Antecedent Conditions and Damage Caused by 2015 Spring Flooding on the Sagavanirktok River, Alaska

The total volumetric runoff during the open water season for the Kuparuk River basin at Deadhorse was calculated from the period of record, which ranges from 1971 to present (USGS Station 15896000). The Kuparuk basin is an adjacent watershed west of the Sagavanirktok River (Fig. 1).

The Kuparuk River originates in the foothills of the Brooks Range and flows north through the coastal plain to the Arctic Ocean. It is a medium-gradient basin with an area of approximately $\mathrm{8,100}{\mathrm{km}}^2$. Nearly 62% of the basin area is within the foothills region, and 38% is within the coastal plain (Kane et al. 2014). The basin elevation ranges from 1,464 m at the headwaters (Kane et al. 2004) to sea level at the outlet, with an average basin elevation of 245 m (McNamara et al. 1998). The stream length is 330 km (D. L. Kane, personal communication, 2016) and the basin length is approximately 250 km. Selected years representing the lowest and highest cumulative runoff are shown in Fig. 3. Data for 2014, when approximately 276 mm of water left the watershed, represents the highest volumetric runoff for the entire period of record, which has an average value of 150 mm. This high cumulative runoff for 2014 was attributed to a deep end-of-winter snowpack and a wet summer (Toniolo et al. 2015). Cumulative runoff during 2013 was the second highest year on record with approximately 250 mm of water.

Due to high uncertainties in the data reported by USGS Station 15908000 during breakup, a similar analysis to the one carried out for the Kuparuk River was not performed. These uncertainties arise from the reduced number and specific time of measurements. A different approach was taken herein when reviewing the runoff for USGS Station 15908000. The summer monthly mean flows were examined, and it was found that 2014 was the third highest flow on record ($n=33$) for mean monthly discharge in the months of July and August. A similar analysis was completed for other nearby rivers such as the Colville ($n=13$) and the Kuparuk ($n=45$) (Table 1). Thus, one could conclude that there was a water surplus (i.e., water storage system was full) in the region’s watersheds at the beginning of winter 2014–2015.

Table 2 shows the monthly mean discharge for the USGS gauging station. Reported 2014–2015 winter monthly values (i.e., October–May) were above the mean of monthly discharge for the entire record, with increasing percentages as the winter progressed (last rows in Table 2). In addition, Table 2 indicates that flows during winter 2013–2014 were also high. Discharge corresponding to June 2015 was below the mean discharge for the entire record, indicating that breakup consisted of one major, single pulse. In other words, snowmelt happened rapidly in the entire watershed. This argument can be supported by the air-temperature plots shown in Fig. 4. The figure indicates that air temperatures throughout the watershed, from low to high elevation, were above freezing most of the time beginning on May 8, 2015, and continuously above 0°C from May 15 to May 30, 2015, in the foothills and mountain areas. The available energy melted the majority of the snowpack in a short time frame, producing a swift breakup.

The plot in Fig. 5 represents the historical monthly mean (period of record: 1987 to 2015) and the 2014–2015 winter hourly air temperatures at the UAF Franklin Bluffs meteorological station. The region experienced a warm spell during mid-February to early March 2015. While the prediction of aufeis (in terms of location and temporal/spatial development) is still obscure to the scientific community (Kane et al. 2013; Yoshikawa et al. 2007), Kane (1981) associated warm temperatures with aufeis formation and growth. The unseasonal temperatures recorded in 2015 coincided with the initial overflow of the Sagavanirktok along the Dalton Highway. Hence, it could be argued that the warm temperatures exacerbated aufeis formation, which eventually reached the Dalton Highway, causing the initial winter flooding of the highway (Fig. 6).

In summary, the available hydrological data clearly indicate that watershed storage was full heading into 2014–2015 winter. When winter overflow began on the Sagavanirktok, plenty of water was available in the system to develop a massive aufeis field from the river to the Dalton Highway. The areal extension of aufeis in both the vertical and horizontal directions was significant. Near the Dalton Highway at MP 395, the river channel and floodplain are constrained to the east by Franklin Bluffs, which caused the overflow to spread more toward the highway and the oil pipeline to the west. In many cases, the ice elevations were above the road’s centerline elevation (Fig. 7).

Historically, with almost 40 years of monitoring, the reach of the Sagavanirktok River that flooded has not been an area prone to significant aufeis accumulation. On February 19, 2015, during a pipeline surveillance flight, aufeis was detected at PLMP 23. From the initial discovery, the aufeis accumulated rapidly. Follow-up ground surveillance on February 27 revealed some spur dikes with only 15 cm of freeboard. The top of dikes are typically 1.8–3.7 m above ground level. By March 14, the aufeis had completely overtopped and encased the first two upstream dikes with 1 m of ice.

The principal methods used to contain and deflect additional aufeis accumulation involve building trenches through ice and using the excavated ice to construct berms. This approach has been effectively implemented for the TAPS in the past at other floodplains. The ice trenches, which are directed down-valley, intercept overflows, providing a drainage path for water.

As the aufeis infills the floodplain, there is a risk that spring melt will be forced outside the active floodplain, potentially altering channel flow patterns. This scenario may ultimately lead to channel avulsions, which can pose an integrity risk to infrastructure. The ice berms, therefore, serve an additional function of containing and guiding the spring flow to the existing active floodplain, reducing the potential for channel migration.

High flows began on May 17, 2015, following a period of record-high ambient temperatures, as previously described. Historically, at the Sagavanirktok River downstream of the confluence with the Ivishak River, breakup occurs in several cycles of high flows followed by a lull where flows subside. The duration of breakup can last 4–6 weeks, with three or more cresting and falling flood waves in between. Floodwaters on May 18 rose rapidly, and the resulting backwater extended about 8 km beyond the farthest upstream limit of the aufeis field. The ice berms and trenches built to contain the flow were outflanked near PLMP 23.6, and the Dalton Highway overtopped (Fig. 8). Floodwaters reached tundra on the west side of the highway and flowed 40 km downstream towards the oil fields of Prudhoe Bay (Fig. 9).

Along the TAPS, which is located west of the Dalton Highway from MP 395 to MP 413, the floodwaters that overtopped stretches of the roadway exposed a 25.4-cm-diameter gas line at five locations and the crude oil pipeline at three locations. However, the pipes incurred no mechanical damage, so the integrity of the TAPS was not an issue. Supporting infrastructure was damaged: maintenance roads, components of the pipe-corrosion protection system, drainage structures, and ground insulation used to protect the permafrost. Costs for repairing damage associated with TAPS were significant.

On the active floodplain of the Sagavanirktok River, where the oil pipeline crosses the highway (PLMP 18.8) and is buried adjacent to the river for several miles, damage to the pipeline was relatively minor. None of the pipe was exposed by erosion. The ice berms and trenches shielded the area from main channel flows.

Prior to spring breakup—on March 13, 2015—water began flowing over the Dalton Highway from aufeis accumulations in approximately eight locations. Water depths reached 0.3 m on top of the highway, and covered stretches 60 m long. Crews of the AKDOT&PF began working to construct snow berms and dig trenches in the ice to divert water from the road. During late March and early April, water continued to flow over the top of the river ice and breached snow berms, forcing traffic to use only one lane and causing intermittent road closures. Crews worked around the clock to keep the Dalton Highway open to traffic. Road conditions deteriorated with the onset of warmer temperatures in early May. Impacted areas along the Dalton Highway continued to expand, and water depth and ice thickness continued to increase. Breakup flows began in mid-May, and by May 18, AKDOT&PF was forced to close the Dalton Highway indefinitely because water was crossing the road (water was up to 0.6 m deep on top of the road in several sections), causing severe road damage and washouts (Figs. 8–10).

Additional workforce resources were engaged by AKDOT&PF, joining the department’s personnel to protect the highway. Up to 30 pieces of equipment including graders, bulldozers, excavators, and side dumps were onsite constructing snow berms, digging trenches, and installing temporary culverts to help facilitate floodwater movement back to the river’s main channel. Water levels in the flood zone continued to rise through spring breakup. On May 22, 2015, Alaska’s Governor declared the Dalton Highway a state disaster. Extreme flooding occurred from approximately MP 395 to MP 413. Floodwater crossed the Dalton Highway to the west and encroached on an airport about 1 km outside of Deadhorse. Flooding had minimal impact on airport operations. Work crews with AKDOT&PF cut breaches across sections of the highway as an outlet to carry water back to the east towards the main channel of the Sagavanirktok (Fig. 10). The Dalton Highway reopened on June 5, 2015, after an 18-day closure. Motorists drove in typical construction conditions, with the road rough and narrow. The extent of the road damage was not entirely known until water levels dropped. Table 3 shows the location and main characteristics of each breach along the Dalton Highway. The total volume of borrow material needed to bring the road back to preflood conditions surpassed $\mathrm{18,000}{\mathrm{m}}^3$.