3.1 TC track and intensity
The observed best track of Hurricane Irene (2011) is shown in Figs. 1 and 2a. UWIN-CM captured well the storm track, especially in the cross-track direction, which is critical for storms that skirt the coast (inset of Fig. 2a). In terms of intensity, the model storm was significantly stronger (~ 30 kt) than the best track during the latter half of 25 Aug. and early on 26 Aug.; nevertheless, the model storm was within 5–10 kts of the best track intensity at the time of landfall (Fig. 2b).
Tropical Storm Lee (2011) formed on 2 September in the northern Gulf of Mexico, from a pre-existing broad cyclonic circulation. Lee’s forward motion in the Gulf was slow, and its track meandered northward before making landfall in Louisiana in the morning of 2 September with an intensity of 45 kts (Figs. 1c and 2c). The UWIN-CM simulated track was similar to the best track, though the storm tracked further north than observed inland on 5—6 Sep. (Fig. 2c). In terms of intensity, the model storm initially quickly intensified to 50 kts., around 12 h faster than the best track, and it maintained higher intensity while over land on 5—6 Sep (Fig. 2d). The simulated intensity at landfall was within 5 kts of the best track. Because the Mid-Atlantic coast was affected by the remnants of Lee, the intensity of Lee at landfall was less important than for Irene.
3.2 Surface wind, waves, and air-sea temperatures
Figure 3a and 3b show the maximum surface wind speed simulated by UWIN-CM for Irene and Lee. As Hurricane Irene made landfall in North Carolina and moved northwards skirting the coast, a ~ 200 km wide swath of winds > 30 m s-1 directly impacted the North Carolina coast (Fig. 3a). The coasts of Virginia, Maryland, and New Jersey experienced somewhat lower wind speeds in the 20–30 m s-1 range, as the storm was in a weakening trend, and the winds shifted eastward further away from the storm center. Wind speeds in Chesapeake Bay peaked at around 25 m s-1. In contrast, the remnants of Tropical Storm Lee were associated with more moderate wind speeds of 10–15 m s-1 (Fig. 3b). Irene’s stronger winds led to much larger surface waves than Lee (Figs. 3c, d). During the passage of Hurricane Lee, the UWIN-CM significant wave height (SWH) reached 12–14 m. The peak SWH occurred 100–300 km offshore. The SWH within Pamlico Sound, Albemarle Sound, Chesapeake Bay, and Delaware Bay were around 4 m. These waves were generated locally, as these waters were protected from the open ocean swells. Corresponding with the much weaker surface wind forcing, SWH remained < 2 m during the passage of Lee.
The UWIN-CM simulation captured well the surface meteorological and oceanographic conditions observed by the NDBC buoys. Figure 4 shows the observed and simulated conditions at a buoy location that was impacted by the inner core region of Hurricane Irene. This buoy had the most complete data of the NDBC buoys that experienced Irene’s inner core. The simulated surface pressure nearly identically reproduced the observations (Fig. 4a). Wind speed and direction were also well reproduced (Figs. 4b, c). Significant wave height was ~ 2 m higher during the passage of Hurricane Irene and lower than the observations after the passage of Irene (Fig. 4d). During the Lee portion of the simulation, significant wave height peaked on 9 September, ~ 12 h earlier than the observations, and dropped down much faster than the observations. This second peak in waves was due to the swell from Hurricane Katia to the east, which is not a focus of this study. Surface air temperature and sea surface temperature were simulated well before Hurricane Irene, but poorly in the post-Irene period (29 Aug. – 1 Sep.; Figs. 4e, f). SST cooling of > 5°C was observed by the NDBC buoy, but this was not reflected in UWIN-CM (Fig. 4f). This is likely related to river plumes extending out into the sea, which is not resolved by UWIN-CM. Because the air temperatures are in approximate equilibrium with the SST, the air temperatures at the NDBC location were also biased high in the post-Irene period (Fig. 4e). For the Lee portion of the simulation, the ocean was re-initialized, and the UWIN-CM SST and surface air temperatures agreed better with the observations.
Figure 5 shows the comparison between UWIN-CM and surface observations within Chesapeake Bay. Sea level pressure was nearly identical to the observations; however, the peak wind speed for Hurricane Irene (29 August) was high in UWIN-CM (Fig. 5a, b). Both UWIN-CM and observations showed southerly winds of 5–10 m s-1 in association with the remnants of Lee (Fig. 5b, c). (Unfortunately, wind observations were not available for part of Irene and for Lee after 8 Sep.) While the peak wind speeds in Chesapeake Bay during the passage of Irene were similar to the open ocean (Fig. 4b and 5b), the significant wave heights were much lower in Chesapeake Bay, which is due to the limited fetch there (Figs. 4d and 5d). Surface air temperature and sea surface temperature were simulated well by UWIN-CM (Fig. 5e, f).
While the accuracy of the simulations varied from location to location, the overall comparison with the 12 NDBC buoys shown in Fig. 1b was favorable (Fig. 6). Wind speeds were simulated with a bias of 0.6 m s-1 and RMS error of 2.4 m s-1, with the largest errors at wind speeds > 15 m s-1, which were associated with the nearby passage of Hurricane Irene (Fig. 6a). This is consistent with the stronger intensity of the simulated Hurricane Irene compared with the best track (Fig. 2b). Similarly, significant wave height was simulated with a bias of 0.2 m and RMS error of 0.8 m, with the highest biases of for wave heights > 3 m (Fig. 6b). The bias at high significant wave height is consistent with the high wind bias for winds > 15 m s-1. Air temperature and SST were both simulated well, with biases well within 0.5°C (Figs. 6c, d). While most observations were simulated nearly identically, the simulation failed to capture the nearshore SST cooling (Fig. 6d). Simulated air temperature had a somewhat lower RMS error (1.3C) compared with SST (1.6°C; Fig. 6c).
3.3 Sea surface height and water level above pre-storm conditions
The storm impact on water heights is reflected in the maximum high tide water height relative to the pre-storm high tide levels. The pre-storm period is affected by upper ocean structures as well as seasonal and interannual variability of sea surface height. For Irene, 0000 UTC 24 August to 0000 UTC 26 August is taken as the pre-storm reference. For Lee, it is 0000 UTC 4 September to 0000 UTC 6 September, which is the period with lowest high tides between Hurricane Irene and Tropical Storm Lee. For observations, both NOAA tide gauges and USGS estuary sites wereused for this evaluation. For UWIN-CM, the pre-storm reference was calculated independently for each grid point.
For Hurricane Irene, the greatest water height changes from pre-storm conditions, 1.2 m up to 1.6 m, were observed in the New York/New Jersey metro area and Long Island (Fig. 7a). Somewhat lower increases were observed over the New Jersey coast and Delaware Bay. In Chesapeake Bay, the storm induced water heights > 0.7 m were in lower Chesapeake Bay, with little effect in the upper bay. Water heights also increased on the west side of the North Carolina outer banks. UWIN-CM successfully simulated the general trend of the largest water height perturbations being in the NY/NJ metro area, lower Chesapeake Bay, and NC outer banks, though the magnitude was somewhat lower in Long Island Sound (Fig. 7b). In the UWIN-CM simulation, the water height increases from pre-storm were limited to within 50–100 km of the coast, and they were greater closer to the coast. It is not possible to map out the spatial extent of storm-induced water height increase using the observational data, and UWIN-CM provides information along the coast in-between observation sites.
For Tropical Storm Lee, the observed water height increase was much lower than for Irene (Fig. 7c). It was near zero along the coast, with increases of ~ 0.5 m in the upper Chesapeake Bay estuaries, Potomac River, and Delaware River. The UWIN-CM simulation similarly showed little water height increase (Fig. 7d). Unfortunately, the UWIN-CM simulation did not provide information for the Potomac and Delaware Rivers. Nevertheless, it did miss water height increases of ~ 0.3 m in the upper Chesapeake Bay and in Delaware Bay. These observed water height increases were due to water discharge from rivers (Sec. 3.5.3).
The water height perturbations associated with Hurricane Irene varied in time as the storm moved northward skirting the coast. When Irene was over North Carolina and Virginia, the mid-Atlantic coast experienced onshore winds, and increases in water height were observed from the mouth of the Chesapeake Bay to Long Island (Fig. 8a). When the storm moved northward to the Maryland/Delaware coast 12 h later, water heights increased in the NY/NJ region (Fig. 8b). Notably, the water heights in lower Chesapeake Bay and at the mouth of the bay had not fallen, despite the offshore flow. This was due to the set-up of northerly flow across the length of Chesapeake Bay, which pushed water from the upper bay towards the lower bay. As the storm moved further north, the water heights at the mouth of the Chesapeake Bay, along the Maryland coast, and in Delaware Bay dropped (Fig. 8c-d). Notably, during the same period, the North Carolina outer banks experienced an extended period of westerly winds, which pushed the water up against the west sides of the barrier islands.
Figure 9 shows the contrast in Chesapeake Bay water heights between Hurricane Irene and the remnants of Tropical Storm Lee. As Irene passed to the east, strong (> 20 m s-1) northerly winds occurred over Chesapeake Bay. Meanwhile, compared with pre-storm high tides, water heights were over 1 m higher in lower Chesapeake Bay and 1 m lower in the upper bay (Fig. 9a). Note that this occurred concurrently with the open ocean storm surge from Irene. Two factors affected this: 1) the storm surge at the mouth of the bay prevented water from leaving the bay, and 2) the northerly winds pushed water from the upper bay towards the lower bay. In contrast, the remnants of Lee brought a period of moderate southerly winds (10 m s-1) to Chesapeake Bay with water heights higher (lower) than the pre-storm reference in the upper (lower) bay (Fig. 9b). The southerly winds would tend to push the water from the lower bay into the upper bay, causing a local storm surge effect, similar to storm surges observed on large lakes such as the Great Lakes. Outside of the bay, Ekman transport associated with the winds blowing parallel to the coast contributed to the lower water levels there, including at the mouth of Chesapeake Bay (Clémente-Colon and Yan 1999).
3.4 Water height above astronomical tides
Coastal flooding hazard is associated with tides that are above the typical astronomical tide range, which the coastline has had time to adapt to. Figure 10 shows the maximum water height obtained from observations and the UWIN-CM simulation for Hurricane Irene and Tropical Storm Lee. Generally, higher water heights occurred from Irene than from Lee. However, in the Delaware River, abnormally high water elevations occurred from both Irene and Lee. The high water heights were limited to the lower portion of Chesapeake Bay due to the northerly winds over the bay from Irene. The simulation captured the general characteristics of where the highest water heights occurred, but they differed from observations somewhat in the magnitudes (Fig. 10b, d). Note that simulated water heights are not available at coastal river locations, which are not included in HYCOM.
Another aspect of storm water impact is the duration of the event. The duration of abnormally high water varied greatly at the NOAA tide gauges (Fig. 11). The longest duration of tides > 0.3 m above astronomical predictions occurred along the Delaware River (Fig. 11a, c). The duration was up to 3–5 days at some inland locations. Notably, the inland high water duration tended to be higher from Lee than from Irene, despite Lee having no significant wind forcing (Fig. 3b). In fact, long durations (> 48 h) of high water occurred in the upper Chesapeake Bay and Potomac River from Lee, but this did not occur from Irene. Along the immediate coats and in Delaware Bay, the opposite occurred—the duration of high water was significantly greater from Irene. The UWIN-CM model was able to capture the longer duration from Irene along the immediate coast, which is due to the wind-induced storm tide; unfortunately, it could not depict the high water further inland along the river, which was associated with river discharge (Fig. 11b, d).
The high-water levels along the immediate coasts of North Carolina, Maryland, New Jersey, and New York were associated with the wind-induced storm tide. Additionally, high water levels above astronomical tides were observed in Chesapeake Bay, Delaware Bay, and further inland along the Delaware and Potomac Rivers. These two components of coastal flooding hazard are illustrated further below.
3.5 Compound effects on coastal flooding
Here we examine change of water height at various coastal locations over three major river basins in connection with the Delaware Bay and Chesapeake Bay during Hurricane Irene and the remnant of Tropical Storm Lee. Effects of rain, wind, storm surge, and river-stream discharge associated with the TCs on the water height and flooding are examined using both observations and UWIN-CM simulations at each station.
3.5.1 Storm surge contribution
The wind-induced storm surge/tide is evident as an onshore surge of water ahead of the storm followed by an offshore push after the storm, which depresses the water levels. Because it is tied to the storm’s wind forcing, it occurs over time scales of 1–2 days. The effect of the storm tide is primarily seen along the exposed coastlines, but it can also affect bays, estuaries, and the lower reaches of coastal rivers which are connected to the ocean and sensitive to ocean tides but sheltered the open ocean. Figure 12 illustrates this for the Delaware Bay and lower reaches of the Delaware River. At Brandywine Shoal Lighthouse (NOAA tide gauge 8555889), which is the NOAA tide gauge most close to the mouth of the bay, the highest tide of the study period (2.7 m above MLLW, 1.0 m above MHHW) occurred at 2354 UTC 27 August 2011 (Fig. 12a). This was significantly higher than the astronomical high tide prediction of 1.8 m above MLLW (0.2 m above MHHW). On the back side of the storm, the low tide at 19 UTC on 28 August was 0.4 m below MLLW, compared with an astronomical predicted low tide of near MLLW. At this time, the storm had moved northward into New York, and winds in Delaware Bay were offshore (Fig. 8d).
The storm had a similar effect on the tides at Ship John Shoal (NOAA tide gauge 8537121) and Reedy Point (8551910), which is located further into Delaware Bay (Figs. 10b, c). The 1/25th deg. resolution of the UWIN-CM ocean is able to resolve water heights within the Delaware Bay. Using the grid points nearest to the NOAA tide gauges, the model simulated well the evolution of the tides at these locations, with the exception of the intermediate high tide at 13 UTC, which was under-predicted by UWIN-CM. The storm tide was also evident in the lower reaches of the Delaware River, including Philadelphia (NOAA 8545240) and Newbold (NOAA 8548989), as a pulse of high water concurrent with the passage of the storm (Figs. 10d, e). The high tide reached 3.2 m above MLLW (1.0 m above MHHW) at Philadelphia, PA at 0536 UTC 29 August and 3.7 m above MLLW (1.1 m above MHHW) at 0642 UTC 29 August at Newbold, PA. However, unlike in Delaware Bay, the post-storm tides did not fall below MLLW at these locations. The delayed timing along the Delaware River of 5–7 h from the high tide at Brandywine Shoal Lighthouse illustrates the upriver propagation of Hurricane Irene’s storm surge pulse. Finally, the storm surge pulse was not evident at Trenton, NJ, which is at a higher elevation and was not affected by the ocean tides (Fig. 12f).
The contribution of the wind-induced storm surge itself can be determined as the difference between the observed tides and the astronomical predictions. This peaked during the low tide, several hours after the highest high tide was observed. The observed low tides at that time were the highest low tides during the study period. At Brandywine Shoal, this low tide was 1.1 m above MLLW at 0642 UTC 29 August (Fig. 12a). This corresponded with the passage of the storm as indicated by the shift in strong winds (> 20 m s-1) from the easterly to from the west-northwest, which was well simulated by UWIN-CM. The flooding impact would have been greater in the Delaware Bay and Delaware River area if the storm surge had peaked coincidentally with the high tides rather than the low tide.
In contrast to Delaware Bay, storm surge from Hurricane Irene peaked during the high tide in lower Chesapeake Bay (Fig. 13a, b). This was shown most prominently at Sewells Point, VA (NOAA 8638610) near the mouth of the bay, where the high tide at 0000 UTC 28 August reached 2.3 m above MLLW (1.4 m above MHHW; Fig. 13a). At Lewisetta, VA (NOAA 8635750), the highest high tide occurred at 0336 UTC 28 August, and it peaked at 1.4 m above MLLW (0.9 m above MHHW; Fig. 13b). The storm surge situation in the lower Chesapeake Bay was somewhat different than Delaware Bay. Unlike at the mouth of Delaware Bay, strong forcing did not occur perpendicular to the coast directed into the bay, but rather, strong northerly winds occurred along the length of the bay, piling up the water in the southern portion (see Fig. 9b). Because the winds did not reverse to the opposite direction, e.g., southerly, the post-storm low tides did not fall much below MLLW near the mouth of the bay. In contrast to the lower bay, the passage of Irene in the upper bay was associated with lower water heights (Figs. 13c, d, and e). Similarly, along the Potomac River at Washington, DC (NOAA 8594900), abnormally low water heights were observed during the passage of Irene (Fig. 14b). A prominent storm surge was not observed; however, the post-storm low tide there fell to 0.2 m below MLLW at 1848 UTC 28 August (Fig. 14c). Similar to Trenton, NJ along the Delaware River, the storm surge had no effect at the higher reaches of the Potomac and Susquehanna Rivers, at higher elevations not affected by ocean tides (Figs. 13f and 14c). The coastal flooding hazard within Chesapeake Bay and its tributary river systems is complex and highly dependent on the local geography.
3.5.2 TC RAINFALL
Heavy rainfall associated with Hurricane Irene from 27–28 August and subsequent remnant of Tropical Storm Lee from 5–10 September in the Mid-Atlantic region is a main contributor to the inland flooding as shown in Kerns and Chen (2022). The effects of heavy rain on water height are evident at the stations farther from the open ocean in Figs. 12–14. The catchment of rainfall over the Delaware, Susquehanna, and Potamic River basins during these widespread heavy rain events contribute to the increased river and stream discharge and elevated water height in all three basins.
3.5.3 River discharge
Effects of river and stream discharge on inland flooding during Hurricane Irene and Tropical Storm Lee have been discussed in detail in Kerns and Chen (2022), especially the record flooding occurred on the Susquehanna River. In addition to local flooding along the river banks, the discharge of fresh water into the coastal zone presents a compounding impact following the storm surge/tides. The increased water height in the inland portions of the Delaware, Susquehnana, and Potomac rivers from Hurricane Irene and Tropical Storm Lee is shown in Figs. 12–14. The peak water heights at those locations were observed 1–2 days following the peak in storm rainfall. This water subsequently flowed down river into the tidal zone, increasing the water heights above astronomical tides there. This impact was more pronounced for Tropical Storm Lee, and it was most pronounced during low tides, i.e., the low tides were not as low as typical. This is likely the case because the discharge from tributary streams and groundwater would be greater during low tide, whereas the high tides can present a “backwater effect” barrier to this drainage (Cai et al. 2014; Leonardi et al. 2015). At Newbold and Washington, DC, some tide cycles had their low tides at levels typical of the high tides! An increase in water level above astronomical levels after the passage of Lee also occurred in the Delaware Bay and lower Chesapeake Bay, though it was not as pronounced as it was up the rivers (Figs. 12a-c, 13a-b).
3.5 Spatial extent of coastal flooding
The coupled model UWIN-CM have reproduced the rain, wind, waves, storm surge/tides and sea surface height during Irene and Lee compared with observations as shown in Fig. 12–14. Although induration is not represented in UWIN-CM explicitly, it is possible to assess whether we can derive useful estimates for the extent of coastal flooding using model simulated SSH. As discussed in Sec. 2.3.2, the maximum spatial extent of coastal flooding hazard can be estimated using the maximum water heights, MHHW data, and a high-resolution (1 m) DEM. Here we compare the estimated coastal flooding using the observed maximum water height to that of UWIN-CM simulated. The regions with the most widespread hazard are the eastern side of lower Chesapeake Bay, Delaware Bay, low lying areas along the Delaware River, New Jersey barrier islands and estuaries, and the New York harbor area (Fig. 15). Elsewhere, the hazard is more localized in vulnerable estuaries and along low-lying streams and rivers. The estimated hazard using the UWIN-CM SSH (Fig. 15b) corresponds remarkably well with the hazard derived from using observed water heights (Fig. 15a). These results used a threshold of 0.3 m above MHHW. Sensitivity tests showed that the potential flood areas in Fig. 15 were similar when using a threshold of 0.6 m above MHHW (not shown).