**1. Introduction**

Compound hazards are those events that occur simultaneously or successively whose combination and interaction with underlying conditions amplify the hazardous impacts from individual events [1], creating storm surge and thus seawater inundation. The storm's heavy rainfall, on the other hand, causes surface runoff, sub-surface flow and river flooding. These two flooding processes can have dependence [2] and complex interactions. A higher downstream coastal water level changes the river's downstream boundary conditions, and thus affects upstream river flow dynamics and inland

freshwater flooding [3–7]. Simultaneously, the flow of river flow into the ocean can a ffect coastal sea level changes [8], which, in turn, can act as a feedback e ffect to further impact the river flooding. Jay et al. [9] and Guo et al. [10] analyzed and documented the relative importance of tides and river flows at di fferent locations along the river channel to coastline at several floodplain wetlands.

An example of a compound hazard occurred in 1999. From late August to early September in 1999, Hurricanes Dennis and Floyd passed along and across the eastern coastal region of North Carolina (NC), and together deposited about 1000 mm of precipitation. The pair created extreme coastal, inland and upland flooding. The combined e ffects of the hurricanes resulted in massive property damage and led to 74 (56 in NC) human fatalities, due principally to the ensuing flooding. The net cost of the damage ascribed to the flood event was in excess of \$6.5B in 1999 and 10.166B in 2019 dollars. The flood event extended from the coast to New Bern NC and Washington NC, well inland and upland. At the time, the flooding inland and upland were not associated with the downstream blocking of the Neuse and Tar-Pamlico Rivers, but rather directly with Floyd's 600 mm of rainfall.

Hildebrand [11] found that from 1887 up to 1999, NC had experienced 83 named tropical storms and 31 hurricanes, but none had resulted in the massive flooding associated with these two 1999 events. However, Pietrafesa et al. [12] further documented the revelation that flooding, instead of winds, was responsible for nominally 65% of hurricane-related property damage in NC. More recently, heavy precipitation events such as hurricanes Joaquin (635 mm rainfall) and Matthew (457 mm rainfall) in South Carolina (SC) in 2015 and 2016, respectively, hurricane Harvey in 2017 in Texas (1828 mm rainfall), and hurricane Florence in 2018 (914 mm rainfall) in NC and SC, are examples of these kinds of heavy precipitation events. Hurricane Matthew resulted in \$10.3B in damage in 2016 (\$10.92 in 2019) dollars in SC alone. Rivers inland in SC crested to unprecedented levels of 5–6 m over mean water levels (NCEI), such as the Waccamaw River at Freeland and the Congaree River in Columbia, well up to the foothills of the Appalachian Mountains. Following Florence (2018), the Waccamaw River crested at 0.9 m above the level reached during the passage of Matthew (2016). There are wide ranges of spatial and temporal physics scales and thus in reported impacts (p.c. from M. McClam of the South Carolina State Guard), from several kilometers to several tens of kilometers downstream, and hours to days downstream to tens to many hundreds of kilometers upstream and days to weeks and back to tens of minutes upstream. To reduce the risks from these kinds of seemingly hidden and then often explosive flooding events in the future, we need to understand the fluid mechanics of these events that are temporally extensive and spatially massive, often transitioning to short period, laterally explosive from inland to upland and from the coasts to the mountains.

This study discusses the 1999 Dennis-Floyd event and the nonlinear fluid physics that ensued, and presents this case as a precursor to future events in-kind. First proposed by Pietrafesa and Dickey at an Eastern Carolina University (ECU) Hurricane Flood Workshop, in the study reported on below, it is found that in 1999, while Hurricane Floyd was attributed solely for the inland flood damage [13], Hurricane Dennis actually set the stage for the massive inland and upland flooding by changing the downstream boundary conditions. This study documents the events that preceded, were present during, and followed the passages of Dennis and Floyd and o ffers the possibility of an improved model prediction scheme for inland and upland flooding in coastal states. Moreover, the need to properly initialize the water levels in prognostic numerical models of incoming heavy precipitation events is suggested as both proof of concept and as a warning for the future. The Dennis-Floyd scenario is described in the section to follow, because of the comprehensive data set that is available to study that combined event. A numerical modeling scheme is envisioned and should be developed and employed in the future, if forecasts of coastal, inland and upland flooding are to improve from present-day mathematical architectures.

#### **2. Data and Study Area**

The study area, which encompasses eastern NC, and the points of in situ observations, are presented in Figure 1. Data used in this study (cf. Figure 1) include time series of atmospheric winds, precipitation, water levels and water currents. Atmospheric data is from the National Weather Service (NWS) first order stations. River discharge and water level data for Little Washington NC are from the U.S. Geological Survey (USGS). Open ocean coastal sea level and sound-side water level is from the National Ocean Service (NOS). Wind data time series are from the NC Coastal-Marine Automated Network (C-MAN station) located along the coast downstream from Ocracoke Inlet where the + sign is shown, and the Kinston, NC Airport and precipitation data are from the National Weather service (NWS). Sea surface temperature (SST) data are from NOAA's polar orbiting satellite. Sea surface and cloud color data are from the NASA SeaWifs and Infrared Imager satellites. All data, including the hurricane tracks, are available from the National Center for Environmental Information (NCEI) at: https://www.ncei.noaa.gov/.

**Figure 1.** Tracks of 1999 hurricanes Dennis and Floyd. The insert is the eastern NC study area. Triangles represent NWS and USGS data collection sites, and stars represent NOS data collection sites. Data are provided by the NCEI: https://www.ncei.noaa.gov/.
