The characteristics of ‘abrupt wave front’ floods on Pennine catchments, northern England, and their transmission downstream

Extremely rapid rates of rise in level and discharge in a subset of flash floods (‘abrupt wave front floods’ (AWF)) are separate hazards from peak level. Such flood events were investigated for Pennine catchments in northern England (Archer & Fowler 2021). Historical data for 122 events are extracted from a chronology of flash floods for Britain freely available on the JBA Trust website (https://www.jbatrust.org/how-we-help/publications-resources/rivers-and-coasts/uk-chronology-of-flash-floods-1/), and hereinafter referred to as ‘The Chronology’. Until recently, such events have been attributed to ephemeral upstream landslides or bridge failure but are now shown to occur predominantly as the result of extreme rainfall on steep upland catchments without upstream blockage but can be transmitted downstream with a steepening front for tens of kilometres (Archer & Fowler 2021). Historically, observers have frequently described such events in Britain as ‘walls of water’ including the flood of 17 July 1983, reported below. Similarly, Viggiani (2020) lists 20 ‘instant floods’ from various parts of the world which caused death by drowning. Collischonn & Kobiyama (2019) also note the relatively frequent occurrence of ‘cabeça d’água’ in southern Brazil as a type of flood in which it is possible to clearly observe the arrival of the flood wave as a visible discontinuity of flow and water level. The focus of this paper is on two such events on the Rivers Wear and Tees in 1983.

The hazard and frequency of events with very rapid rates of rise in level and discharge is not adequately recognised in Britain. AWF events are a threat to river users even when the peak flow is not severe or extreme. The principal objective of the paper is therefore to raise awareness and to encourage development of procedures with agencies responsible for monitoring, forecasting and warning to river users.

The description and analysis is based on contemporaneous observations of the effects of the floods and interviews of affected residents and on archived rainfall and level and flow records at 15-min intervals for the period 1982–2014 held at the Environment Agency.

The flood of 17 July is described by Carling (1986a) and Archer (1992) but is considered here in the context of the concept of AWF floods. July 1983 had at the time the highest recorded July temperature in central England and is now only exceeded by 2006 (Met Office Hadley Centre 2022). In the northeast temperatures reached 29° on 14th and 15th July. Monthly rainfall totals were less than 25% of the average for many stations in England but thunderstorms were widespread during the month including the storms of 17 July in the upper Wear and Tees. The single recording gauge in the storm area at Burnhope Reservoir registered the start of the storm at 15.24 but failed after a few minutes probably as the result of a lightning strike. The adjacent daily gauge registered a total of 87 mm. A rainfall of 104.8 mm was measured at Ireshope Plains between 15.30 and 18.00 but the observer noted that the most intense rainfall fell in 1.25 h between 15.45 and 17.00. The observer also noted that the storm was more intense over the moors to the south. The most intense rainfall seems to have occurred on Noon Hill on the margins of which five peat slides occurred, three towards the Ireshope Burn, and one each to the West Grain (Wear catchment) and the Langdon Beck (Tees catchment) (Carling 1986b; Archer 1992). It was from these three tributaries that the observed AWF on the main river gauging stations at Stanhope (Wear) and Middleton (Tees) originated.

The Ireshope Burn rises on the north side of Noon Hill and the occurrence of three peat slides suggests the storm intensity was at least of the same order as on the West Grain. Information was gained from interviews with residents near the mouth of the stream which suggested a steep wave front which dislodged two riverside caravans and carried one (unoccupied) into the River Wear and another (occupied by a family) carried off but held against a wall until the wave subsided. The rapid rise in level precluded the possibility of escape. The catchment area of the Ireshope Burn to the Wear confluence is 7.9 km².

The raingauges on the Burnhope Burn indicate that significant flow could also have arisen on that catchment but the reservoir, drawn down to summer levels, captured most of that flow and there was no overflow.

At the Stanhope gauging station (catchment area 172 km²), 10.5 km downstream from the East/West Grain confluence the discharge rose from 0.76 to 71 m³/s between 16.15 and 16.30 with an equivalent rise in level of 1.53 m. The peak flow of 97.8 m³/s was 1 h later at 17.15. A police observer along the river reported a ‘wall of water’ to the flood warning control room at Northumbrian Water.

The estimated flow from the 1.86 km² catchment of the West Grain of 16–22 m³/s represents 17–23% of the peak flow at the Stanhope gauging station (172 km²). With similar intensity of runoff, the peak flow at Stanhope of 94 m³/s could have been created from an area of 8–11 km². Since the combined area of the Ireshope Burn and West/East Grain catchments is 13.2 km², this small contributing area seems credible. The total volume of storm flow at Stanhope was approximately 1,410,000 m³. Assuming contributing areas of 8 or 13.2 km², the effective rainfall for the storm was 176 or 107 mm.

The next gauging station downstream at Sunderland Bridge (catchment area 658 km²) was out of operation during the event, and at Chester le Street (catchment area 1,008 km²), the rising limb was much less steep with the 15-min rise less than the median annual maximum.

At the Middleton gauging station (catchment area 242.1 km²), the flood wave arrived at 17.15 when the level rose from 5.2 to 60.0 m³/s (from 0.53 to 1.45 m). However, the contributing waves seem not to have completely merged as the first peak was followed by a brief trough, then a rise to a second main peak of 85.7 m³/s at 18.15. The combined peak flows from the Langdon Beck and the Harwood Beck are sufficient to account for the peak at Middleton without any further contribution.

A chart record from Barnard Castle gauging station (catchment area 509.2 km²), 16.4 km downstream shows a sudden rise from 0.52 to 1.17 m (6.3–77.6 m³/s) at 19.30. However, the high level was maintained for 2 h, indicating that the contributing waves have still not fully merged.

At Broken Scar gauging station (catchment area 818.4 km²), a further 28.4 km downstream, the hydrograph shows considerable attenuation with the initial increase in level from 0.52 m at 23.45 to 1.22 m at 00.30 on 18th (3.5–60.0 m³/s). However, the maximum 15-min rise in discharge was 41.6 m³/s, an increase which could still endanger river users.

The June 7/8 1983 flood entirely escaped notice at the time both by the press and by Northumbrian Water, then responsible for flood risk management and warning. The measured rainfall for the event, although heavy, was insufficient to account for the extreme flow (Ireshopeburn – 30.7 mm; Burnhope – 27.5 mm and Greenhills – 33.1 mm). The most significant feature of the storm was the fact that it caused an AWF on the Wear and Tees at Stanhope and Middleton at exactly the same time of 01.45. This observation provides conclusive evidence that AWF floods are not necessarily caused by landslide blockage and subsequent release as this would require simultaneous blockages on both sides of the divide (Archer 2021). On both rivers, the recession was rapid and by the following morning, the level had fallen to a normal summer level. The AWF observed on the Langdon Beck indicates that intense storm rainfall occurred on its headwaters at Noon Hill, the same location of intense rainfall as in the 17 July flood. The June storm may therefore have influenced the peat and substrate making it more vulnerable to the peat slides which occurred in the July storm.

The storm seems to have overlapped with the east side of the storm of 17 July causing an AWF on the Langdon Beck with an instantaneous rise in level from 0.12 to 1.38 m at 00.00 on 8 June. However, the Harwood Beck was unaffected. At the Middleton, the discharge rose from 3.1 to 123 m³/s at 01.45 and fell in the next 15-min interval to 108 m³/s. Although the Langdon Beck rise in level was rapid, the associated discharge was insufficient to account for the flow at Middleton, and further inflow from downstream tributaries is assumed. The Barnard Castle record appears to be influenced by partial blockage of the stilling well with the rise from 7.9 to 88.7 m³/s spread over 45 min and the recession broken. The Tees was not much influenced by tributary inflows, but at Broken Scar, there was a gradual increase over a period of 6 h before a 1-h rise from 7.7 to 69 m³/s with a maximum 15-min increase of 44.3 m³/s. The equivalent rise in level was only 0.36 m given the insensitivity of the multiple-crested weir.

Although the focus here is on the floods of 1983, there were four other AWF events at Middleton in the record period from 1982 to 2014 during which the water level rose by more than 1 m in 15 min (Table 1). Three of these occurred at the same time as AWF events on the South Tyne with which the Tees shares headwater sources on the Cross Fell massif. On 20 July 2002, a recording raingauge at Alston on the South Tyne recorded 26.2 mm in the first 15 min of the storm. At Middleton on the Tees, the water rose 1.12 m (76.3 m³/s) in 15 min, while on the Tyne, the AWF persisted to the lowest gauging station at Bywell with a catchment area of over 2,000 km² (Archer & Fowler 2018). Similarly, AWF events were experienced on the Tees and South Tyne on 9 August 2004 and 20 July 2007. On both occasions, thunderstorms were widespread and properties were flooded from surface water following intense rainfall. In addition, on 20 July 2007, the overflow of tributaries caused flooding in Alston, Haltwhistle and Bardon Mill and overwhelmed and nearly drowned a fisherman standing on a flood bank at the river's edge on the lower South Tyne (Archer & Fowler 2022). Flash flood hydrographs for the South Tyne are shown in The Chronology (Northeast region). On 4 August 1994, storms were more dispersed but a remarkable AWF also occurred on the lowland River Wansbeck (Archer 1994; Archer et al. 2016).

Table 1 shows the contrast in 15-min rise in level and discharge between AWF and normal summer and winter annual maximum floods. While the median rise in AWF floods was 1.12 m (81.4 m³/s), for summer AMAX floods, it was 0.22 m (30 m³/s) and for winter AMAX floods, it was 0.16 m (33.0 m³/s). In each case, normal flood rises from heavy persistent rainfall were preceded and followed by similar but smaller 15-min rises.

Annual maximum flood peaks were, with one exception, greater than the peaks of the AWF floods, indicating the greater risk to property from such normal floods. However, the concentration on flood peaks as the only measure of severity misses the quite separate risk posed by the rate of change.

Such unusual catchment response has been quantified by a metric of hydrograph skewness as the volume before peak (VBP) of the hydrograph (Collischonn et al. 2017). The VBP is calculated as the proportion of total hydrograph volume that occurs before the occurrence of the peak. The VBP ranges from 0 to 1, with a low value when the hydrograph is positively skewed and a high value when the hydrograph is negatively skewed. An alternative simpler measure was used to demonstrate the contrast in hydrograph skewness between AWF and normal floods on the same catchments. The maximum 15-min rise in discharge (m³/s) is compared as a percentage of peak discharge (m³/s).

There is a clear difference between these ratios for AWF events versus normal floods at the same gauging stations. AWF events have an average proportion of 72.4%, with a maximum and minimum of 97.5 and 31%, respectively, while normal annual maximum events have an average of 12.6%, with a maximum of 36.2% and a minimum of 5.2%.

Although not all rapid rise events on the basis of level records with a minimum 15-min time step can be conclusively defined as a ‘wall’ or a wave front increasing within a few seconds, observers’ descriptions of the event of July 1983 and numerous historical events (Archer & Fowler 2021) provide strong collaborative evidence. The flow records are not in contrast with the occurrence of such phenomena. Even at a 15-min time step, AWF events are clearly different in character from normal flood events on the same catchment (Figures 6 and 7).

The Wear and Tees are not ‘rapid response catchments’ (Francis 2010) but these rare events are ‘rapid response events’. But what are the causes of the difference between normal floods and AWF floods on the same catchment? The character of the AWF response in the vertical or near-vertical wave front and rapid increase in both level and discharge points to the occurrence of a kinematic shock wave (Archer 2022).

Lighthill & Whitham (1955) recognised the intrinsic non-linear property of kinematic flood waves and the associated shock potential, creating a visible discontinuity in the water level and discharge. However, until recently, the real-world occurrence of such AWF or shock waves has been doubted in spite of the theoretical basis for their existence. Such scepticism has been abetted until recently by the very rare observation of such events and it has been concluded by distinguished hydraulic engineers that shocks are simply a result of the approximations made in the development of the theory (Henderson 1966). However, nearly 300 historic and recent events have been described in the chronology of flash floods (Archer & Fowler 2021) and their occurrence can hardly be described as very rare. In this paper, we have given a detailed account of two events from which to account for the difference from normal floods.

Ponce (1991) notes that the shock is a direct consequence of the non-linear steepening tendency of the wave front, which is abetted when the following conditions occur:

First, the wave is kinematic as opposed to diffusive (or dynamic). Diffusion is a mechanism acting to oppose the non-linear steepening tendency. The more diffusive a wave is, the less kinematic, and therefore, the less the steepening tendency. The wave form at Stanhope (Wear) and Middleton (Tees) with the near-instant rise in level provides evidence that the wave is kinematic. Further downstream diffusion sets in and the wave front becomes less steep but diffusive effects seem to be greater on the river Wear than on the River Tees.

A second suggested requirement is that there is a low base-to-peak flow ratio. The steepening tendency is promoted when the flow is subject to large relative changes, with base flow being only a small fraction of peak flow. This is certainly the case both for the small West Grain catchment and for the larger upper Wear and Tees catchments. For the West Grain, Figure 2(a) shows the contrast between the normal very small summer flow observed a week after the event and the effects of the flood wave. At Stanhope, the flow increased in the 15-min interval from 2.19 to 74.4 m³/s on 7 June and from 0.76 to 71 m³/s on 17 July. At Middleton, the 15-min increase in flow was from 3.06 to 123 m³/s on 7 June and from 5.19 to 85.7 m³/s on 17 July. Figure 6 also shows the very low initial flow for the AWF events compared with normal winter and summer flood events.

Ponce (1991) also suggests that since wave steepening is gradual a sufficiently long channel is necessary to give the shock a chance to develop. Distance from the West Grain confluence to the Stanhope gauging station on the Wear is 10.8 km with a fall of 48 m and on the Tees from the Langdon Beck confluence to the Middleton gauging station is 23.5 km with a fall in level of 162 m. However, Ponce suggests strong steepening tendencies may require a shorter reach, and in the case of the steep West Grain, there is strong observational evidence that an AWF was already in existence within 3 km of the source of the storm – notably the swash mark of a wave on the grass (Figure 2(a)). Several other Pennine catchments where AWF events were observed on areas less than 5 km² are included in the Chronology. It is postulated that the larger discharge from the Ireshope Burn steepened and merged with the wave from the West Grain in the main Wear channel.

A further condition for shock development is that the channel is hydraulically wide, that is, one in which the wetted perimeter is nearly constant where the wetted perimeter increases very little in comparison with the increase in flow depth and area. This condition may be satisfied for in-bank flows as was the case for the Wear and Tees for these two events. The steepening tendency is counteracted and shock development is arrested in shallow-overbank flow situations but may still occur in extreme events such as the River Rye at Helmsley (Wass et al. 2008) or the Boscastle flood (Fenn et al. 2005).

Ponce also notes that the steepening tendency is promoted at high Froude numbers, at or above critical, to the point where the shock may develop. In hydraulically wide channels, high-Froude-number flows lack sufficient diffusion to effectively counteract the steepening tendency (Ponce & Simons 1977). Such high-Froude-number flows are rare in natural streams and rivers and are unlikely to occur on the main Wear or on the Tees with the exception of the steep rapid and fall section at High and Low Force. However, the steep tributaries on either side of the divide, where the AWF are generated, are likely to experience such high Froude numbers.

It is concluded that the events on the Wear and Tees in 1983 meet most of the conditions for kinematic shock specified by Ponce (1991). However, the most important feature of these events was the intense localised rainfall that generated rapid rates of rise on small upland tributaries, which were then transmitted downstream on the main rivers with a steepening front for more than 10 km. The account given here is descriptive and based on general observations but hydrological and hydraulic analysis with modelling capable of ‘shock capturing’ is required to provide a basis for forecasting and warning of the occurrence of such events including their generation and downstream transformation.

• 1.

  Five AWF events with 15-min rise in level greater than 1 m occurred on the Tees at Middleton and three on the Wear at Stanhope in a 30-year period. The hazard and the risk to life need to be taken more seriously for flood forecasting and warning procedures.

• 2.

  AWF events floods are categorically different from normal floods on the same catchment. The Rivers Wear and Tees are not normally ‘rapid response catchments’ but they experience rapid response events on rare occasions.

• 3.

  AWF floods are generated from intense rainfall on upland tributaries and very rapid rise in level already occurring on the West Grain catchment of 1.5 km².

• 4.

  Tributary flood waves merge in the main channel with the wave front steepening and persisting downstream for more than 10 km on the Wear and more than 20 km on the Tees.

• 5.

  Diffusion and attenuation occur further downstream but rates of rise 45 km downstream on the River Tees at Broken Scar are still sufficient to be a serious hazard to river users.

• 6.

  AWF floods differ from normal floods on the same catchment in their much more rapid short-period rate of rise and in terms of the comparative maximum 15-min rise in discharge as a percentage of peak discharge.

• 7.

  AWF floods satisfy conditions for characterising kinematic shock waves and in spite of previous reservations they should be recognised as such.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.