**4.2. Radar detection of the 2013 US East Coast meteotsunami**

and (b) beyond the shelf, where depths change slowly, from 760 m at 25 km to 1510 m at 40 km, the Green's Law approximation described in Section 2.3.2 should be valid. However, the observed velocity is nearly constant, which contradicts Eq. (7). These effects may be due to

An unusual storm system moved eastward across the US on June 13, 2013, commonly called a "derecho," and appears to have launched a meteotsunami that impacted the US East Coast. The existence of the meteotsunami was confirmed by several of the 30 tide gauges along the East Coast up through New England and was seen as far away as Puerto Rico and Bermuda. The event, which occurred during daylight hours, attracted widespread attention after several media reports were released focusing on local impacts including people being swept off a breakwater at Barnegat Light, New Jersey, some damage to boat moorings, and minor

Meteotsunamis generally do not have sufficient heights/energies to cause catastrophic loss of life, as do severe seismic tsunamis, although damage to harbors and coastal structures is common. The June 13, 2013 event, however, attracted significant attention among many agencies and scientific groups, probably due to its proximity to heavily populated areas.

A meteotsunami is generated by an atmospheric pressure disturbance traveling across the sea. An atmospheric anomaly (a low‐ or high‐pressure center) will produce a small peak or trough moving at the same speed on the sea surface beneath it. This results in a freely propagating surface wave that increases in amplitude when the speed of atmospheric anomaly *vaa* matches the shallow‐water wave phase velocity *vph(d)*. This is known as Proudman resonance, see [17]. The speed *vaa* of the June 13, 2013 derecho was about 21.1 m/s [15]. Substituting this value for *vph(d)* into Eq. (5), it follows that the onset of the independent wave occurs at a depth *d* equal

This meteotsunami was unusual because it was generated by a frontal pressure anomaly traveling offshore. Yet coastal sensors including HF radars indicate that the meteotsunami approached the coast. Numerical models [8] indicate that a strong reflection occurred at the shelf edge about 110–120 km offshore, where the depth decreases from 100 to 1200 m over a distance of 20 km. The reflection is greater when a wave interacts with a drop‐off rather than a step‐up with the same slope. Data from New Jersey radars confirm the existence of a wave reflected from the shelf edge back toward the coast. This wave was also detected by coastal

To explain these results, we consider the interaction of the tsunami with a hard boundary, assuming a single pulse of water approaching the coast, that is, a traveling wave. The forward velocity is maximum at the wave crest. As a boundary is approached, there is a hard reflection:

**4. Radar observations of the 2013 US East Coast meteotsunami**

**4.1. Origin of meteotsunamis and nature of the June 13, 2013 event**

to 45 m, which lies about 60 km off the New Jersey coast.

signal aliasing, as discussed later in Section 7.

inundation.

92 Tsunami

tide gauges.

We analyzed data sets from three SeaSonde HF radar systems located in New Jersey: BRNT, BRMR, and BELM. Radar transmit frequencies and range cell widths were approximately 13.5 MHz and 3 km, respectively. Radar results were compared with data from NOAA tide gauges at Atlantic City and Sandy Hook, New Jersey. **Figure 12** shows the locations of the radars and tide gauges, and the offshore bathymetry. The meteotsunami height at the neighboring DART buoy, located about 240 km to the east, was only 5 cm [15].

**Figure 12.** The radar stations at Brant Beach (BRNT), Brigantine (BRMR), and Belmar (BELM); the NOAA tide gauges at Sandy Hook (tide gauge 1) and Atlantic City (tide gauge 2) and the offshore bathymetry contours, with depths in meters.

**Figure 13.** NOAA tide gauge observations June 13, 2013 at Atlantic City, NJ.

Readings show a maximum negative meteotsunami signal at approximately 18:42 UTC, indicated by the sharp water‐level decrease. This is followed at approximately 22:00 UTC by a sharp increase in water level and subsequent oscillations.

As described in Section 3.3, the radar coverage area is divided into rectangular area bands 2‐ km wide and approximately parallel to the depth contours. Radial vectors within each area band were resolved parallel and perpendicular to the depth contour. These velocity compo‐ nents are then averaged over the bands.

**Figure 14** shows time series of four perpendicular band velocities from BRNT and BRMR and the corresponding *q*‐factors, obtained from the four bands [6].

The arrival of the meteotsunami is signaled by a marked decrease in the perpendicular band velocity component, indicating an outflow, followed by correlation between different area bands. The parallel component did not display the tsunami signature. The water level measured by closest tide gauge at Atlantic City decreases when the tsunami arrives, as shown in **Figure 13**, also indicating an outflow of water.

The tsunami signal at BELM was far less, which is consistent with tide gauge measurements at Sandy Hook, 30 km to the north, which barely registered the tsunami arrival.

About 4 h later, after 22:00 UTC, BRNT velocities first increase and then sharply decrease, as is also shown by the Atlantic City tide gauge (see **Figure 13**). This effect was not seen at BRMR or BELM.

Atlantic City tide gauge data obtained from the NOAA website [16] are shown in **Figure 13**.

Readings show a maximum negative meteotsunami signal at approximately 18:42 UTC, indicated by the sharp water‐level decrease. This is followed at approximately 22:00 UTC by

As described in Section 3.3, the radar coverage area is divided into rectangular area bands 2‐ km wide and approximately parallel to the depth contours. Radial vectors within each area band were resolved parallel and perpendicular to the depth contour. These velocity compo‐

**Figure 14** shows time series of four perpendicular band velocities from BRNT and BRMR and

The arrival of the meteotsunami is signaled by a marked decrease in the perpendicular band velocity component, indicating an outflow, followed by correlation between different area bands. The parallel component did not display the tsunami signature. The water level measured by closest tide gauge at Atlantic City decreases when the tsunami arrives, as

The tsunami signal at BELM was far less, which is consistent with tide gauge measurements

About 4 h later, after 22:00 UTC, BRNT velocities first increase and then sharply decrease, as is also shown by the Atlantic City tide gauge (see **Figure 13**). This effect was not seen at BRMR

at Sandy Hook, 30 km to the north, which barely registered the tsunami arrival.

**Figure 13.** NOAA tide gauge observations June 13, 2013 at Atlantic City, NJ.

a sharp increase in water level and subsequent oscillations.

the corresponding *q*‐factors, obtained from the four bands [6].

shown in **Figure 13**, also indicating an outflow of water.

nents are then averaged over the bands.

or BELM.

94 Tsunami

**Figure 14.** The meteotsunami arrival observed by radars BRNT and BRMR: band velocity components and the corre‐ sponding *q*‐factors plotted against hours from 00:00 June 13, 2013. BRNT: (a) 6–8 km (blue), 8–10 km (red), (b) 10–12 km (black),12–14 km (green), (c) Corresponding *q*‐factors. BRMR: (d) 2–4 km (blue), 6–8 km (red), (e) 14–16 km (black), 20– 22 km (green), (f) Corresponding *q*‐factors.

To demonstrate more clearly the meteotsunami velocity trough as it approached the coast, BRNT band velocities were further processed as follows: the band velocities were first detrended over time, removing effects with time scales longer than 1.5 h, such as those due to tides. The detrended band velocities were then low‐pass filtered and, to further reduce noise, averaged over two adjacent bands.

**Figure 15** shows the smoothed velocities plotted as a function of time vs. range from shore, the dashed line indicating the progression of the first tsunami trough. Tsunami hindcast modeling [15] confirms this time‐distance progression of the meteotsunami as it moved toward shore.

**Figures 14** and **15** show that the tsunami arrived first at the most distant ranges and progres‐ sively later moved toward the coast. To compare these results with theory, the tsunami arrival time at BRNT was calculated using Eq. (10), based on an initial detection at range 23 km. The bathymetry contours offshore from BRNT shown in **Figure 12** were approximated by parallel contours, giving depth as a function of distance. As discussed in Section 2, this approximation is valid, as the tsunami is not affected by perturbations in depth with spatial scales far less than its wavelength. This analysis assumes no coastal boundary and results are expected to differ somewhat from radar‐observed arrival times, as the orbital velocities are affected by shallow water.

**Figure 15.** BRNT band orbital velocities plotted as a function of time, dstance from shore: (a) 6 km, (b) 10 km, (c) 14 km, (d) 18 km, and (e) 22 km. The progress of the first tsunami trough minimum is shown by the dashed line.

**Figure 16(a)** shows the different arrival times plotted vs. distance from shore. For the radars, arrival time was defined to correspond to the minimum tsunami velocity value; for the tide gauge, it was defined to correspond to the minimum water level. The solid curve shows the arrival times at BRNT calculated using Eq. (10), using the approximate depth vs. distance from shore plotted in **Figure 16(b)**.

its wavelength. This analysis assumes no coastal boundary and results are expected to differ somewhat from radar‐observed arrival times, as the orbital velocities are affected by shallow

**Figure 15.** BRNT band orbital velocities plotted as a function of time, dstance from shore: (a) 6 km, (b) 10 km, (c) 14 km,

**Figure 16(a)** shows the different arrival times plotted vs. distance from shore. For the radars, arrival time was defined to correspond to the minimum tsunami velocity value; for the tide gauge, it was defined to correspond to the minimum water level. The solid curve shows the arrival times at BRNT calculated using Eq. (10), using the approximate depth vs. distance from

(d) 18 km, and (e) 22 km. The progress of the first tsunami trough minimum is shown by the dashed line.

shore plotted in **Figure 16(b)**.

water.

96 Tsunami

**Figure 16.** (a) The arrival time of the first tsunami trough observed by the radars plotted vs. distance from shore. Blue: BRNT, black: BRMR, green: BELM, red asterisk: arrival time at the Atlantic City tide gauge, and magenta: tsunami ar‐ rival time calculated from Eq. (10). Time is measured in hours from June 13, 2013, 00:00 UTC. (b) Depth plotted vs. distance from BRNT perpendicular to depth contours.

It can be seen from **Figure 16(a)** that the meteotsunami arrived 23 km from the coast at BRNT and BRMR at about the same time. It then traveled toward shore at approximately 30 km/h. The meteotsunami arrived about 14 min later 23 km offshore from BELM. The two BELM readings indicate that it then moved toward shore at a higher speed, probably due to deeper water near the Hudson Canyon. The tsunami arrived at the Atlantic City tide gauge 47 min after it was first observed by the BRNT radar.

The initial velocity observed by the radars was offshore, indicating a "trough" on the ocean surface. This was also observed by closest tide gauge at Atlantic City. However, as shown in **Figure 15**, the tsunami wave itself approached the coast due to a strong reflection occurring at the shelf edge 110–120 km from shore, see Section 4.1.

The meteotsunami was detected by the radars 23 km from the coast. It arrived at the shore 47 min later, as indicated by the tide gauge measurement of water level shown in **Figure 13**. The measured tsunami height was approximately 50 cm. These observations suggest that for similar tsunami height and bathymetry conditions, HF radar can provide a three‐quarter hour warning alert before the wave strikes the shore.
