Fast Ferry Traffic As a Qualitatively New Forcing Factor Of

Environmental Fluid Mechanics (2005) 5: 293–323 © Springer 2005 DOI 10.1007/s10652-005-5226-1

Fast Ferry Trafﬁc as a Qualitatively New Forcing Factor of Environmental Processes in Non-Tidal Sea Areas: A Case Study in Tallinn Bay, Baltic Sea

TARMO SOOMERE Institute of Cybernetics, Tallinn University of Technology, Akadeemia tee 21, 12618 Tallinn, Estonia (E-mail: [email protected])

Received 3 February 2004; accepted in revised form 8 April 2005 Abstract. The impact of wake wash from high-speed ferries on the coastal environment in non-tidal seas is analysed in terms of wave energy and power, and properties of the largest waves. Shown is that hydrodynamic loads caused by heavy high-speed traffic may play a decisive role not only in low-energy coasts but also in certain areas with high wind wave activity. For example, ship-generated waves form, at least, about 5–8% from the total wave energy and about 18–35% from the wave power in the coastal areas of Tallinn Bay exposed to dominating winds. The periods of wake waves from high-speed ships frequently are much larger than dominating periods of wind waves. The leading waves typically have a height of about 1 m and a period of 10–15 s. Such waves extremely seldom occur in natural conditions in many regions of semi-enclosed seas. They cause unusually high hydrodynamic loads in the deeper part of the nearshore. The fast ferry traffic thus is a qualitatively new forcing component of vital impact on the local ecosystem. It is demonstrated that wakes from high-speed ferries may trigger considerable changes of the existing balance of coastal processes. Owing to their low decay rates combined with their exceptional compactness after crossing many kilometres of the sea surface, such wakes may cause considerable remote impact of the ship traffic. This feature has to be addressed in the analysis of the impact of harbours and associated ship traffic in the neighbourhood of vulnerable areas.

Key words: Baltic Sea, high-speed ships, Tallinn Bay, wake wash, wave measurements

1. Introduction Marine and coastal resources play a major role in sustaining the economic and social development of society. Today, about 80% of all international trade is carried by sea. According to some estimates, in the nearest future, by the year 2020, about 75% of the world’s population will live within 60 km of sea coasts and estuaries ([2], pp. 29–30). The continuous increase 294 TARMO SOOMERE of population in the coastal areas and the ever increasing density of marine transport combined with a limited number of convenient anchorage places results in an extreme concentration of both cargo and passenger traffic in certain sea areas. The concerns related to heavy ship traffic are traditionally associated with possible accidents (ship collisions or grounding, technical and navigation problems caused by severe weather or human errors, etc.) that may lead to either loss of lives or property, or to environmental pollution. However, these concerns have been and are being effectively managed by international shipping and harbour communities. When considering safety aspects of water surface transport, traditionally it is assumed that the risks are localised within a small area around the ship. For example, a ship acci- dent occurs at a certain point and possible oil contamination is transported relatively slowly owing to winds and currents. The continuing introduction of evermore faster ship services during the last two decades has created another set of major worries. They consist in (i) a massive growth of exhaust emissions per passenger mile (compared, for example, to passenger cars or buses), (ii) a great increase of external noise, and, last but the largest, (iii) the waves generated by large high-speed ships [68]. These by-products of the fast ship traffic are no more located in small areas. In the contrary, they may travel much faster than the ship itself (in particular, the ship-generated noise), may become a part of global troubles (for example, exhaust emissions, Hobbs et al. [20]), or may result in violent energy concentration in remote sea areas [17]. The importance of the contribution of the ship traffic to the local hydrodynamic activity in rivers, inland channels and narrow straits has been recognised for a long time. Ship wakes can essentially contribute to the shoreline erosion [3, 14, 43], cause an extensive erosion and resuspension of bottom sediments [33], trigger ecological disturbance and cause harm to the aquatic wildlife [1, 38], to mention a few studies. Also, damage to structures and archaeological sites, and safety problems for navigation and for users of the beach and nearshore may arise (see Parnell and Kofoed-Hansen [45] and references therein). These aspects are not related to particular types of ships. They appear in many places all over the world and for a variety of ships, and only represent a locally important contribution of shipping in the hydrodynamic activity. Historically, these aspects have been studied perhaps in the greatest detail for the archipelagos of the Baltic Sea ([50–53]; for an overview see Madekivi [40] and references therein). The introduction of large high-speed vessels extended the above threats from inland waterways, particularly narrow straits, and archipelagos to much larger confined sea areas with low natural wave and tide activity [16, 45]. These vessels have a high ratio of propulsion power to vessel dis- placement and are able to sail at much larger speeds than conventional FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 295 ships. They are able to cross the two firm ‘barriers’ that restrict the speed of conventional vessels [66]. First, the hump speed occurs when the half of the length of ship-generated waves travelling in the sailing direction is close to the vessels length. Sailing at this speed means that the ship is continu- ously moving ‘upwards’ along the slope of her wake. Second, the critical speed corresponds to the maximum phase speed of surface waves in finite depths. At this speed the ship generates a wave system that is non-dispersive and may remain compact for a long time. The crossing of either of the ‘barriers’ requires much more energy than sailing at lower speeds, and is accompanied by a fast increase of wave resis- tance [66]. It results in much higher wake waves and, accordingly, in much higher wave loads on the coastal environment near the ship lane. Large high-speed craft sailing at specific speeds in shallow areas may produce fundamentally different wave systems than conventional vessels [4, 7, 36, 57]. Ship-generated solitonic disturbances usually have moderate height [70] and possess no immediate danger to the people or to the beach environment. However, superposition of solitonic waves (occurring, for example, when by two systems of ship waves simultaneously reach a shallow area) may lead to a fourfold amplification of the local surface elevation [46]. Depression areas accompanied with the generation of solitary precursors may be responsible for a drastic dropdown of water surface in small harbours adjacent to narrow waterways [13]. The increase in wave heights at large speeds is accompanied by an increase of the typical periods and lengths of single waves. High and long waves from high-speed ships may create explicit danger to the safety of people [17] and their property, and other vessels [29, 31]. This situation was recognised as inadmissible in many countries (in particular, in Denmark, the United States, and New Zealand) and necessary restrictions were imposed [45, 68]. Today, the experience in managing of wake wash from high-speed vessels in confined sea areas has been generalised in a sys- tematic manner. A description of the accompanying adverse effects on the local ecosystem and possibilities of their mitigation has been presented in [16, 45]. On the one hand, it has become clear that heavy ship traffic has a great damaging potential in the vicinity of waterways and the adjacent shoreline, in particular, in areas that are sheltered from large wind waves (such as wetlands, low-energy coasts, etc., see Bourne [3], Schoellhamer [57]). On the other hand, it is generally believed that ship wakes are negligible and that their effect is sporadic in coastal areas that are open seawards and where natural waves are frequently much higher than the wakes [38]. This assumption is true for coasts exposed to high tides or large wind waves indeed. 296 TARMO SOOMERE

The purpose of the current paper is to demonstrate that wake wash may be of particular importance in certain parts of coasts that are exposed to significant natural hydrodynamic loads and that are already subject to intense beach erosion in natural conditions. The reason is a combination of specific features of (i) the existing hydrodynamic loads (that are restricted, for example, to a particular direction or to a certain frequency interval) and (ii) the coastal environment (that has reached a near-equilibrium stage of its evolution) with (iii) particularly high anthropogenic wave loads in a different frequency range. It is first demonstrated that the share of ship waves in the total wave activity may be remarkably high in the vicinity of open sea waterways even in terms of wave energy. The description of properties of ship waves at different sailing regimes as well as their changes in the far field and in the coastal areas, and relevant references can be found in [16, 45, 66]. The current study is focussed on the occurrence probability and basic properties of waves of different origin, and on energy density and its flux (wave power) in the coastal zone owing to wind waves and ship wakes. The local background of Tallinn Bay (the Baltic Sea) is introduced and the set-up of field experiments is described in Section 2. The basic properties of ship wakes and wind waves in the area in question are discussed in Sections 3 and 4. A comparison of the energy and power of waves of different origin shows that the annual mean energy of ship-generated waves is about 10% from the total wave energy at a distance of a few km from the main ship lane (Section 5). This is not unusual in itself; for example, [57] have identified much larger portions of ship-generated waves in the total wave activity. The specific feature of the coasts in question is that they are open to a part of dominating winds, may experience significant wave heights 3–4 m during strong storms, and undergo extensive coastal erosion owing to natural wave activity. An energy-based comparison of waves of different origin is equivalent to a comparison of relevant wave heights. The comparison of wave power (that characterises the rate of wave energy supply to some area) shows much larger share of ship waves in the total wave activity. This disparity reflects the fact that a part of ship waves usually become longer when the ship’s speed increases. It is shown in Section 6 that the energy of waves from large fast ferries is mostly concentrated in components with relatively large periods that greatly exceed the typical periods of natural waves in the area in question. It is well known that low-energy coasts are vulnerable with respect to changes of the wave activity. The presented results extend this threat to a part of high-energy coasts. In particular, the difference in periods of natural and anthropogenic waves results in greatly different properties of wave- bottom interaction and suggests that the heavy high-speed ship traffic may FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 297 add a qualitatively new forcing factor to the local ecosystem in specific sea areas. This conclusion mainly concerns micro-tidal or non-tidal seas (such as the Azov Sea or the Baltic Sea) where near-bottom hydrodynamic activity is governed by the local windseas and currents. Specific wind regime and complex geometry of some sea areas (for example, Tallinn Bay in the Baltic Sea) may additionally enhance the influence of high-speed traffic. In specific sea areas, the new component of wave activity – highly energetic long waves – may considerably change the overall sediment redistribution pattern. In unfavourable cases, it may greatly accelerate coastal erosion owing to a specific multi-stage sediment transport mechanism sketched in Section 7. Generic implications of the contribution of wake waves from intense high- speed traffic are discussed in Section 8.

2. Study Area and Set-Up of Field Experiments The Baltic Sea (Figure 1a) is one of the largest brackish water bodies in the world. Its hydrophysical conditions vary largely both in time and space [72], thus allowing only the particularly strong species to survive. It is located at relatively high latitudes with severe climate and each year a large part of it is covered by ice. Many species here are of the border of their area of spreading. As a result, the local marine environment is particularly vulnerable with respect to any changes of the external forcing. Tallinn Bay is an example of a semi-open sea region (ca. 10 × 20 km) in the central part of the Gulf of Finland, Baltic Sea. It hosts extremely

Figure 1. (a) Location of Tallinn Bay, (b) the mounting scheme of the wave recorder (not to scale). 298 TARMO SOOMERE heavy fast ferry traffic since 2000. A variety of different types of high-speed ships cross the gulf nearly 70 times daily during the high season. The water depth along the ship lane is such that large high-speed ships frequently sail in the near-critical1 velocity range. The resulting wave systems frequently cause violent plunging breakers in certain areas located at a distance of many kilometres from the ship lane, creating appearance of greatly induced overall wave load and serving as a potential hazard to moored vessels, danger to the general public using beaches or small craft, and supposedly intensifying coastal or bank erosion [60]. A sketch of some aspects of the influence of ship wakes based on an intense field and theoretical study in 2001–2002 is presented in [65]. A large number of various devices to measure wave characteristics have been used in different studies of ship wakes [16]. The pressure sensors are relatively inexpensive, reliable and easy to operate. The pressure data well represents wave periods and heights but not necessarily the water surface time series and the wave profile. The sensors may fail to register short waves (because pressure fluctuations caused by short waves fast decay with the increase of the depth) unless mounted at small depths not exceeding 2–3 m (Figure 1b). Since the periods of waves from high-speed ships typically vary from 3 to 40 s [29, 30], pressure sensors can be used for adequate description of their basic parameters. Three devices were used: pressure sensor SD204 (SAIV Ltd., Norway), wave recorder SBE26 (Sea-Bird Elec- tronic, USA) and a domestic water level recorder (PTR Group, Tallinn, Estonia) with an extended memory capacity. Although several witnesses have claimed that ship waves have been as high as 4 m at specific sites of Tallinn Bay, a series of preliminary experiments in 2001 showed that the ship wave heights at the depths of 5–10 m apparently did not exceed 2 m and generally were less than 1 m [60]. Yet the actual possibility of occurring of up to 2 m high waves suggests that the wave properties should be measured at somewhat larger depths than the reference depth 3 m for the wake height criterion in Denmark and New Zealand [32, 45]. The highest ship-generated waves may already be at the breaking stage at this depth and their heights may be severely underestimated from the pressure time series.

1 Generally,√ navigational speeds are distinguished according to the depth Froude number Fd = ν gh, that is the ratio of the ship speed ν and the maximum phase speed of surface gravity waves, where h is water depth√ and g is acceleration of gravity, or according to the length Froude number Fl = ν gL, where L is the length of the ship’s waterline. Operating at speeds resulting Fd <1 is deﬁned as subcritical, at Fd >1 as supercritical and at Fd =1 as critical. There is a relatively wide transcritical speed range 0.84

Figure 2. Tallinn–Helsinki ship lane and wave measurement sites in the coastal area of Tallinn Bay. Shown are depth isolines at 2, 5, 10, 20 and 50 m.

The wave properties were measured in areas with the depth of 5–7 m in different parts of Tallinn Bay and at different distances (2–8 km) from the ship lane. The measurement sites (Figure 2) were chosen near the coastal areas that apparently were the most vulnerable with respect to the ship waves according to Kask et al. [27]. Owing to extremely heavy regular and leisure trafﬁc in the study area the devices were mounted in shallow areas not used by vessels with draught exceeding 1 m. An overview of the measurement procedure and the description of details of both natural waves and ship wakes can be found in [64]. The analysis of the pressure data follows the classical procedure of establishing the surface wave properties from the pressure time series [56, 67]. The data of each record (with duration of 4–34 min, separated by pauses of 30–60 s) is ﬁrst de-meaned and de-trended. The mean is used for calculating the instrument depth. A Hanning window, followed by a multi- plication by a proper scale factor, is applied to suppress the spectral leak- age that occurs when the data contain a periodic signal with a frequency that does not coincide with any of the exact frequencies of the Fast Fou- rier Transform (FFT) for the particular record. 300 TARMO SOOMERE

Further, the FFT of the records, with the number of points equal to the number of pressure snapshots in each record, is performed. To prevent measurement noise from being mapped into unrealistic short wave heights, the amplitudes of waves with the frequencies exceeding a certain threshold are set to zero. The truncated spectral density of the pressure is converted to the energy spectrum of surface waves with the use of the attenuation rates of the subsurface pressure for the given sensor position, water depth, and wavelength. The estimates of water surface time series, obtained by inverting of the described procedure, not necessarily represent the actual water surface time series and only a few properties of this series are used in the further analysis. Wind wave field is usually characterised by the significant wave height Hs (the mean height of one third of the highest waves, IAHR [22]). This measure is not applicable for ship wakes, in particular, for their long-wave part that consists of transient waves. Their height is estimated using the classical zero-upcrossing method [22] applied for the smoothed pressure time series. A de-meaned and de-trended pressure record is filtered with the use of a low-pass filter. Typically, the cut-off frequency was set to 0.2 Hz, equivalently, waves with periods < 5 s were excluded. The resulting pressure time series contains a very small amount of wind waves (see below). The parameters of the remaining pressure disturbances are estimated individ- ually for each single pressure wave. The pressure attenuation rate for the particular wave period and the sensor position, and water depth is applied to convert the pressure wave amplitude to the surface wave height. Doing so implicitly presumes that a sinusoidal wave with the period coinciding with that of the single pressure wave contains all long-wave energy between the relevant zero-crossings of the pressure disturbance. However, an error caused by this assumption is relatively small because the attenuation rate varies insignificantly for the typical period range of long ship waves in the measurement conditions. The longer the waves, the smaller is the error, because subsurface pressure in very long waves in shallow water simply follows the behaviour of the water surface [41].

3. Typical Properties of Ship Wakes The majority of conventional passenger and cargo ships operating in Tallinn Bay excite wakes that are practically indistinguishable from the natural background in the coastal zone even in calm days. They sail at moderate length and depth Froude numbers, and excite waves that have typical heights of 20–30 cm (a maximum height of 40 cm) and periods of 3–4.5 s at a distance of about 2 km from the ship lane. Hydrofoils are capable to sail at near-critical speeds in a large part of the bay [64] but their wakes are FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 301 even smaller. Generally, their height does not exceed 10 cm in the coastal area of the bay whereas the wave periods are 3–4 s. The fast ferries (four high-speed catamarans and one monohull vessel capable of carrying cars and operating already in 2001–2002, and another monohull operating from 2003) frequently create wakes that are clearly dis- tinguishable from the wind wave field even in moderate and strong wind conditions [64]. Below we consider mostly waves generated by these ships and call them ship waves. Technically, the listed vessels may sail in the transcritical velocity range between Tallinn passenger harbour and Aegna jetty. A part of the resulting wave system disperses slowly and may remain more or less compact for a long time [70]. A wake of a fast ferry operating at high speeds (both in sub- and supercritical regime) typically consists of two or three wave groups (Figure 3; cf. [19, 29–32, 70]). It arrives into the measurement sites of the eastern coast of Tallinn Bay (located at a distance of 2–3.5 km from the ship route, Figure 2) about 5–10 min after the ship has passed, and lasts about 10 min. The first group contains the longest waves (periods 15–9 s; the longer waves arriving first). The second group arrives a few minutes later and usually consists of waves with periods 9–7 s. An underlying swell wave with periods exceeding 10 s and probably representing either transverse waves or open sea swell (cf. Kirk McClure Morton [29]) was not identified in Tallinn Bay. The third wave group of highly monochromatic relatively short waves with periods of 3–4 s and with a typical duration of less than one minute was recognised in a small number of recordings. These groups usually are smaller than other ship waves in the eastern part of the bay. However, they are easily identifiable in the spectral representation as a sharp and narrow

Figure 3. Pressure fluctuations caused by the wake of SuperSeaCat IV near the western coast of Aegna. The first wave group has the maximum height of 45 cm and the second group of 25 cm. The third group is the highest (52 cm). The significant height of the natural wave background is about 30 cm (From [64]). 302 TARMO SOOMERE peak provided a proper part of a record is selected. Their components are much shorter than the leading waves, thus their interaction with the seabed is also much weaker. Owing to a fixed celerity of its components, the group remains compact during a long time also in deep water where other parts of the wake experience strong dispersion. As a result, the third group well maintains its height and frequently contains the highest waves of the wake in remote areas. It was occasionally found severely high (about 70 cm) as far as 8–10 km from the ship lane in the remote coastal zone of Naissaar where it arrived more than one hour after the leading waves [64]. It is well known that ships sailing at transcritical speeds frequently excite solitonic wake components that propagate ahead of the ship and are called precursors (see [36] and bibliography therein). They resemble single moving water elevations and are usually interpreted as solutions to the Korteweg-de Vries equation for one-dimensional waves ([35], chapter 12) or the Kadomtsev–Petviashvili equation for the weakly two-dimensional case [18]. The third group also exhibits solitonic properties. However, it is basically different from precursors, because it is highly oscillatory and resembles an envelope soliton [4]. The non-linear Schrödinger equation [15] might be a proper tool for its analysis. The highest individual waves of wake patterns occur near Aegna jetty. This area may be hit by large bow waves generated by ships sailing at transcritical speeds [36]. The number of wave crests with the heights close

0.8

0.6

0.4

Maximum wave height (m) 0.2

0 08:00 10:00 12:00 14:00 16:00 18:00 Time on 14 April 2002 (hours:minutes) Figure 4. The maximum height of the long wave components (periods >5 s, bold line, restored from the ﬁltered pressure record with the use of the zero-upcrossing method) and of the whole wake of fast ferries (dotted line, estimated from the water surface time series) near Aegna jetty in a relatively deep water (6.7 m). The tempo- ral resolution is 5 min. The signiﬁcant height of wind waves was <10 cm and the height of long-period (>5 s) natural wave components was a few centimeter. The data represent the beginning of the navigation season when the number of depar- tures of high speed ships was less than in the high season. The peaks at 10:20, 13:10 and 15:10 are caused by simultaneous passing of two or three vessels [64]. FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 303 to 1 m may be many tens per day in this area (Figure 4). The highest waves from individual wake patterns, however, are moderate and in most cases do not exceed 60–80 cm at other sites of the eastern part of Tallinn Bay about 2–3 km from the ship lane. In more remote areas (Naissaar, 8–10 km from the ship lane) the maximum and typical heights of the leading waves are 47 and 20–30 cm, respectively.

4. Natural Wave Regime For a proper comparison of the impact of waves of different origin one needs a reliable picture about local wave regime. For that purpose, his- torical wave atlases such as Rzheplinsky and Brekhovskikh [54] are inap- propriate because of their insufficient resolution. A valuable experimental data set collected in 1974–1980 [44] only describes wave properties during the relatively calm spring and summer seasons. Although visual wave measurements at the Tallinn Harbour apparently extend back to Novem- ber, 1805 (R. Vahter, personal communication, 2002), they describe properly wave properties neither in the open part of the bay nor in other parts of the coast [44]. Contemporary wave measurements in the central part of the Gulf of Finland [26, 47] cannot be directly extended to Tallinn Bay area because of a very specific combination of geometry of the bay and wind regime in this area. Wave measurements in the bay itself have been performed sporadically only starting from 2001 [64]. Owing to the lack of experimental data, the comparisons below rely on the numerically modelled wave climate of Tallinn Bay. The average and extreme properties of wind waves are estimated on the basis of a simplified scheme for long-term wave hindcast with the use of a high-resolution multi- nested version of the WAM model. This model, although constructed for open ocean conditions where relatively sparse spatial grid and large time step can be used for an adequate representation of the wave field [34], gives good results in the Baltic Proper provided the model resolution and the wind information is correct. Owing to the complex geometry of the area in question (Figure 2), a triple nested model is used. The grid step of the innermost model is 1/4 along latitudes and 1/2 along longitudes (about 1/4 nm). In each sea point, 600 spectrum components (24 evenly spaced directions and 25 frequencies ranging from 0.042 to 0.41 Hz with an incre- ment of 1.1) were calculated. An extended frequency range 0.042–2.08 Hz (42 evenly spaced frequencies) was used as occasion requires in order to correctly represent wave growth in low wind conditions after calm situa- tions. Since waves are relatively short in Tallinn Bay, the wave fields pre- dicted by this model are reliable until the depth of about 5 m and as close to the coast as 200–300 m [59]. 304 TARMO SOOMERE

The basic idea of speeding up the computations consists in reducing long- term calculations of sea state to an analysis of a cluster of wave field maps precomputed with the use of single-point wind data. To the first approxi- mation, it is assumed that an instant wave field in Tallinn Bay is a function of the instantaneous wind speed and direction, and (for a small number of directions) of their persistence. This technique is justified provided wave fields rapidly become saturated and have a relatively short memory of wind history in the area of interest, and that remote wind conditions insignificantly contribute to the local wave field. These assumptions are justified in Tallinn Bay for most of the time (about 99% according to wind data from a measurement site in the open part of the Gulf of Finland) but frequently not applicable for larger sea areas, for example, for the Baltic Proper. If relevant, they make it possible to split the long-term wave calculations into a number of short independent sections corresponding to steady and homogeneous wind conditions. Details of the technique and its verification can be found in [59]. Since wind data from measurement sites near Tallinn do not adequately represent open sea wind conditions [28], the wave model is forced with the wind data at observation sites not affected by the shore [59]. The presence of ice in both the Gulf of Finland and Tallinn Bay is ignored. Doing this leads to a certain bias of the results, because the mean number of ice days is from 70 to 80 annually [8]. Statistically, the ice cover damps wind waves either partially or totally during the most windy winter season [42]. There- fore, the computed annual mean parameters of wind waves in Tallinn Bay are somewhat overestimated and represent average wave properties during the years with no extensive ice cover. The wind regime in this area as well as in the whole Baltic Sea basin is strongly anisotropic [42, 58, 62]. The most probable wind direction in the Gulf of Finland is south-west. Moderate and strong winds have a secondary maximum in north-west. During certain seasons, eastern winds blowing along the axis of the gulf may be the strongest [63]. Tallinn Bay is mostly well sheltered from winds and waves coming from those directions. South- east winds are infrequent and weak. Only western winds may excite high waves in this area. As a result, the local wave climate is relatively mild compared with the open part of the Gulf of Finland and with the sea areas adjacent to Tallinn Bay. The significant wave height exceeds 0.5–0.75 m in the bay only with a probability of 10% and 1.0–1.5 m with a probability of 1% (Figure 5). An important role in forming of the local wave climate play numerous shallow areas (that effectively damp waves coming from certain directions) located at the entrances of the bay [59]. However, very high waves occasionally appear in Tallinn Bay. The reason is that the angular structure of strong winds in the Gulf of Finland does not match the structure of all winds. Strong gales at times blow from FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 305

Figure 5. Distributions of signiﬁcant wave heights occurring with the exceedance probability of 10% (a) and 1% (b) during the years with no ice cover.

Figure 6. Distribution of the 1-year return values of the signiﬁcant wave height.

directions where winds generally are infrequent [63]. This feature causes relatively large 1-year return values of significant wave heights (Figure 6). Very strong winds at times blow from north-west. Tallinn Bay is fully open to this direction and extreme north-west winds may excite very high waves in its certain parts. This feature well explains why extreme significant wave heights in the bay are comparable with those in the open part of the Gulf of Finland, and why most of the coasts of Tallinn Bay show features of intense erosion and are considered as high-energy coasts [27, 39]. The significant wave height exceeds 2 m each year and may reach 4 m in extreme NNW storms in the central part of the bay (Figure 7). The significant wave height in the regions Gulf of Finland adjacent to Tallinn Bay slightly exceeded 5 m during the hindcasted storm. These results well match the measurements by the Finnish Institute of Marine Research [48]. 306 TARMO SOOMERE

Figure 7. Signiﬁcant wave height in Tallinn Bay during an extreme storm on 15.11.2002.

The average wave properties have significant seasonal and spatial variability. The mean wave energy during the relatively windy autumn and winter season is about twice as high as during the calm spring and summer season. In specific coastal areas both wave energy and extreme wave heights are much less than in the centre of the bay. The most important property of the local wave regime in the context of the current study is the moderate dominating period of wind waves (cf. [44, 64]). It originates from the limited fetch length. The fetch is maxi- mally 300 km for certain directions (but, statistically, no long-lasting strong storms occur from these directions) and 50–100 km for the directions of the strongest storms. In weak wind conditions (≤ 5 m/s) the significant wave height Hs ≤ 25 cm whereas the typical wave periods are 1–2 s. In moderate wind conditions (6–8 m/s) the wave height is about 50–60 cm, and is reached within a couple of hours. The dominating wave period may vary from 2–3 s to 3–4 s depending on the duration of the wind. If moderate wind blows many hours, a significant swell component with the period about 4 s appears in the wave field. The winds with speeds ∼10 m/s excite wave fields with Hs ∼1 m, depending on the site, wind direction, and duration. At these wind speeds, a superposition of wind waves with a frequency of about 3 s and a swell with a frequency of 4–5 s is typical (cf. below, Figure 10d). In long-lasting gales with the wind speed ∼15 m/s the significant wave height may be about 2 m (cf. [64]). The dominating wave period may largely vary depending on the duration of the gale but high waves with periods > 5 s appear very seldom (cf. [44]). Even in cases when the 1-year return value of the significant wave height is reached, the dominating wave periods normally do not exceed 5–6 s. FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 307

Figure 8. (a) Comparison of the daily maximum heights of the long-wave components with periods >7 s of the wake waves of the fast ferries with the signiﬁcant height of wind waves occurring with the probability of 1% and 5%; (b) comparison of the wake energy and its ﬂux (wave power) with the wind wave energy and power round the year and during the spring and summer seasons. For the measurement site at the western coast of Aegna, the average values of three measurement days from Soomere and Rannat [64] have been used.

5. Contest of Ship Wakes and Wind Waves The above has shown that the heights of the wake waves of fast ferries normally do not exceed 1 m in the coastal zone of Tallinn Bay. In particular, they are much smaller compared with the highest wind waves reaching this area each year. The typical annual maximum significant wave height at the measurement sites is Hs ∼ 1.5 m (Figure 2) and the highest single waves are 2.5–3 m high. However, waves of this height occur seldom and the daily highest ship waves mostly belong to the annual highest 1–5% of wind waves (Figure 8a). Since properties of waves in different wakes vary significantly (cf. [64]), the height of single waves and the number of wave crests are not proper parameters to characterise the impact of ship traffic. To the first approxi- mation, the gross impact of waves of different origin can be quantified in terms of wave energy [67]. The energy of a single wake is small compared with the energy of waves occurring during strong storms in the coastal area of Tallinn Bay [64]. Yet the fast ferry traffic is so heavy there that ship waves do play a notable role in the wave energy balance. Comparison of field data [64] and numerical estimates [59] shows that the mean energy of ship-generated waves at the 5 m isobath of Tallinn Bay is as large as 5–8% from the annual mean wind wave energy, and 6– 12% from the wave energy during the spring and summer seasons (Table I, Figure 8b). Only in a few coastal areas (that are open to the dominating wind directions) the ship wave energy is about 3.5% from the annual and 5% from the summer mean wave energy. Note that a part of wind waves that propagate through a particular near-coastal site are excited by offshore 308 TARMO SOOMERE

Table I. Ship wave recordings 2002a.

Date and Site Depth, Average properties Background wind time of the m of ship waves wave ﬁeld session total (sensor) Energy, Power, J/m2 W/m

14.04 Aegna jetty 6.7 (3.5) 15.8 110 Calm, Hs <10 cm 08:18–18:08 24◦45.25 E, 59◦34.2 N

12.05 Pringi jetty 5.2 (2.1) 13.9 83 Hs ∼20–30 cm 08:25–21:40 24◦47.6 E, 59◦31.15 N

08:00 31.05– Viimsi museum (2 days) 5.2 (2.2) 9.1 51 31.05: Hs ∼40–60 cm; ◦ ◦ 22:30 01.06 24 47.4 E, 59 31. 6 N 01.06: Hs ∼20–30 cm

10:56, 17.06– Aegna, western coast 5.2 (2.2) 9.9 63 Hs up to 70–80 cm at the afternoon of 05:00, 18.06 24◦44.6 E, 59◦34.5 17.06

12:36, 18.06– Aegna, western coast 5.2 (2.2) 9.5 61 Hs ∼30–40 cm 09:00, 19.06 24◦44.6 E, 59◦34.5

08:13, 20.06– Aegna, western coast 5.1 (2.4) 9.3 59 Hs up to 80–90 cm at 09:13, 22.06 24◦44.6 E, 59◦34.5 the morning and 30– 40 cm at the evening of 20.06

06–08.07 Naissaar harbour (2 days) 5.7 (2.5) 8.5 48 Hs <20 cm 24◦33.2 E, 59◦33.7 E winds and propagate towards the open sea, but all the registered ship waves hit the coast. Thus, the above figures should be nearly doubled in the context of the possible influence on the coast. Since the hindcasted annual mean energy of wind waves apparently is overestimated [59], the relative role of ship waves in the total wave activity may be even more significant. A proper comparison of wind waves and ship wakes is only possible when the selected set of wakes well represents the basic properties of a typical wake. The wake patterns are highly variable and may strongly depend on the vessel load and trim, and navigation details [16]. The comparison is based on about traced 240 wakes (from total 280 passages of high-speed ships) on 9 days of five different months of the measurement campaign in 2002. This set contains about 3% from the total number of wakes of high-speed ships this year [64]. Statistical analysis of the properties of wakes of single ships is not performed in Soomere and Rannat [64], because ships pass the measurement site so frequently that the wakes often cannot be separated from each other. The recordings are performed during various weekdays and different FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 309 sailing conditions from perfectly calm conditions to developed windseas with Hs ∼ 1 m. (Figure 5 shows that the latter wave conditions generally occur with a probability of about 1% in the area in question.) No time intervals or weekdays containing wakes with specific properties seem to exist [64]. Also, the highest wakes appear more or less randomly on different days. Based on the listed arguments, the resulting estimates apparently are trustworthy. Solely energy-based comparison of waves of different origin is equivalent to a comparison of the squared wave heights and ignores other wave properties. Since many ship waves are longer than typical wind waves, such a comparison may considerably underestimate the role of ship waves in the nearshore. A more convenient measure is the energy flux or wave power. The bulk power carried by waves per unit of length of wave crests equals to the product of the wave energy density and group speed. It implicitly accounts for the wave periods since longer waves have larger group velocities. For wind wave fields, it is assumed that the wave energy propagates with the group velocity of the wave corresponding to the spectral maximum [59]. For ship wakes, it is assumed that energy propagates with the group velocity of the wave, which period equals to the mean weighted period of the wake [64]. Ship waves play a much larger role in terms of wave power. The annual mean power of ship waves in the coastal area of Tallinn Bay is about 50–100 W/m (Table I, [64]). It constitutes 18–35% from the total wave power at the western coast of the bay, about 13% in areas open to the Gulf of Finland and about 17% at the eastern coast [64]. Owing to strong seasonal variation of the natural wave regime, the power of ship waves forms up to 54% of the power of wind waves. Since the wave propagation direction is not included in the calculations, the ship waves may cause approx- imately from one third to a half of the total wave power at the coasts of the bay during the spring and summer seasons.

6. Quality of Energy in Ship Wakes and Wind Waves The huge difference of the share of ship waves in the bulk wave energy compared to their share in the bulk wave power suggests that a large part of ship waves are longer than typical wind waves. This difference can be quantiﬁed by comparing spectral properties of ship wakes and wind waves. Doing this is not justiﬁed for a single wake, because the properties of its different parts may vary considerably (Figure 9). However, spectral representation of an ensemble of ship waves occurring during a day or during a week apparently reveals some information about the structure of the wakes. This approach allows discussing the properties of large ensembles of transient ship waves in terms of traditional wave energy spectra. It is not 310 TARMO SOOMERE

(a) (b) 0.2 ) -1 -1 Hz ) Hz 2 2 ← 23:50 – 00:24 0.15 0.1

0.1

0.05 ← 01:00 – 01:34 0.05 Spectral density of energy (m Spectral density of energy (m 0 0 0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5 Frequency (Hz) Frequency (Hz) Figure 9. (a) Spectral representation of the wake of Autoexpress sailing to Tallinn on 31.05 near the Viimsi museum: the whole wave field at 21:25–21:49 (solid line), during 2 min of the intense waves at 21:40.5–21:42.5 (dashed line) and during the following record at 21:50–22:24 (dotted line); (b) Spectral density of the energy of the different parts of the wake pattern of SuperSeaCat IV near the coast of Nais- saar 06–07.08. Dashed line represents the energy spectrum at 00:25–01:00 [64]. necessary to assume that ship waves satisfy the linear dispersion relation for surface waves, because an analogous expansion in the Fourier components is widely used, for example, for cnoidal waves [41, 55]. The daily average wave energy spectra for the measurement sites (Figure 10) are simply called energy spectra in what follows. They can be, in principle, directly calculated by applying the FFT to the set of properly prepared pressure recordings, with a subsequent rescaling of the amplitudes of the wave components (see Section 2). However, in order to avoid calculations of the FFT with a very large number of points (∼106) and the influence of the recording pauses (see above) on the results, the spectra are obtained as an arithmetical mean of the properly rescaled energy spectra for particular recordings of Soomere and Rannat [64]. The spectra in Figure 10 are normalised to correctly represent the daily mean properties of ship waves at the measurement sites, i.e. they not necessarily represent the exact amount of wind wave energy on the measurement days. Although the spectra of wakes from single ships are very sharply peaked (Figure 9), in average the wave energy is distributed much more evenly among the wave components (Figure 10). The measurement conditions near Aegna jetty were the most favour- able. The significant height of wind waves was well below 10 cm and the height of long-wave components with periods >5 s was only a few cm. Thus, this day was perfect for recording wakes from conventional vessels and hydrofoils that have modest heights and periods. The energy spectrum in Figure 10a undoubtedly demonstrates that the role of wakes from these vessels is negligible in the total energy budget of the ship-generated waves. FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 311

0.035 0.035 ) ) -1 -1 (a) (b)

Hz 0.03

Hz 0.03 2 2 0.025 0.025

0.02 0.02

0.015 0.015

0.01 0.01

0.005 0.005 Spectral density of energy (m Spectral density of energy (m 0 0 0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5 Frequency (Hz) Frequency (Hz)

0.035 0.035 ) ) -1 (c) -1 (d)

Hz 0.03 Hz 0.03 2 2 0.025 0.025

0.02 0.02

0.015 0.015

0.01 0.01

0.005 0.005

Spectral density of energy (m 0 Spectral density of energy (m 0 0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5 Frequency (Hz) Frequency (Hz)

0.035 0.035 ) ) -1 -1 (e) (f)

Hz 0.03

Hz 0.03 2 2 0.025 0.025

0.02 0.02

0.015 0.015

0.01 0.01

0.005 0.005 Spectral density of energy (m Spectral density of energy (m 0 0 0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5 Frequency (Hz) Frequency (Hz) Figure 10. Daily average spectral energy density of waves (a) at Aegna jetty on 14.04; (b) at Pringi jetty on 12.05; (c) at the Viimsi museum on 31.05-01.06; (d), (e) at the western coast of Aegna on 17.07 and on 18.07, respectively; (f) at Naissaar harbour on 06-08.08.

This conclusion is evermore unambiguous because of the fact that this measurement site is the second closest site to the ship lane (Figure 2). Figure 10a also shows that the third group of wakes from high-speed ships (with periods of 3–4 s) carries a very small fraction of the total wave energy that is practically indistinguishable in the average spectrum. This feature is somewhat unexpected, because this group frequently is the 312 TARMO SOOMERE highest part of wake patterns in remote areas, and evidently occurs because high and short waves appear relatively seldom, and that the duration of their packet is short. The energetically dominant waves of different origin are well separated in the spectral domain in most of the wind conditions. The separation occurs at different frequencies depending on both the location of the site and the background wave field. It is unambiguous in measurements at Pringi jetty (Figure 10b) and at Naissaar harbour (Figure 10f). On these days, the significant height of wind waves was about 30 or 20 cm, respectively (that is typical for summer conditions in the area in question, [44]). The presence of wind waves is visible as a weak maximum for waves with periods about 3 s in these spectra. Somewhat more overlapping wave components are found near the Viimsi museum (Figure 10c) where Hs ≈ 60 cm during a part of the day. The separation is less spectacular but still evident near the western coast of Aegna on 18.06 (Figure 10e), and occurs even in the conditions of rough windseas [64]. An extensive overlapping of waves of different origin in the frequency spectra only occurs at the western coast of Aegna (Figure 10d). It is caused by a strong wind on this day. The significant wave height was close to 1 m during a part of this recording. Wave fields of this intensity occur quite seldom, with a probability of a few per cent (Figure 5a). The site is totally open towards the central part of the Gulf of Finland (cf. Figures 5 and 6) and open sea swell frequently occurs there. Indeed, on 17.06 a portion of swell with periods about 5 s was pres- ent in the background wave field. The spectral separation of waves of different origin in most of the wind conditions allows to estimate the average energy density of the long-wave fraction of ship waves and to perform comparison of waves of different origin. The energy spectra of wind wave fields typically have a very steep slope at the long-wave side of the spectral maximum [34]. Therefore the spectra in Figures 10d,e apparently correctly describe the energy of ship waves that are slightly longer than wind waves or swell corresponding to their spectral maximum. Most probably, a part of the relatively modest slopes in the frequency range 0.15–0.19 Hz in Figure 10d,e is caused by the presence of ship wakes. The major part of the energy of wake waves of high-speed vessels lies outside the frequency range of wind waves occurring in typical conditions. This property is most remarkable at Aegna jetty (Figure 10a) where about 77% of the wake energy is concentrated in waves with periods from 8 to 15 s (cf. Figure 4). The well-defined peak corresponding to the longest waves in most of the energy spectra in Figure 10 suggests that, in average, the energy of the leading (that are also the longest) waves of the wash considerably exceeds the energy carried by the rest of the wake. The two peaks close to each other, FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 313 for example, in the spectrum at Aegna jetty apparently represent different properties of leading waves of different (types of) ships operating in this area. Although the typical periods of the first and the second wave group within a wake differ substantially [16, 29, 45, 64, 70], these groups are identifiable only in Figure 10b (at Pringi jetty) where ships normally do not use their service speed. The sharp sub-peaks in some panels of Figure 10 apparently represent leading waves from a small number of near-critical wakes. The prevalence (in terms of energy) of the first wave group and, in particular, of the longest (resp. leading) waves (except at Pringi jetty in Figure 10b and at Naissaar in Figure 10f), is somewhat unexpected. The duration of the second wave group frequently exceeds that of the first group [16, 70], and one might expect that energy is divided more or less evenly among waves of different length. The conclusion that the longest (leading) waves are also the highest is essential in the areas such as Tallinn Bay where long wind waves have very modest heights. The majority of the energy of wakes from high-speed ships is concentrated in wave components with heights close to 1 m and periods of 10–15 s in the coastal zone of Tallinn Bay. Such waves extremely seldom, if ever, exist in the natural conditions in the area in question. Although one should “never say never”, this formulation is fully relevant in this context, because the Baltic Sea in its contemporary size and shape is less than 10,000 years old [49]. It is important to emphasize that Naissaar (that is located at a much larger distance from the ship lane than other measurement sites) receives nearly the same amount of wake energy as the sites close to the ship lane [64]. Therefore, the area of possible impact of heavy fast ferry traffic can be considerably wider than the immediate vicinity of the ship lane. The energy of waves with periods of 6–10 s at Naissaar coincides with that in other sites. Only the portion of even longer waves is smaller. This is so apparently because of more intense wave-bottom interaction of long waves between the ship lane and Naissaar, and because that only divergent waves reach the coastal zone of Naissaar. The location of the spectral peaks of the average energy distribution varies greatly at different sites. Therefore, different areas of the coastal zone of Tallinn Bay receive wake wave energy in different frequencies (or wave- lengths). The area at Aegna jetty receives a large portion of very long waves with periods exceeding 10 s. The neighbouring area at the western coast of Aegna receives some energy of waves with periods of 9–10 s but mostly wave energy is concentrated in components with periods of 6–8 s. The most energetic waves at Pringi (the closest site to the Tallinn Harbour) are relatively short (with periods <9 s). This difference is consistent with the fact that the wakes of high-speed ships are non-stationary and that the 314 TARMO SOOMERE particularly long waves are formed only after a ship had sailed a certain time at a high speed [36].

7. Excessive Wake-Induced Near-Bottom Velocities and a Potential New Pattern of Sediment Transport Analysis of geological setting and nearshore processes in the Tallinn Bay area is presented in [27]. The shores are often covered by coarse-grained terrigenous deposits (gravel, pebbles, cobbles), which overlie glacial deposits. The pebble and cobble pavement in the nearshore protects the shore from further erosion. Ship waves with modest heights normally are not able to erode these shores directly. The abrupt increase of the ship wave loads during the last decade apparently plays a modest role in the sediment transport processes in backshore and at depths 0–5 m where natural wave activity dominates [64]. The inﬂuence of ship waves may, however, be much larger in deeper parts of the nearshore. The most critical issue in the area in question is the difference of prevailing periods of the highest parts of the ship wash and the windseas. For a ﬁxed wave height, the wave-induced near-bottom velocity depends essentially on the wave period. For the depths of 5–30 m, it has the highest variation when the wave period increases from 5 to 8 s (Figure 11). Since the major part of the energy of the wakes from fast ferries is concentrated in waves with the periods exceeding 6–7 s (Figure 10), the impact of a typical ship wake on bottom sediments and aquatic wildlife at these depths is comparable with or even exceeds the impact of wind waves occurring in the most violent storms. In particular, the leading waves may cause unusually high near-bottom velocities in non-tidal areas at the depth of 10–30 m. Many components of

1 ) -1 H=3 m 0.8

H=5 m 0.6

H=10 m 0.4 H=20 m 0.2 H=30 m Maximum nearbottom velocity (ms 0 0 2 4 6 8 10 12 14 16 Wave period (s) Figure 11. Dependence of the maximum wave-induced near-bottom velocity on the wave period for the wave height of 1 m and the water depth of 3–30 m. FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 315 such marine systems, historically, have been adjusted to low near-bottom velocities. For that reason the waves from high-speed ships may appear as particularly hazardous new forcing component in such areas at certain depths. The increase of hydrodynamic activity at certain depths becomes first evident in re-suspension and sediment transport processes. An abrupt intensification of long-wave activity may cause considerable changes in the existing balance of sediment distribution. An analogous effect has been noticed for ocean conditions where already relatively small levels of long-period wave energy in combination with wind waves can cause greater beach response than an equal amount of energy in the windsea only frequencies [10]. The response of the sea bed may be particularly large in semi-enclosed micro-tidal sea areas where natural waves are limited to a certain height and/or frequency range. In such areas, near-bottom currents usually are relatively weak and there occurs a little sediment transport in the deeper part of the coastal zone in natural conditions. Long wake waves from fast ferries probably re-suspend sediments at deeper parts of the nearshore than the natural waves. If this part of the coastal area consists of unprotected fine sediments, they may be actively re-suspended and transported to other locations (probably offshore by undertow, [6]). An extensive joint influence may happen in areas where intense long- wave wash occurs simultaneously with strong short-wave activity. In such an environment, waves of different origin transport sediments at different water depths, and wake waves may trigger a more complicated pattern of sediment transport processes. A combination of these features does happen in a large part of the coast of Tallinn Bay where wind waves have much smaller periods than the leading ship waves. The velocity of near-bottom currents normally does not exceed 10 cm/s in the deeper parts of the nearshore (U. Raudsepp and J. El- ken, personal communication). A part of the bottom sediments is not protected and the presence of fine-grained bottom sediments in the deeper part of the coastal zone (depths > 10 m; [27, 39]) indicates a low hydrodynamic activity. At the depths 5–10 m the bottom deposits are usually represented by sand with some silt (up to 25%). Between 10 and 20 m, the sea floor is covered with silt containing ca. 30% of pelite, and in places outcrops of glaciolacustrine deposits (varved clay) occur. A qualitative description of a principally possible new phase of the coastal evolution, that may happen if a coast with a low hydrodynamic activity and a proper structure of bottom sediments is exposed to a multi- frequency wave system during a long time, is sketched in [61]. A typical shore of Tallinn Bay mostly consists of post-glacial sand and gravel with some pebbles and cobbles, which overlie till. The initial evolution of such 316 TARMO SOOMERE a shore owing to the wave activity is sketched in classical texts and manu- als [5, 9, IV-3-51, 71]. Waves erode the shore in the immediate vicinity of the waterline and transport finer sediments to the deeper area. At a certain distance from the original shoreline an abrasional scarp is formed (Figure 12a). A part of the nearshore becomes less steep and, finally, a nearly flat terrace is built. The finer the particles and the larger the wave activity, the further away from the shoreline they are transported. Larger stones mostly retain their initial position and occasionally cover a large part of the area directly influenced by waves. The flattened nearshore (where large waves usually break) and the stone pavement jointly protect the shore from further erosion, and the erosion and transport processes slow down (Figure 12b). After some time, the evolution of the coastal zone reaches almost an equilibrium stage whereas the shore profile and the structure of its sediments are basically defined by the local wind wave regime. The shore erosion reinforces for short periods only during very strong storms accompanied by unusually high water levels. As a result, additional scarp(s) may be formed at the upper boundary of the backshore, but the profile of the nearshore remains more or less steady. Long wake waves from fast ferries add a new constituent to this scheme. They may excite velocities that occur extremely seldom or never under natural conditions in the deeper part of the nearshore that often consists of relatively fine sediments and has no protecting pavement. They possibly trigger a new stage of coastal zone evolution through fast eroding the sea- side part of the nearshore at certain depths and resuming the total transport of bottom sediments to the deeper area (Figure 12c). There are no experimental data concerning the intensity and direction of the described sediment transport yet. However, the perceptible energy of wakes, the significant amount of ship wave power as compared to the bulk wave power, the existing structure of bottom sediments at depths of 10–25 m and the estimates of the wave-induced near-bottom velocities (that are comparable to those excited by wind waves at the depths of a few metres) suggest that its intensity might be comparable to that occurring in the initial stage of the erosion of a smooth coastal profile. Schoellhamer [57] found that in Hillsborough Bay, a shallow, microtidal, subtropical estuary in West-central Florida, large vessels in a dredged ship channel can generate long solitary waves that cause large water velocities and sediment resuspension. The annual mass of sediments resuspended by long ship waves is by one order of magnitude greater than the annual mass of sediment brought into motion by wind waves. A secondary impact of the long waves is that sediments that are resuspended and newly deposited are more susceptible to resuspension by natural currents than undisturbed bottom sediments. This is particularly important in the context of the current FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 317

Figure 12. Qualitative scheme of evolution of an initially smooth coastal profile owing to natural wave activity (a, b, not to scale) and a possible effect of long waves from high-speed ships (c, d, not to scale) [61]. 318 TARMO SOOMERE study; because wakes from high-speed ships are extremely strong impulse loads that at times essentially exceed similar loads caused by natural fac- tors, and may act just as agents of resuspension of sediment fractions that are further transported by natural processes. This transport may have twofold influence on the processes near the shoreline. First, it apparently creates a deficit of sediments in some inter- mediate part of the nearshore that will be balanced by a more intense transport of material from the vicinity of the shoreline. Another, more subtle, consequence is that after some time the underwater terrace may be partially or totally eroded. If this happens, the intensity of wind wave breaking decreases, and more wind wave energy penetrates to the shoreline (Figure 12d). As a result, the overall intensification of wave-induced sediment transport processes (possibly accompanied by a certain shoreline reduction) might be to a large extent higher than estimated from plain energy-based quantities. Recently, Webster [69] examined the enhancement of solute diffusivities within coastal surficial sediments as a result of wave action. Wave amplification of diffusivities is the greatest for waves of period 10 s. It might be a simple coincidence that the leading waves of the wakes from high-speed ferries have comparable periods. However, this feature once more suggests that these wakes may have unexpectedly high influence on the sediment transport processes.

8. Conclusions Ship-generated waves frequently have the height of about 1 m and a period of about 10–15 s. Such waves are extremely seldom in the area in question in natural conditions. The estimates in terms of wave energy, power and wave-induced near-bottom velocity suggest the ship traffic adds a qualitatively new forcing factor into the local marine ecosystem in certain semi- enclosed sea areas such as Tallinn Bay. Ship waves are of modest height but may have considerable larger periods than wind waves and swell. Their role is impressive in terms of energy (∼ 10% from the bulk wave energy) but striking in terms of wave power where ship waves constitute 18–35% (27–54% during the summer season) from the total wave power at the coasts of Tallinn Bay. The anthropogenic waves may even dominate during a part of the relatively calm high navigation season (April–June) when the biological productivity is at its seasonal maximum. At least theoreti- cally, they may cause considerable overall intensification of certain coastal processes. Thus, we are faced with the unique situation where a principally new component of the wave regime with intensity comparable to that of the existing one has abruptly emerged in open sea conditions. The reasons of FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 319 this phenomenon are that (i) Tallinn Bay is a non-tidal area where wind waves may be high but, owing to relatively short fetch, have moderate periods, and that (ii) the density of high speed ship traffic is so excessive that ship waves form a substantial part of the bulk wave intensity. Literally, the contribution of ship waves to the wave climate is comparable to the increase of wave activity that would happen if open ocean swell from the North Atlantic reached the Gulf of Finland. The major threats to the coasts and the ecosystem in the vicinity of ship lanes may arise from the distinctive properties of the wake waves from fast ferries. An extensive reaction of fine bottom sediments at the deeper part of the nearshore is conceivable. The accompanied reduced water transpar- ency [11, 60], besides of the impact of direct mechanical disturbances, may have suppressing feedback on the bottom vegetation. Suspension and re- sedimentation of finer sediments may considerably worsen fish spawn- ing conditions. Another potential mechanical effect of ship waves is an enhancement of vertical mixing along the ship lane [24, 25]. This may intensify the eutrophication effects and influence harmful algae blooms due to the transport of nutrients from sediments into the euphotic layer [38]. The extension and many properties of this influence are still unclear and have to be addressed in further studies. Many important features of ship wakes, namely the possibility of excitation of soliton-like wave groups, the actual shape and nature of the leading waves, and their implicit influence on sea water properties and local ecosystem also request further investiga- tions. The most important aspect is the remote impact of the ship traffic. The amount of the wake wave energy that reaches fairly remote areas is comparable with the wave energy in the vicinity of the ship lane. Although this feature of surface waves has been discussed earlier [16, 66, 70], it should be particularly underlined that waves excited by high speed ships under certain conditions are practically non-dispersive compact entities carrying massive amounts of energy. The wave energy may become active in the form of violent plunging breakers far from the ship lane and a long time after the ship has passed [17]. Apart from the spectacular sights, remote influence of wakes apparently is responsible for drastic thermal changes in shallow inlets several kilometres away from the fairway [37] and even at the open sea [12]. This feature has to be taken into account in the analysis of environmental impact of harbours and associated ship traffic in the neighbourhood of vulnerable areas, in particular, if long-living compact wave groups form a part of the wake. Since the location of the harbours and vulnerable areas (either densely populated or protected because of environmental issues) cannot be changed in practice, it has become necessary to introduce a new paradigm of treat- ing the ship traffic as a potential source of remote influence. In particular, 320 TARMO SOOMERE it is natural to extend the definition of pollution (that today commonly is interpreted as releasing certain substances or noise into the environment) towards including the releasing of energy in general into the marine environment [23, 67].

Acknowledgements The field works of the presented study were mostly performed in the frame- work of the project “The influence of ship wake on beaches of the Viimsi Peninsula and Naissaar and Aegna islands, and the possibilities of its neutralising” (financed by the Estonian Environmental Investments Cen- tre). Follow-up field campaigns in 2003–2004 were supported by the Centre of Excellence for Non-linear Studies. Financial support from the Estonian Science Foundation (Grant 5762) as well as permission of the Estomian Academy publishers to use figures from [61, 64] are gratefully acknowl- edged. A part of the studies were performed in the Estonian Marine Insti- tute and the Marine Systems Institute. Special thanks are to Andres Kask who prepared Figures 2 and 12.

References 1. Ali, M.M., Murphy, K.J. and Langendorff, J.: 1999, Interrelations of river ship trafﬁc with aquatic plants in the River Nile, Upper Egypt, Hydrobiologia 415, 93–100. 2. Boissonnas, J., Connolly, N., Mantoura, F. and d’Ozouville, L. (eds.): 2002, Integrating Marine Science in Europe. Marine Board, European Science Foundation, Strasbourg, 148 pp. 3. Bourne, J.: 2000, Louisiana’s Vanishing Wetlands: going, going..., Science 289 (5486), 1860–1863. 4. Brown, E.D., Buchsbaum, S.B., Hall, R.E., Penhune, J.P., Schmitt, K.F., Watson, K.M. and Wyatt, D.C.: 1989, Observations of a nonlinear solitary wave packet in the Kelvin wake of a ship, J. Fluid Mech. 204, 263–293. 5. Bruun, P.: 1988, The Bruun Rule of erosion by sea-level rise: a discussion of large- scale two- and three-dimensional usages, J. Coastal Res. 4, 627–648. 6. Carter, R.W.G.: 2002, Coastal Environments, 8th printing, Academic Press, London, 2002, 617 pp. 7. Chen, X.-N. and Sharma, S.D.: 1995, A slender ship moving at a near-critical speed in a shallow channel, J. Fluid Mech. 291, 263–285. 8. Climatological Ice Atlas for the Baltic Sea, Kattegat, Skagerrak and Lake Vänern (1963–1979): 1982, SMHI, Swedish Meteorological and Hydrological Institute, Nor- rköping, Sweden, and Institute of Marine Research, Helsinki, Finland, Norrköping, 220 pp. 9. Coastal Engineering Manual: 2002, Department of the Army. U.S. Army Corps of Engineers. Manual No. 1110-2-1100. 10. Coates, T.T. and Hawkes P.J.: 1999, Beach recharge design and bi-modal wave spectra. In: B.L. Edge (ed.), Coastal Engineering 1998: Proceedings of the 26th Interna- tional Conference 22–26 June 1998, Falconer Hotel, Copenhagen, vol. 3, pp. 3036–3045, American Society of Civil Engineers (ASCE), USA. FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 321

11. Erm, A. and Soomere, T.: 2004, Influence of fast ship waves on optical properties of sea water in Tallinn Bay, Baltic Sea, Proc. Estonian Acad. Sci. Biol. Ecol. 53, 161–178. 12. Fagerholm, H.P., Rönnberg, O., Östman, M. and Paavilainen, J.: 1991. Remote sensing assessing artificial disturbance of the thermocline by ships in archipelagos of the Baltic Sea with a note on some biological consequences. International Geoscience and Remote Sensing Symposium Digest 2, 377–380. 13. Forsman, B.: 2001, From bow to beach, SSPA Highlights No. 3, 4–5. 14. Gaskin, S.J., Pieterse, J., Al Shafie, A. and Lepage, S.: 2003, Erosion of undisturbed clay samples from the banks of the St. Lawrence River, Canadian Journal of Civil Engineering 30, 585–595. 15. Grimshaw, R., Pelinovsky, D., Pelinovsky, E. and Talipova, T.: 2001, Wave group dynamics in weakly nonlinear long-wave models, Physica D 159, 35–57. 16. Guidelines for managing wake wash from high-speed vessels: 2003, Report of the Work- ing Group 41 of the Maritime Navigation Commission. International Navigation Association (PIANC), Brussels, 32 pp. 17. Hamer, M.: 1999, Solitary killers, New Scientist No. 163 (2201), 18–19. 18. Hammack, J., Scheffner, N. and Segur, H.: 1989, Two-dimensional periodic waves in shallow water, J. Fluid Mech. 209, 567–589. 19. Hannon, M.A. and Varyani, K.S.: 1999, The wake effect of high speed ferries in coastal and inland waterways. In International Conference on Coastal Ships and Inland Waterways, RINA, London, February 1999,11pp. 20. Hobbs, P.V., Garrett, T.J., Ferek, R.J., Strader, S.R., Hegg, D.A., Frick, G.M., Hop- pel, W.A., Gasparovic, R.F., Russell, L.M., Johnson, D.W., O’Dowd, C., Durkee, P.A., Nielsen, K.E., and Innis, G.:. 2000, Emissions from ships with respect to their effects on clouds, J. Atmos. Sci. 57, 2570–2590. 21. Hüsig, A., Linke, T. and Zimmermann, C.: 2000, Effects from supercritical ship oper- ation on inland canals, J. Waterway, Port, Coastal Ocean Eng. 126, 130–135. 22. IAHR working group on wave generation and analysis: 1989, List of sea-state parameters, J. Waterway, Port, Coastal Ocean Eng. 115, 793–808. 23. IMO: 2000, A Technical Protocol for the Assessment of Abnormal wave (wash) Effects Generated by High Speed Craft (HSC) and Ships: 2000, MEPC 45/INF.19, 28.07.2000, Marine Environment Protection Committee, International Maritime Orga- nisation, London, 3 pp. 24. Jacobsen, T.S.: 1988, Ship traffic – an important source of mixing in Great Belt? In: Proceedings of 16th Conference of the Baltic Oceanographers, Kiel, September 1988, vol. 1, pp. 505–525, Institute of Marine Research, Kiel. 25. Jürgensen, C.: 1991. Vertical mixing due to ship traffic and consequences for the Bal- tic Sea. In: Report from IABSE Colloquium, pp. 187–194. Nyborg, Denmark. 26. Kahma, K. and Pettersson, H.: 1994, Wave growth in a narrow fetch geometry, The Global Atmosphere and Ocean System 2, 253–263. 27. Kask, J., Talpas, A, Kask, A. and Schwarzer, K.: 2003, Geological setting of areas endangered by waves generated by fast ferries in Tallinn Bay, Proc. Estonian Acad. Sci. Eng. 9, 185–208. 28. Keevallik, S.: 2003, Possibilities of reconstruction of the wind regime on Tallinn Bay, Proc. Estonian Acad. Sci. Eng. 9, 209–219. 29. Kirk McClure Morton: 1998, Investigation of High Speed Craft on Routes near to Land or Enclosed Estuaries. Research Report JR226. The Maritime and Coastguard Agency, UK. 30. Kirkegaard, J., Kofoed-Hansen, H. and Elfrink, B.: 1999: Wave wake of high-speed craft in coastal areas. In: B.L. Edge (ed.), Coastal Engineering 1998: Proceedings of 322 TARMO SOOMERE

the 26th international conference 22–26 June 1998, Falconer Hotel, Copenhagen, vol. 1, pp. 325–337, American Society of Civil Engineers (ASCE), USA. 31. Kofoed-Hansen, H.: 1996, Technical Investigation of Wake Wave from Fast Ferries. Danish Hydraulic Institute, Report No. 5012, 41 pp. 32. Kofoed-Hansen, H. and Mikkelsen, A.C.: 1997, Wake wave from fast ferries in Den- mark. In: Proceedings of the 4th International Conference of Fast Sea Transportation FAST’97, Sydney, 1997, vol. 1, pp. 471–478, Baird Publications, Hong Kong. 33. Kohonen, T., Rytkonen, J., Virtasalo, J. and Vuorinen, I.: 2001, Sediment erosion and resuspension near navigation routes and dredged material disposal sites. In: Baltic Sea Science Congress 2001, Abstract volume, p. 226. Stockholm Marine Research Centre. 34. Komen, G.J., Cavaleri, L., Donelan, M., Hasselmann, K., Hasselmann, S. and Janssen, P.A.E.M.: 1994, Dynamics and Modelling of Ocean Waves, Cambridge Uni- versity Press, Cambridge 532 pp. 35. Lakshmanan, M. and Rajasekar, S.: 2003, Nonlinear Dynamics, Springer, Berlin, Hei- delberg, 619 pp. 36. Li, Y. and Sclavounos, P.D.: 2002, Three-dimensional nonlinear solitary waves in shallow water generated by an advancing disturbance, J. Fluid Mech. 470, 383–410 37. Lindholm, T.: 1997, Färjor ger fart åt alger (Ferry traffic speeds up phytoplankton). Vatten 53, 133–136 (in Swedish). 38. Lindholm, T., Svartström, M. Spoof, L. and Meriluoto, J.: 2001, Effects of ship traffic on archipelago waters off the Långnäs harbour in Åland, SW Finland, Hydrobiologia 444, 217–225. 39. Lutt, J. and Tammik, P.: 1992, Bottom sediments of Tallinn Bay, Proc. Estonian Acad. Sci. Geol. 41, 81–87. 40. Madekivi, O. (ed.): 1993, Alusten aiheuttamien aaltojen ja virtausten ympäristöva- ikutkset (The environmental effects of ship-induced waves and currents). Vesi ja Ympäristöhallinnon Julk. Sarja A, 166, 1–113 (In Finnish). 41. Massel, S.R.: 1989, Hydrodynamics of Coastal Zones, Elsevier, Amsterdam, 336 pp. 42. Mietus, M. (co-ordinator): 1998, The Climate of the Baltic Sea Basin, Marine mete- orology and related oceanographic activities, Report No. 41, World Meteorological Organisation, Geneva, 64 pp. + figures. 43. Nanson, G.C., von Krusenstierna, A. and Bryant., E. A.: 1994, Experimental measurements of river-bank erosion caused by boat-generated waves on the Gordon River, Tasmania. Regul Rivers, Res Manage 9, 1–14. 44. Orlenko, L.R. (ed.): 1984, Studies of the Hydrometeorological Regime of Tallinn Bay. Gidrometeoizdat, Leningrad (in Russian). 45. Parnell, K.E. and Kofoed-Hansen, H.: 2001, Wakes from large high-speed ferries in confined coastal waters: Management approaches with examples from New Zealand and Denmark, Coastal Manage. 29, 217-237. 46. Peterson, P., Soomere, T., Engelbrecht, J. and van Groesen, E.: 2003, Soliton interaction as a possible model for extreme waves in shallow water, Nonlinear Proc. Geophys. 10, 503–510. 47. Pettersson, H.: 2001, Directional Wave Statistics from the Gulf of Finland. MERI, vol. 44, Finnish Institute of Marine Research, 37 pp. 48. Pettersson, H. and Boman, H.: 2002, High waves and sea level during the November storm. In: Annual Report 2001, p. 7, Finnish Institute of Marine Research, Helsinki. 49. Raukas, A. and Teedumäe, A. (eds.): 1997, Geology and Mineral Resources of Estonia, Institute of Geology, Estonian Academy Publishers, Tallinn, 436 pp. 50. Rönnberg, O.: 1975, The effects of ferry traffic on rocky shore vegetation in the southern Åland archipelago. Merentutkimuslaitoksen Julk./ Havsforskningsinst. Skrift FAST FERRY TRAFFIC IN NON-TIDAL SEA AREAS 323

239, 325–330. 51. Rönnberg, O.: 1981, Traffic effects on rocky-shore algae in the Archipelago Sea, SW Finland. Acta Acad. Aboensis Ser. B 41, 1–86. 52. Rönnberg, O. and Lax, P.-E.: 1980, Influence of wave action on morphology and epi- phytic diatoms of Cladophora glomerata (L.) Kütz. Ophelia 1(Suppl.), 209–218. 53. Rönnberg, O., Östman, T. and Ådjers, K.: 1991, Fucus vesiculosus as an indicator of wash effects of ships’ traffic. Oebalia 27 (Suppl), 213–222. 54. Rzheplinsky, G.V. and Brekhovskikh, Yu.P. (eds.): 1967, Wave Atlas for Gulf of Fin- land. Gidrometeoizdat, Leningrad (in Russian). 55. Sarpkaya, T. and Isaacson, M.St.Q.: 1981, Mechanics of Wave Forces on Offshore Structures, Van Nostrand Reinold, New York, 651 pp. 56. SBE26 Seagauge Wave and Tide Recorder: 2002, Operating manual. Sea-Bird Electron- ics, Inc., Bellevue, Washington, USA. 57. Schoellhamer, D.H.: 1996, Anthropogenic sediment resuspension mechanisms in a shallow microtidal estuary. Estu Coast Shelf Sci 43, 533–548. 58. Soomere, T.: 2003, Anisotropy of wind and wave regimes in the Baltic Proper, J Sea Res. 49, 305–316. 59. Soomere, T.: 2005, Wind wave statistics in Tallinn Bay, Boreal Env. Res., 10, 103–118. 60. Soomere, T., Elken, J., Kask, J., Keevallik, S., Kõuts, T., Metsaveer, J. and Peterson, P.: 2002, The Influence of Ship Wake on Beaches of the Viimsi Peninsula and Naissaar and Aegna Islands, and the Possibilities of its Neutralising. Research report, Marine Systems Institute at Tallinn Technical University, Tallinn, 243 pp (in Estonian). 61. Soomere, T. and Kask, J.: 2003, A specific impact of waves of fast ferries on sediment transport processes of Tallinn Bay, Proc. Estonian Acad. Sci. Biol. Ecol. 52, 319–331. 62. Soomere, T. and Keevallik, S.: 2001, Anisotropy of moderate and strong winds in the Baltic Proper, Proc. Estonian Acad. Sci. Eng. 7, 35–49. 63. Soomere, T. and Keevallik, S.: 2003, Directional and extreme wind properties in the Gulf of Finland. Proc. Estonian Acad. Sci. Eng. 9, 73–90. 64. Soomere, T. and Rannat, K.: 2003, An experimental study of wind waves and ship wakes in Tallinn Bay. Proc. Estonian Acad. Sci. Eng. 9, 157–184. 65. Soomere, T., Rannat, K., Elken, J. and Myrberg, K.: 2003, Natural and anthropogenic wave forcing in Tallinn Bay, Baltic Sea. In: C.A. Brebbia, D. Almorza and F. López-Aguayo (eds.), Coastal Engineering VI, pp. 273–282, WIT Press, Southampton, Boston. 66. Sorensen, R.M.: 1973, Ship-generated waves. Adv. Hydrosci. 9, 49–83. 67. Stumbo, S., Fox, K., Dvorak, F. and Elliot, L.: 1999, The prediction, measurement, and analysis of wake wash from marine vessels. Marine Technol. SNAME News 36, 248–260. 68. Wood, W. A.: 2000, High-speed ferry issues for operators and designers. Marine Tech- nol. SNAME News 37, 230–237. 69. Webster, I.T.: 2003, Wave enhancement of diffusivities within surficial sediments. Envi- ron Fluid Mech 3, 269–288. 70. Whittaker, T.J.T., Doyle, R. and Elsäßer, B.: 2001, An experimental investigation of the physical characteristics of fast ferry wash. In: V. Bertram (ed.), Second Interna- tional EuroConference on High-Performance Marine Vehicles HIPER’01 Hamburg, 2001, pp. 480–491. Technical University Hamburg-Harburg. 71. Wright, L.D. and Short, A.D.: 1984, Morphodynamic variability of surf zones and beaches: a synthesis, Marine Geol. 56, 93–118. 72. Wulff, F.V., Rahm, L.A. and Larsson, P. (eds.): 2001, A Systems Analysis of the Baltic Sea, Ecological Studies 148, Springer, Berlin, 430 pp.