UBIQUITY OF METEOTSUNAMIS IN THE NORTHERN AND EASTERN GULF OF

MEXICO

Maitane Olabarrieta* and Arnoldo Valle-Levinson

Civil and Coastal Engineering Department, ESSIE, University of Florida

365 Weil Hall, Gainesville, FL, 32611, United States

(*) Corresponding author

E-mail addresses: maitane.olabarrieta@essie.ufl.edu (Maitane Olabarrieta) and

arnoldo@coastal.ufl.edu (Arnoldo Valle-Levinson)

ABSTRACT

On March 28th 2014 non-tidal sea level oscillations >0.8 m were reported at Panama City Beach

(FL). Those oscillations were linked to a prefrontal squall line that originated in the Alabama

coast on March 28th. The squall line expanded radially at 19 m/s in the southeastward direction

along the west coast of Florida, reaching the southern tip of Florida after 1000 PM on March

29th. Atmospheric pressure dropped across the squall line by up to 5.5 hPa. Accompanying the

pressure drop, air temperature decreased by 3-4 ºC and wind changed by 135º. Three main

meteotsunami waves associated with different atmospheric pressure drops were identified during

this two-day period. High correlation between atmospheric pressure changes and radar

reflectivity allowed characterization of atmospheric pressure fields at the spatial scales needed..

Numerical simulations were forced with simplified atmospheric pressure fields (from reflectivity

mosaic maps). Although the maximum height of the meteotsunami was underestimated the

arrival time and the generation of the main meteotsunami was correctly captured in the model.

1

Proudman resonance was identified as relevant in producing the meteotsunamis but Greenspan

resonance was also an important contributor. During the year of 2014, three more meteotsunami

events were identified in the northern and eastern Gulf of Mexico. Further analysis of maximum

surges between 1996 and 2014 has shown that meteotsunami events are not unusual in the

eastern gulf. Four of the 10 highest surges identified in Naples, FL, were originated by squall

lines propagating in the alongshore direction.

INTRODUCTION

Meteorological tsunamis or meteo-tsunamis are sea level oscillations with periods from few

minutes to few hours. They are generated by atmospheric disturbances associated with fronts

such as squalls, atmospheric gravity waves, and fast pressure changes. Unlike seismic or

landslide tsunamis, the occurrence and impact of meteorologically generated tsunamis is rather

local. However, their effects can be as severe as those caused by tsunamis and become

catastrophic (Rabinovich et al., 2009). Meteotsunamis have been reported worldwide under

different names: Rissaga in Ciutadella (Spain), Marrobbio in Sicily (Italy), Abiki in Nagasaki

Bay (Japan), milghuba in Malta, šćiga in Croatia, and Seebär in the Baltic Sea. Although not

regularly, these waves have also occurred in Western Australia (Pattiaratchi and Wijeratne,

2014), the Yellow Sea, the English Channel, the Great Lakes, the northwestern Atlantic coast,

the Argentinian coast, and the New Zealand coast. Meteotsunamis have been described

extensively by Monserrat et al. (2006) and Rabinovich et al. (2009). A recent special issue of

Natural Hazards (74(1), 2014) has been dedicated to the study of meteotsunamis. However, no

studies to date have reported their development in the Gulf of Mexico. This study fills that void

in information and also documents the ubiquity of meteotsunami occurrence in this marginal sea.

2

Atmospheric disturbances are typically transmitted to the ocean’s surface through the inverse

barometer effect. These disturbances can grow when they travel at speeds Ca over water depths

where the long wave celerity matches Ca. This situation is known as Proudman resonance

(Proudman, 1929) and maximizes the energy transfer from the atmosphere to the ocean. In this

case, atmospheric forcing is bound to the ocean’s surface wave. The amplitude of the

meteotsunami generated under these circumstances depends on the intensity of the atmospheric

perturbation and on the time or distance in which the meteotsunami was bounded to the

atmospheric pressure disturbance. If the water depths change and the celerity of the atmospheric

pressure disturbance remains unmodified, the formerly bounded meteotsunami becomes a free

wave and does not absorb more energy from the atmosphere. Although Proudman resonance is

the best-known process that drives meteotsunamis, there are other resonance processes. The two

most relevant are the Greenspan (1956) and the shelf (Monserrat et al., 2006) resonance.

Greenspan resonance occurs when the alongshore component of the atmospheric disturbance

velocity equals the phase speed of one of the edge wave modes. Shelf resonance takes place

when the atmospheric disturbance and the associatedocean wave have periods and/or

wavelengths equal to the resonant scales of the shelf region. Because meteotsunamis are shallow

water waves (wavelength 20 times the local depth), their amplitude can increase by refraction,

shoaling, diffraction, and harbor resonance.

One of the best known examples of meteotsunamis affecting the east coast of the United States is

the event of Daytona Beach of July 4th 1992. It produced a water surge of 3 m that caused at

least 75 minor injuries to individuals and damage to dozen vehicles near the beach. This extreme

wave was generated by a southward- propagating squall line (Churchill et al., 1995; Sallenger et

al., 1995). A comparable event occurred on March 25th 1995, when a 3-m high wave hit the west

3

coast of Florida. This ocean wave was generated by a train of atmospheric gravity waves

(Paxton and Sobien, 1998). A third example on the east coast of the United States occurred in

Boothbay, Maine, on October 28th 2008. Apart from the meteotsunami described by Paxton and

Sobien (1998), meteotsunamis in the Gulf of Mexico had not been reported but this is the first

account of their occurrence and their ubiquity there.

OBSERVATIONS OF THE METEOTSUNAMI OF MARCH 28-29TH 2014

On March 28th 2014 unusual non-tidal sea level oscillations were reported at Panama City

Beach (FL) soon after 17:20 UTC. The maximum surge (water level above predicted) reached

1.2 m, as a consequence of the combined contribution of a long period (3 days), Gaussian shape

oscillation (0.25 m high) and a fast (1 hour period) solitary wave. The fast oscillation was

associated with a sudden and abrupt change in weather marked by a squall. Atmospheric

pressure dropped by 5.5 hPa across the squall, indicating that this tsunami-like ocean wave was

meteorologically driven. When the meteotsunami hit Panama City Beach overnight, the beach

was only occupied by chairs and umbrellas. The meteotsunami did not cause harm to humans

but it caused material damage and beach erosion. This account attempts to elucidate the

conditions and mechanisms that favored the development of the meteotsunami of March 2014,

and documents the occurrence of other similar pulses in the area.

Free surface elevation and atmospheric pressure analysis

Water levels, wind velocity, air temperature and atmospheric pressure records were obtained

from archives of the US National Oceanic and Atmospheric Administration (NOAA) at 10

coastal stations between Louisiana and the western coast of Florida (Fig. 1). Sampling interval

4

for both water levels and meteorological data was 6 minutes. A high-pass Kaiser-Bessel filter (cf.

Emery and Thomson, 2001) centered at 3-h was used to separate the low and high frequency

surge and atmospheric pressure residuals. The resulting high frequency surge (black lines) and

barometric pressure (blue lines) time series illustrated the nature of these oscillations (Fig.1b).

The red dot represents the peak of high frequency surge.

The time series (Fig. 1) show the occurrence of three main meteotsunamis associated with the

passage of this specific March 2014 storm. The main meteotsunami (Wave 1) appeared in the

coastal area between Dauphin Island and Panama City Beach (stations 3 and 5 in Fig. 1). Its

generation was associated with an atmospheric pressure drop of 5.5 hPa propagating eastward at

~19-20 m/s, which is equivalent to the phase celerity of a long wave propagating over water

depths of 40 m (the depth of the shelf between Mississippi Delta and Pensacola). The

atmospheric pressure drop appeared near Dauphin Island and its shape remained unchanged

between 15:00 and 18:00 h of March 28th. The drop in atmospheric pressure damped between

Panama City Beach and Apalachicola station. At Dauphin Island the meteotsunami amplitude

was a few centimeters but increased as it propagated toward Panama City Beach, where it

attained its maximum elevation (0.85 m). Records also showed that the atmospheric pressure

drop and the meteotsunami traveled together. This indicated that the wave was bound to the

atmospheric pulse, which allowed its amplification. However, the atmospheric pressure pulse

was not visible by Apalachicola (station 7 in Fig. 1) but the meteotsunami did arrive with

diminished amplitude and with a delay of 2.5 h relative to Panama City Beach. Disengagement

of the surface wave from the atmospheric pulse in this case suggested a free wave behavior.

Panama City and Apalachicola tidal gauges are located inside estuaries and the wave must have

damped and slowed down inside the estuaries.

5

Figure 1. High frequency storm surge and atmospheric pressure disturbance time series at the tidal gauges along Alabama and West Florida coasts. a) Location of the tide gauges (the maximum observed high frequency surge height and the arrival times are indicated in the text; three main meteotsunami waves are detected associated with thestorm analyzed. The area affected by each wave is indicated with arrows of different color on the map; b) Time series of the surge (black lines) and atmospheric pressure (blue lines) measured at each of the stations considered.

A second meteotsunami (Wave 2) could also be observed from 6:40 to 7:55 AM on March 29 th

propagating eastward between New Canal (Station 1) and Shell Beach (Station 2). This wave

was related to an atmospheric pressure change of ~2 hPa. The second meteotsunami was not

detectable eastward of Shell Beach nor the atmospheric pressure pulse that generated it. A third

meteotsunami (Wave 3) occurred between 18:00 and 20:00 h on March 29 th. This was mainly

6

generated by another southward propagating atmospheric pressure drop of 2 hPa at Cedar Key

(station 8) that intensified to 3 hPa. The maximum elevation of this meteotsunami (0.5 m) was

observed at Clearwater Beach (station 9) after a drop of atmospheric pressure, suggesting that

this was a bound wave. However, the meteotsunami at Naples (station 10) appeared just before

the atmospheric pressure drop, meaning that the wave was free at this particular site.

Air temperature and wind changes associated with the passage of the squall

Intense wind shifts and temperature inversions accompanied the passage of the atmospheric

pressure drops that originated the main three meteosunamis. At Shell Beach a 3 ºC abrupt drop or

the air temperature and an increase of 15 m/s of wind intensity followed the atmospheric

pressure drop which forced Wave 2. At Dauphin Island, the main atmospheric pressure drop was

measured at 15:10 UTC on March 28th. Temperature decreased 4ºC 30 minutes after the

atmospheric pressure minimum., The wind gust increased from 10 m/s to 25 m/s and changed

from southeasterly to northerly. The meteotsunami at Clearwater Beach also coincided with a

drop in the air temperature of 2º, a local increase of the wind intensity of ~8 m/s, and again a

sudden change in the wind direction from southeasterly to northerlyThese atmospheric pressure

drops were followed by smaller amplitude gravity waves (which did not produce strong

meteotsunami waves), but the associated temperature variations (< 0.3 degrees) and wind

intensity changes (up to 15 m/s) were much weaker. Wind direction shifts associated with these

gravity waves were not that noticeable (< 5ºchanges).

Synoptic weather conditions and observed reflectivity mosaics

Surface weather maps for March 28th and 29th of (Figure 2) showed a cold front connecting two

main low pressure systems. The first cold front was initially found in the boundary between

7

Arkansas and Oklahoma and another one located in Ohio. South of the cold front, between the

coast of Louisiana and West Florida, a high precipitation area could be identified in the surface

weather maps. This precipitation was related to a surface through or prefrontal squall line

(orange dashed line). The squall line emanated for the region of strong rainfall and thunderstorm

and expanded radially (southward) from March 28th to March 29th. Prefrontal squall lines are

associated with the formation of a cold air wedge ahead of the cold front (reference?). This

results from intense thunderstorm and rainfall associated with the cold front. The cold air wedge

reinforces the convection of warm, moist, and unstable air at the head of the wedge. Energy and

moisture related to the warm air mass are released in the form of thunderstorms and showers

along the squall. The wind shifts cyclonically with the passage of these squall lines. The

temperature decreases because of the high precipitation and the pressure rises after the passage of

the squall.

Returning to the weather map, a high pressure system located in midwestern United States

propagated south-eastward and pushed the cold front southward. During March 28th there was a 8

hPa eastward atmospheric pressure gradient along the Gulf of Mexico. This pressure gradient

relaxed on the 29th and reversed to a westward gradient on the 30th.

Radar mosaics showed high reflectivity in regions of sharp changes in atmospheric pressure.

Figure 2.b depicts the reflectivity maps corresponding to the meteotsunami peaks as observed in

Pensacola, Panama City Beach, Clearwater Beach and Naples. All the maps clearly show squall

lines. The location of the squall line coincided with the position of the meteotsunami. The

linkage between abrupt changes in atmospheric forcing and surface waves was explored with

numerical experiments inspired by the events of March 28th and 29th, 2014.

8

Figure 2. a) Surface weather maps for March 28th and 29th; and b) National reflectivity mosaic maps derived for the NOAA climate service.

NUMERICAL MODELLING

Simulations of the generation and propagation of one meteotsunami (Wave 1) associated with

the March 28-29, 2014 storm were carried out with the Regional Ocean Modeling System

(ROMS). The model is three-dimensional, free surface, terrain-following and solves finite-

difference approximations of the Reynolds-Averaged Navier-Stokes equations. It uses the

hydrostatic and Boussinesq approximations (Chassignet et al., 2000; Haidvogel et al., 2000) with

9

a split-explicit time stepping algorithm (Shchepetkin and McWilliams, 2005; Haidvogel et al.,

2008). In this study, the model was implemented in a depth-integrated mode, disregarding

baroclinic flows. Because the focus of this study was to characterize the meteotsunami, the

forcing conditions were simplified by ignoring wind, heat fluxes, and astronomic tides. The

horizontal grid had a constant resolution (250 m) with rectangular cells, and was constructed

with the bathymetry derived from the Northern Gulf Coast Digital Elevation Model of NOAA.

This elevation model integrates bathymetry and topography along the northern coast of the Gulf

of Mexico. It extends from Louisiana's Mississippi River Delta to Cape San Blas (near Station 7)

in Florida.

One of the most challenging aspects of simulating the generation of a meteotsunami is

characterizing the atmospheric forcing that feeds it. This requires knowledge on the propagation

direction and the speed of the atmospheric pressure disturbance. A priori, the area of the local

minimum radar reflectivity seems to reveal the location of the atmospheric pressure drop. This

allowed reconstruction of reflectivity time series from the 5-minute resolution maps at the

location of different tidal gauges as outlined next. Figure 3a shows the time series of the

measured atmospheric pressure and radar reflectivity at Shell Beach, Louisiana, and Pensacola,

Florida. In all stations analyzed (including Panama City Beach) the pressure drop occurs

exactlyat the maximum reflectivity zone. The reflectivity gradients are thus correlated with

atmospheric pressure gradients. A tight reflectivity gradient implies a strong updraft/downdraft

interface and greater threat for an active and potentially damaging squall line.

On the basis of these connections between reflectivity and atmospheric pressure, radar

reflectivity maps were used to analyze the evolution of the surface pressure trough and

characterize the forcing field that fed the first meteotsunami (Wave 1). The forcing field was

10

simplified in the model as a radial sudden drop of 5.5 hPa traveling with the minimum

reflectivity area observed in the reflectivity maps. The modeled squall line moved eastward at

the same speed as the observed squall line (on average 19.0 m/s).

Figure 3. a) Time series of reflectivity and atmospheric pressure at Shell beach and Pensacola. b) Spatial distribution of the simplified atmospheric pressure drop on March 28 th at 16:30 and at 17:30 UTC, and c) Spatial distribution of the modeled free surface elevation on March 28 th at 16:30 and at 17:30 UTC (Panama City beach is indicated with a red dot).

11

Figure 3b shows the location of the atmospheric trough at 16:30 UTC. Figure 3c depicts the

spatial distribution of the sea surface elevation at 16:30 and at 17:30 UTC showing a wave that

moves together with the atmospheric pressure drop, meaning that the modeled meteotsunami in

the region between Dauphin Island and Panama City Beach was a bound wave. Model results

showed similar magnitudes between the observed and predicted meteotsunami, although the

modeled meteotsunami height was ~0.5 m, almost 60% the observed peak. Underestimation of

the modeled meteotsunami could be related to the simplified atmospheric pressure pulseand to

the fact that wind forcing was not included in the model. However, it can be concluded that the

main mechanism producing the meteotsunami was the atmospheric pressure drop associated with

the passage of a prefrontal squall line.

The inverse barometer effect from a 5.5 hPa atmospheric pressure drop should produce a 0. 055

m increase of the free surface elevation. Numerical results indicated that Proudman resonance

over the shelf generated a 0.1 m meteotsunami offshore of Panama City Beach (amplification

from Proudman resonance was 1.8, i.e., 0.055 m to 0.1 m). This value was in agreement with the

estimates obtained with the analytic model of Hibiya and Kaijura (1982), considering that the

distance traveled by the squall line over the shelf was 210 Km. As the meteotsunami propagated

onshore, still forced by the atmospheric pressure disturbance, the oblique incidence of the ocean

wave with respect to the local bathymetry produced wave refraction and shoaling. When the

atmospheric pressure pulse arrived at the coast (in the area of Seagrove Beach) westward of

Panama City Beach around 5 PM, the maximum modeled meteotsunami wave height (including

refraction and shoaling) was 0.3 m. Shoaling caused the amplification factor in this onshore

propagation stage to become 5.5 (relative to the inverse barometer) and 3 (relative to the 0.1 m

offshore wave).

12

After this onshore propagation stage, the squall line continued moving parallel to the coast. Over

a span of 30 minutes the meteotsunami intensified to 0.5 m (amplification factor from Greenspan

resonance of 1.7, from 0.3 to 0.5 m), while the squall line celerity remained unmodified at ~19

m/s. The meteotsunami wave energy was maximum at the shore and the energy exponentially

decreased in the cross-shore direction. The bathymetry in this coastal area is characterized by a

slope of 0.0013. In the model results, different oscillation modes could be identified in the coast

after the passage of the squall line. All these facts suggest that the fastest increase of the

meteotsunami height occurred after the squall began to propagate parallel to the coast. In this

along-coast propagation stage the energy increase was mainly caused by Greenspan resonance

over the sloping beach, rather than by Proudman resonance. Once the squall line reached the

inlet in Saint Andrew Bay (Panama City station inside), the meteotsunami lost height because of

the coast discontinuity at the inlet.

DISCUSSION AND CONCLUSIONS

Several meteotsunami studies (e.g. Pattiarachi et al. 2014, Pellika et al. 2014) consider Proudman

resonance as the primary process by which ocean waves forced by atmospheric pressure are

amplified in the open ocean. Less documented are meteotsunamis in which Greenspan resonance

is relevant. The contribution of Greenspan resonance depends on the beach slope, the shape of

the atmospheric pressure disturbance, the pressure anomaly intensity, and the moving speed of

the pressure system. As recently explained by Seo and Liu (2014), resurgent edge waves from a

given atmospheric pressure disturbance that moves in the along shore direction, are possibly

triggered in the range 0.7 V 1, where V is the normalized pressure disturbance moving

speed. In the case of the Panama City Beach meteotsunami, the length of the atmospheric

pressure disturbance was computed considering the period of the pressure drop (1 hour) and the

13

celerity at which it was moving (19 m/s). In this case the atmospheric pressure drop length was

70 km. Given that the bed slope in Panama City Beach is 0.0013, the resulting non-dimensional

propagation velocity for the atmospheric pulse producing that meteotsunami ranged between 0.6

and 0.75, which may fall in the range proposed by Sea and Liu (2014). Although Proudman

resonance has been identified as the main process causing resonance in other meteotsunamis

(e.g. the Daytona Beach meteotsunami in 1996), Greenspan resonance can be a considerable

contributor, especially if the squall line propagates in the alongshore direction in areas with

relatively constant beach slopes. In the case of the Panama City Beach meteotsunami, the

amplification factors from Proudman and Greenspan resonance were similar. However,

Proudman resonance acted over 250 km, compared to the 40 km of Greenspan resonance action.

Numerical results revealed that bathymetric discontinuities along the coast, such as inlets and

capes, can disrupt Greenspan resonance and thereby reduce the meteotsunami height.

Because of the simplified atmospheric forcing considered in the numerical simulations, the

simulated meteotsunami was 0.35 m smaller than observed. This could be related to the

simplified atmospheric pressure drop but it could also be explained by the omission of the

sudden change of wind stress associated with the passage of the squall. Coupled atmospheric and

oceanic simulations with the COAWST modeling system (Warner et al., 2010) are underway to

analyze if mesoscale models such as WRF (Weather Research Forecast system) can correctly

simulate the development of pre-frontal squall lines for a better characterization of the observed

meteotsunami. Although the presence of squalls can be identified with radar data, a tight relation

between reflectivity and atmospheric pressure is needed. This study indeed found a qualitative

relation, for the first time. However, further analysis, including other meteotsunami events, is

14

needed to ascertain whether the reflectivity data can provide the detailed characteristics of

atmospheric pressure drops and serve as an accurate predictor of meteotsunami occurrence.

Meteotsunamis have affected the Gulf of Mexico coast of Florida in several more occasions than

that in late March of 2014. Exploration of sea level data since 1996 at Naples station indicated

that Meteotsunamis are ubiquitous in the northern and eastern Gulf of Mexico: ten events of high

frequency surges (period < 3h, surge height > 0.8 m) were detected. Out of the 10 events, 6 were

associated with the passage of tropical storms or hurricanes between the months of June and

October (Figure 4a). But the other 4 events were associated with 4 extreme surges during winter

months. Figure 4b shows the reflectivity conditions during the occurrence of 3 of the 4 surge

events. In all cases, the surge was associated with a high reflectivity band oriented perpendicular

but propagating parallel to the coastline. The high reflectivity region indicated the presence of a

squall. These observations suggested that meteotsunamis are more ubiquitous than previously

thought. Three more meteotsunamis were measured recently in the analysis region. On

November 24th, 2014. the maximum observed meteotsunamis exceeded the 0.8 m elevation at

Panama City Beach. On November 17th maximum elevation of 0.6 m was measured at

Clearwater Beach (Figure 4.c). These meteotsunamis were generated under very similar

atmospheric forcing conditions to those of late March, and were associated with squall lines

propagating parallel to the coastline (Figure 4.d). It is evident that the development of

meteotsunamis will depend on the direction of propagation of a squall, the celerity of the squall,

the pressure gradient associated with the squall, the wind velocity in the squall, and the

bathymetric slope at the coast.

15

Figure 4. a) Storm surge events exceeding 0.8 m at Naples (FL) tidal gauge. The red boxes indicate those events not associated with Tropical storms; b) Atmospheric reflectivity maps associated with the major winter surges observed at Naples; c) surge time series during the passages of the main meteotsunamis identified in 2014, and d) atmospheric reflectivity maps associated with the generation of the meteotsunamis.

Although meteotsunamis can contribute to extreme surge levels and to increased hazards to

coastal communities, their generation mechanisms and their forecast feasibility are still open

questions and need further research.

ACKNOWLEDGEMENTS (NEED TO INCLUDE)

REFERENCES

Rabinovich AB (2009) Seiches and harbour oscillations. In: Kim YC (ed) Handbook of coastal and ocean engineering. World Scientific, Singapore, pp 193–236.

Paxton CH, Sobien DA (1998) Resonant interaction between an atmospheric gravity wave and shallow water wave along Florida’s west coast. Bull Am Meteorol Soc 79:2727–2732.

16

Chassignet, E.P., H.G. Arango, Dietrich, D., Ezer, T., Ghil, M., Haidvogel, D.B., Ma, C.-C., Mehra, A., Paiva, A.M. & Sirkes, Z., 2000. DAMEE-NAB: The Base Experiments, Dynamics of Atmospheres and Oceans, 32, 155-183.

Haidvogel, D.B., H.G. Arango, K. Hedstrom, A. Beckmann, P. Malanotte-Rizzoli, and A.F. Shchepetkin, 2000. Model Evaluation Experiments in the North Atlantic Basin: Simulations in Nonlinear Terrain-Following. Dynamics of Atmospheres and Oceans, vol. 32, pp. 239–281.

Pattiaratchi, C. and E.M.S., Wijeratme, 2014. Observations of meteorological tsunamis along the south-west Australia. Natural Hazards, 74: 281-303. DOI 10.1007/s11069-014-1263-8.

Monserrat S, Vilibic´ I, Rabinovich AB, 2006. Meteotsunamis: atmospherically induced destructive ocean waves in the tsunami frequency band. Nat Hazards Earth Syst Sci 6(6):1035–1051.

Greenspan HP, 1956. The generation of edge waves by moving pressure distributions. J Fluid Mech, 1(6):574–592.

Haidvogel, D. B. , Arango, H.G., Budgell, W. P., Cornuelle, B. D., Curchitser, E., Di Lorenzo, E., Fennel, K., Geyer, W. R., Hermann, A. J., Lanerolle, L., Levin, J., McWilliams, J. C., Miller, A. J., Moore, A. M., Powell, T. M., Shchepetkin, A. F., Sherwood, C. R., Signell, R. P., Warner, J. C. & Wilkin, J., 2008. Regional Ocean Forecasting in Terrain-following Coordinates: Model Formulation and Skill Assessment. Journal of Computational Physics 227, 7, (3595-3624).

Shchepetkin, A.F., and McWilliams, J.C., 2005. The regional ocean modeling system (ROMS): a split-explicit, free-surface, topography-following-coordinates ocean model. Ocean Modelling 9, 347-404.

Hibiya T, Kajiura K (1982) Origin of the abiki phenomenon (a kind of seiche) in Nagasaki Bay. J Oceanogr

Soc Jpn 38:172–182.

Pellikka, H., Rauhala, J., Kahma, K.K., Tapani, S., Boman, H., and Kangas, A., 2014. Recent Observations of meteotsunamis on the Finnish coast. Natural Hazards, 74, 197-215. DOI 10.1007/s11069-014-1150-3.

Seo, S-N. and Liu, P.L.F., 2014. Edge waves generated by atmospheric pressure disturbances moving along a shoreline on a sloping beach. Coastal Engineering, 85, 43-59. DOI 10.1016/j.coastaleng.2013.12.002.

17