South Aegean Sea and inundated many places of the Aegean Sea with heights up to at least 15m. However, the tsunami did not propagated to more remote places. The three tsunamis mentioned were regional, megatsunamis given that they inundated many places of the east Mediterranean Sea with high wave amplitudes. However, they did not propagated to the western side of the Mediterranean Sea, that is they were not transoceanic tsunamis. The propagation properties of the Mediterranean Sea tsunamis indicate the importance for developing mainly local and regional tsunami warning systems.
e) Inundation Maps and Tsunami Intensity
The technology of the tsunami numerical simulation provides important possibilities for the determination of the zones expected to be inundated by future tsunami waves. Inundation mapping may include the area of flooding as well as the wave height and direction of water flow. However, inundation maps do not describe the expected impact of the wave. Therefore, there is need to develop further this technology so that to translate expected tsunami inundation to expected tsunami impact. One of the most practically applicable parameter to describe the tsunami impact is the tsunami intensity. Traditional tsunami intensity scales developed since the 20’s are 6-grade scales. However, they are not detailed and sensitive enough to describe adequately the several components of the tsunami impact. More recently, a 12-grade scale was introduced by following the long seismological tradition in this field. The new scale needs further tests and calibration on the basis of actual tsunami impact data.
f) Identification of Coastal Vulnerabilities
No standard methodology has been generally adopted for the assessment of the vulnerability of the anthropogenic environment to the impact of the natural hazards including the tsunami hazard. Parameters that may be used include types and properties of the engineering structures, population density, the land use/land cover peculiarities and the existence or not of critical facilities. For example, in a pilot study developed as part of the GITEC and GITEC-TWO projects for the tsunami risk assessment in a coastal segment of Crete island, Greece, such parameters were semi-quantitatively introduced for the vulnerability assessment. The timing of the event, that is the season, the day and the time of occurrence, make a very important function for the vulnerability assessment. There is no doubt that there is important room for improvements of the methodology mainly towards making it more quantitative.
g) Evaluation of Potential Economic and Social Losses
The quantitative evaluation of the expected economic and social losses from natural hazards is a very difficult task. Again no standard methodologies have been generally adopted. A direct measure of the economic impact could be the local GNP of the area considered. Other parameters that could be taken into account may include the population density, the road network and other infrastructures, the critical facilities and again the timing of the event.
4. What exists for the moment in Greece
4.1 Instrumental Networks
4.1.1 Real-Time Seismograph Networks
In Greece, a well-established national seismograph system consisting from about 30 permanent, on-line stations is operated on a 7/24 basis by the Institute of Geodynamics, National Observatory of Athens (NOAGI). All the stations are equipped with BB seismometers. The seismic signals recorded by the stations are transmitted to NOAGI in real-time via dedicated telephone lines. Scientific and technical staff is on duty on a 7/24 basis and announce publicly moderate and strong earthquakes in routine times ranging between 10 and 15 minutes by processing manually the data. The earthquake announcement is also transmitted directly to governmental bodies like the General Secretary of Civil Protection and the Earthquake Planning and Protection Organization. This system is capable to transmit simultaneously and in real-time up to 13 different geophysical signals via the same telephone lines. Recently, GPS signals started to be transmitted from some of the stations. In addition, following international standards a system for the automatic process of the data is under development in NOAGI. It is expected that in a time prospect of about two years from now this system may provide the possibility to routinely announce earthquakes within about 5 minutes from their generation. NOAGI also operates seismic stations of international networks, like MEDNET, under the supervision of INGV, Italy, and GEOFON under the supervision of GFZ, Potsdam, Germany. Moreover, NOAGI has participated in research projects which incorporated seismic monitoring with Ocean Bottom Seismographs (OBS) in several seismogenic places of Greece.
As a part of GITEC-TWO, NOAGI developed an experimental tsunami warning system in the South Aegean Sea consisting of five additional, temporary digital seismographs and two tide-gauges connected on-line with NOAGI (see more details below).
In Greece, other academic institutes maintain local seismograph networks. However, they either have no personnel on duty for the monitoring of their area or their capability is only limited. Currently a new project was submitted with the aim to unify the local networks with the NOAGI national network by developing special software which is expected to make compatible the presently different networks. It is expected that the government will approve this important 3-year project.
4.1.2 Tide-Gauge Network
A system of about 20 analog tide-gauge stations is operated by the Greek Navy mainly for oceanographic purposes related to national security aspects. No data transmission is made by this system. As a part of GITEC-TWO, NOAGI developed an experimental tsunami warning system in the South Aegean Sea consisting of five additional, digital seismographs and two tide-gauges connected on-line with NOAGI (see more details below).
4.1.3 Experimental Tsunami Warning System
On 1998, as a part of GITEC-TWO, NOAGI developed an experimental tsunami warning system in the South Aegean Sea consisting of two subsystems: one seismograph subsystem, incorporating five digital seismographs additional to those of the national system, and one sea-level changes subsystem incorporating two new, digital tide-gauges equipped with pressure meters. Both subsystems were connected with NOAGI with dedicated telephone lines that transmitted in real time the seismic and tide-gauge signals. The experiment proved successful and important know-how was obtained.
4.2 Tsunami Hazard Assessment
4.2.1 Tsunami Simulation
The main research group involved in tsunami modeling in Greece is that of the Dept. of Civil Engineering, University of Thessaloniki, which in collaboration with NOAGI simulated several tsunamis observed in Greece. Other groups (e.g. from Japan and Norway) also tried to simulate Greek tsunamis. The most important result is that the algorithms used reproduce well-enough the observed wave heights in the near-field but usually they fail to reproduce wave heights observed in the far-field. Therefore, it is of great importance to investigate further the factors that control this inconsistency. The most important of them include source properties, bathymetry and simulation techniques.
4.2.2 Tsunami Hazard Assessment – Sources
Although the historical record of tsunamis is incomplete, the existing data basis is adequate enough to define the main tsunamigenic zones not only in Greece but in the Mediterranean Sea in general (Fig. 1 and Table 1). However, what remains unknown is why only some of the characteristic earthquakes occurring in the same seismic fault or plate boundary segment produce tsunamis while others do not. For example, in the eastern Hellenic Arc, it is historically and geologically well-documented that in the Rhodes island region from the five large earthquakes that took place in the last six centuries only three were tsunamigenic while the other two were not. This is of extreme importance for the tsunami hazard assessment and, therefore, further research for understanding better the properties of the seismic tsunamigenic sources is needed.
4.2.3 Tsunami Hazard Assessment – Recurrence
The recurrence of large tsunamis is only roughly estimated in a few tsunamigenic regions. This is due to that the historical record is incomplete. Important improvement is expected from further research in the historical documentation but also from the application of the palaeotsunami method for the identification of past tsunamis from geological field techniques combined with analytical laboratory methodologies. The palaeotsunami method has been applied very successfully in the Aegean Sea, the Corinth Gulf and the Marmara Sea by NOAGI in collaboration with Japanese and Turkish institutes. About ten strong palaotsunami events were identified in the last ten years.
4.2.4 Vulnerable Areas Identification
In Greece the most tsunami-vulnerable coastal areas are those that are highly involved in tourist activities mainly in the broad region of the South Aegean Sea (e.g. Cyclades island complex, Crete island, Dodecanese island complex), the south part of the Ionian Sea (west and south Peloponnese and nearby islands) and the Corinth Gulf, central Greece.