sign in sign up
<<
Home , Boscastle, Canworthy Water, Cloudburst, Hendraburnick, Lesnewth, Lynmouth

– August 2005, Vol. 60, No. 8 219 copters. copters. Local lifeboats were scrambled when it was feared that people had trapped been in cars and washed out to sea. It is clear that without the prompt response of the coastguard, local emergency and services the military, many been lives lost and it is indeed fortunate that no- might have one was killed in this major rainstorm in a popular tourist area at the peak of the holi- day . More details and photographs of the are damage given in Doe (2004) and in eye- and witness accounts given in around a Radio 4 docu- later months four exactly broadcast mentary (Kirby 2004). on 16 August 2004. UK Met Office analysis – Crown Copyright. UTC Cloudburst Cloudburst upon Hendraburnick Hendraburnick Down: noptic situation at 1200 Fig. 1 Sy 1 Fig. flooded and several destroyed (at least one of which was over 70–80 400 cars were years swept through the old) village while by the force of the , damaging bridges and property during their passage. Many of to out washed subsequently were cars these sea. All the boats in destroyed. Several the bridges harbour were were severely damaged or destroyed and in places road The away. completely surfaces scoured were immediate danger to local tourists residents alike and was sufficiently serious about that 130 adults and children had rescued (winched to from roofs be and trees) by seven coastguard and heli- from from , UTC The The Boscastle storm of 16 August 2004 is shown in Fig. 1; the upper-air UTC In 70 over Boscastle propertiesitself, were – Cornwall about 70 km south-west/upwind in Golding (this issue, of Boscastle - is shown northand Cornwall Inwest 231). page 2, Fig. convergence effects between winds and local topography resulted in onshore the repeated development of intense storm cells over high ground just inland from the coast of between and 2 to (see for locations Fig. referred in the text). The repeated generation same the much in cells these of propagation and afternoon the during hours several over area resulted in enormous quantities of falling in a narrow strip parallel to and a few kilometres inland from the coastline. area The of heaviest rainfall headwaters lay close of to several the which coastal rose streams, very rapidly during noon resulting in the destruction the of bridges after- and property particularly in the coastal set- tlements of Boscastle Haven. and Less Crackington severe flood reported across damage north-west was Cornwall into north and . ascent ascent made at 1200 In an unstable cyclonic thundery situation, showers heavy, developed in south-west parts by late of morning on 16 August 2004. The synoptic 1200 situation at The The Boscastle storm On 16 August 2004 the coastal Boscastle in north village Cornwall was devastated of torrential of series a of result a as flooding by depositing over 200 mm of rainfall within four hours into rocky catchment. This the article describes small the meteorological circumstances of this excep- tional storm, in particular the rainfall obser- vations made in the immediately locality after the event, during and compares and it to other intense rainstorms observed in the British Isles. Stephen Burt Stratfield Mortimer, Berkshire West 3 km south-west of ; the fall in this area probably exceeded 250 mm and may have surpassed 300 mm over perhaps 1 km2. Just as remarkable as the very large rainfall totals are the exceptional gradients on the south and east flanks of the storm system. At Lower Moor on the outskirts of – and only 5.5 km south-south-west of Hendraburnick Down – just 2.0 mm fell during the day (a rainfall gradient of almost 50 mm/km), while just 6 km east of Otterham only 15.0 mm fell at Canworthy The The Boscastle storm of 16 August 2004 Water (a gradient of over 30 mm/km). There are no raingauges in Boscastle itself, but radar rainfall evidence would suggest that perhaps only 30–50 mm fell there. Figure 5 shows 15 minute totals from eight recording raingauges across north Cornwall on 16 August 2004. Totals from these and other gauges shown in Fig. 2 are given in Table 1. Based upon both radar and raingauge evidence, the main sequence of events appears to have been as follows: Fig. 2 Locations of places referred to in the text; inset shows location of detailed map. Standard daily-read ● raingauges are shown by green circles, recording gauges by red circles. HD marks the location of The first shower cell originated to the Hendraburnick Down. Base map courtesy of Environment Agency. south-west of Padstow about 1115 UTC,

Weather – August 2005, Vol. 60, 8 – No. Vol. August Weather 2005, giving a short burst of intense rainfall at Roserrow shortly before 1130 UTC: 1.4 Rainfall records low by a factor of about 2 in the central area mm fell in 2 minutes (42 mm/hr). As the of the storm*, the radar rainfall evidence cell developed and moved north-east on Rainfall observations are available for this suggests that the highest totals were proba- date from 21 locations in the area shown by the mid-level winds (see the Camborne bly recorded over or close to the moorland ascent in Golding, this issue, Fig. 2, page Fig. 2. Eleven of these are recording gauges, of Hendraburnick Down (rising to a little the majority being telemetered tipping 231), it produced 5.5 mm in the 30 min- over 300 m above mean sea level) about utes commencing 1200 UTC at Slaughter- bucket raingauges (TBR) operated by the Environment Agency, providing 0.2 mm or bridge and 18 mm in 30 minutes 0.5 mm rainfall resolution at fixed 15 minute * The Boscastle area is on the edge of the radar commencing 1215 UTC at . ● telemetering intervals. At the time of writing coverage – Otterham is 70 km west-south-west A second major cell formed on its south none of the raingauge data presented in this of the Cobbacombe Cross radar. At this flank in the Delabole/Camelford area at paper had yet been subjected to Met Office distance the radar beam (at 0.5° elevation) is about 1215 UTC. This cell was responsible approximately 800 m above mean sea level. Quality Control procedures. for the majority of the observed rainfall At five sites, the rainfall for the rainday 16 August 2004 (i.e. the 24 hours com- 34.8 mencing at 0900 UTC on 16 August) exceeded 75 mm, the highest daily fall 46.7 58.0 being at Otterham (National Grid Reference (NGR) SX 169 916), about 7 km east of 57.2 Boscastle, where 200.4 mm was recorded. 75.5 The highest fall at a site with an autographic rainfall record was at Lesnewth (NGR SX 131 123.0 902), about 4 km west-south-west of 200.4 Otterham, where 184.9 mm was recorded in 15.0 a standard raingauge and 155.2 mm in the 184.9 adjacent logged tipping-bucket gauge. 10 Figure 3 shows the approximate distribu- 2.6 tion of rainfall for the rainfall day of 76.5 6.0 16 August 2004. This is based upon the 21 54.8 2.0 daily raingauge totals (including ‘as read’ 11.8 totals from the tipping-bucket gauges 5.0 2.0 where no checkgauge value was available, 5.9 although these are likely to be significantly 3.0 lower than the true particular- 2.4 ly in areas affected by intense rainfall), together with the composite 2 km × 2 km Cobbacombe Cross radar rainfall accumula- Fig. 3 Approximate distribution of rainfall for the rainfall day of 16 August 2004. Units are millimetres: iso- tions for the period 1100–1800 UTC shown in hyets are at 5, 20, 50, 100 and 200 mm. The approximate area covered by the radar rainfall accumulation Fig. 4. Although the radar-derived totals are map (Fig. 4) is shown by the dashed rectangle. 220 The Boscastle storm of 16 August 2004 Weather – August 2005, Vol. 60, No. 8 221 . UTC UTC onwards, in Fig. 6). A , again in the UTC UTC UTC , two-thirds of the , 88, at and mm/hr) the character of the UTC UTC , including 21.5 mm in UTC UTC from from this cell whereas at total total of 74.5 mm) fell between (see below). ) while the cells developing to Data courtesy of Environment Agency UTC UTC UTC UTC large large area of intense rainfall affected the Bude area and eastwards Devon into from north 1530–1600 the 15 minutes commencing 1300 storm system changed, from a very line intense, of narrow cells to wider, less a intense much decaying multi-cell storm system (compare the images for 1500, 1600 and 1700 the north-east produced intense rainfall minutes 15 in mm (22 Bridge Crowford at 1315 commencing A further developing storm cell intensi- fied the rainfall at 1530 Lesnewth around afternoon’s rainfall at this site). afternoon’s 1245 and 1415 At At Lesnewth the commencement of this highly organised, narrow linear vective con- storm regime was marked by a prolonged spell of (30 mm intense in 30 rainfall 1300 minutes commencing Another very developed about 1430 intense storm cell At about 1615 at : here 80% (59.5 rainfall 1200– the of mm afternoon’s of the 1800 (86 mm/h). From about 1300 a series of extremely narrow (approximately 10 convective km wide) developed cells close to Delabole, just 5 km or so inland from the Atlantic coast. A line of similar intense narrow, extended quickly north-east in an almost unbroken chain of echoes across , the Channel and into South hourly rainfall radar 6). images, Fig. Wales (see Delabole area. The storm grew in inten- sity very rapidly as it moved north-east, such that Slaughterbridge received only 8 mm in the 45 1430 minutes commencing Lesnewth, Lesnewth, only 5 km about away, 96 mm in fell 75 minutes commencing 1445 Canworthy Canworthy Water (9.5 mm in 15 minutes commencing 1330 ● ● ● . The area covered UTC

TBR details rainfall (mm) Total hour Wettest 15 min Wettest . The axes are the same for all plots – 0–35 mm (line rules every on time axis. UTC UTC and are taken directly from original records UTC osite osite 2 km × 2 km accumulations for rainfall radar the period 1100–1800 Rainfall Rainfall data from gauges recording in north-west on Cornwall 16 August 2004 All times are in Site Roserrow Golf Course SW 945 780De LankSX 132 765 0.2 mmSlaughterbridge 0.2 mm 10 secTrevalec Lesnewth, 15 minWater Canworthy NGR SX 109 857 SX 131 902 Logged Telemetered 0.5 mm 0.2 mm BridgeCrowford SX 227 916 15 min 10 sec 0.5 mm No Capacity Resolution 15 min SX 290 991 Telemetered No Source Logged 0.5 mm* Adjusted to adjacent checkgauge total where available. Telemetered 15 min No 3.0 Adjusted* No Telemetered 5.0 Yes 09-09h 1.6 12-18h 76.5 No 0.8 184.9 15.0 74.5 1.8 mm 183.2 Commencing 2.6 14.5 75.5 46.0 85.7 mm 74.5 12.5 1745 mm/hr 1804 Commencing 34.0 1300 1.4 1450 1330 2.4 21.5 32.6 6 1300 9.5 10 86 130 22.0 38 1745 1804 88 1300 1530 1330 1315 Table 1 Table SW 988 727 0.2 mm 15 min Telemetered MillWoolstone No SS 227 018 0.2 mm 15 min 2.4 Telemetered nil No 0.4 57.2 57.0 1815 39.6 0.4 2 1545 14.0 2300 56 1615 Tamarstone SS 282 056 0.5 mm 15 min Telemetered No 58.0 57.0 34.5 1600 20.0 80 1630 Yeolmbridge SX 317 873 1 mm 15 min Telemetered No 10 nil 7 1830 5 20 1845 5 mm) on value axis, 1200 to 1800 by by the plot is shown in 3. Fig. Met Courtesy Office – Copyright © Crown Fig. 5 Fig. Rainfall totals over 15 minute periods from eight recording raingauges in north-west on Cornwall 16 August 2004, period 1200 to 1800 Fig. 4 Comp 4 Fig. The The Boscastle storm of 16 August 2004 Weather – August 2005, Vol. 60, 8 – No. Vol. August Weather 2005,

Fig. 6 High-resolution (2 km × 2 km) radar rainfall data from the radar at Cobbacombe Cross, Devon at hourly intervals from 1200 to 1800 UTC on 16 August 2004. The time of the image is shown in the bottom-left-hand corner of each frame. Boscastle is located close to the white pixels close to the extreme left of the first image, very close to the edge of the radar coverage circle. Courtesy Met Office – © Crown Copyright.

eventually extending to cover the thunderstorms stretching as far as the hours the trigger point for individual cells western Bristol Channel and reaching upper Severn estuary. This system remained anchored in the Delabole area. South Wales. At Woolstone Mill, south- persisted for about 2½ hours, a little Significant flooding occurred on all the east of Bude, 39.6 mm fell in the hour more than half as long as the Boscastle major rivers in the locality, particularly the commencing 1545 UTC while at event, but as these cells did not develop Valency in Boscastle. That the flooding Tamarstone, east of Bude, 20 mm fell in repeatedly above a single area, point southwards on the Camel through 15 minutes from 1630 UTC. By this time, rainfall totals were not exceptional Camelford and Wadebridge was not worse heavy rainfall had ceased in the Boscastle although intensities probably reached can be attributed to the heaviest of the rain area and although further showers 100–150 mm/hr at times. after about 1430 UTC affecting only the top- continued during the evening and most headwaters. In fact, the heaviest of the overnight, they were much less intense in The extraordinary rainfall distribution rainfall fell upon the watershed separating nature. At the same time, another very shown in Fig. 3 thus resulted mainly from the Valency, Ottery, Inny and Camel rivers narrow band of intense storm cells the repeated generation of storm cells a and this fortuitously dispersed the developed quickly a little north-east of little to the south of Boscastle followed by a enormous volume of runoff across several Okehampton (about 50–60 km east- longer declining tail of very heavy rainfall for catchments. Had the storm been centred wards of the ‘Boscastle storms cell line’ a many kilometres north-east. Throughout more squarely within one of the smaller couple of hours earlier, and with an origin the afternoon sequence of radar rainfall catchments the effects would have been close to the Cobbacombe Cross radar truly catastrophic. Downstream flooding on 222 images it is striking how very narrow the site), generating a line of showers and band of intense rainfall was, and how for 3–4 the Ottery and Inny (and eventually the The Boscastle storm of 16 August 2004 Weather – August 2005, Vol. 60, No. 8 223 , ; . the the UTC UTC UTC , then UTC UTC , 72 mm had , the rain built UTC extrapolate . After this burst of burst this After . UTC UTC , a further 31 mm had UTC . In the 5 minutes ended UTC and was less intense than the and built very rapidly in inten- about 20.3 mm fell, an average . In the 90 minutes from , approximately 109 mm fell at UTC UTC UTC UTC UTC decreased decreased somewhat in rate (‘eased off’ seems hardly the correct here) term to a sustained to 60–70 mm/h before use a single, short and remarkable peak just after 1530 first two spells, reaching a peak 5 minute 50under just intensity 1415 at mm/h already been and recorded; quickly to an intensity of 120 mm/hr in the 5 minutes ended 1500 lasted lasted about 30 minutes 5 with a minute peak rainfall 75 mm/h; about 21 mm fell in this rate spell; of close to 1307 at seconds 10 of period a for the day’s spell added 19 mm to third This total, such that by 1440 Lesnewth. 1440 sity to almost 120 mm/h in the 5 minutes ended 1310, attaining a peak intensity of 156 2 mm/h 1306 minutes from over rainfall intensity probably exceeded 400 mm/h. By the time briefly the at 1335 rain stopped The The first, commencing about 1220 The second 1303 spell commenced at The third spell lasted 1440 from 1350 to longest the was spell final fourthand The and heaviest spell of rain Lesnewth recorded at during this mencing about 1440 event. Com- 1535 rainfall intensity of just over 240 mm/hr. Over a 20 second period 5 within this minute approached spell, and the possibly 5001534 at mm/hr exceeded rainfall very rate intense rainfall, rates 40–80 declined the tailed off, mm/h and the storm to last rainfall tip 1619 being recorded at fallen; From From Fig. 7 we can see there were four It is clear from the observed distribution main periods of heavy rainfall observed at Lesnewth: Inferences on peak depth-duration peak on Inferences storm data Table 2 shows the depth-duration data for described as corrected record, Lesnewth the peak the than higher much are These above. rainfall rates of around 50 mm/hr quoted in other accounts of the Boscastle Storm (e.g. Doe 2004). that storm rainfall varied spatially and temporally, and it hugely is therefore both unsafe to attempt to draw clusions with detailed regard to the con- depth-duration profile at any other 2 in Table site. However, an attempt is made to ● ● ● ● depth-duration data based Lesnewth record to upon Otterham, some 4 the km , comprising the main storm UTC During During periods of intense rainfall, a period. period. There was a break of a little over an hour in the rainfall after 1620, followed by further, much lighter, showers during the evening and overnight. tipping-bucket gauge significantly can compared to a under-read standard rain- gauge owing both to losses in the tipping process and to a whereby tipping’, process the inflow of of ‘continuous water from the raingauge funnel is so smooth large that the mechanical mechanism operation slows down or even ceases*. At of a the rainfall rate of 500 mm/hr, for example, it takes only 1.4 seconds to capacity tipping bucket. fill An increasing pro- a 0.2 mm portion of the incoming rainfall will be lost as the rainfall intensity and increases, above about 50–100 mm/hr the loss becomes increasingly significant. It of is there- record fore likely that between the tipping-bucket gauge more and the of the checkgauge occurred at shortfall the higher inten- sities at the peak of the storm, than at lower intensities. Applying a heuristic correction to the amount of every tip on the original record in linear proportion to the observed rainfall intensity such that the 24 hour rain- fall total from the equals tipping-bucket the gauge total rainfall can we in the checkgauge, period measured over the same construct Fig. 7 which shows intensity the rainfall (in millimetres Lesnewth over 5 minute periods from per 1215 hour) to at 1630 * Hailfall can also lead to considerable uncertainty in short-period estimates of precipitation intensity owing both to loss of hailstones (bouncing out of the collecting funnel) and to the build-up and subsequent slower release of a mass of precipitation, frozen but on the basis of eyewitness accounts of the Boscastle storm does not appear to have contributed significantly to observed precipitation amounts, if at all. – just UTC on 16 August 2004 UTC Recording Recording raingauges normally read Totals from Totals this gauge at 15 minute reso- slightly lower than standard raingauges for a variety of reasons, and it is therefore stan- dard practice wherever possible to adjust gauge the from recording tabulated records to agree with the manual from a co-located measurement standard raingauge (the where ‘checkgauge’) both records are avail- able (Met Office 1981). It should be noted, however, that not all of the tipping-bucket raingauge records considered here have adjacent checkgauges and thus this correc- tion cannot be undertaken for 1). Table (see every site lution are shown in Fig. 5. Fortunately, the records from this gauge are available from the original logger records at a much finer temporal resolution (10 seconds) and using the original tip-time data and correction applying to a the observed totals to agree with the checkgauge total, we can examine the fine structure of the rainfall event at this location close to the centre of the storm in greater This detail 7). is (Fig. the highest res- olution record yet obtained of any rainfall event in the British Isles. major over over 4 hours. The Lesnewth rainfall The record rainfall Lesnewth The tipping-bucket gauge at Lesnewth was close to the centre of the area highest and rainfall fortunately remained in working order throughout the storm. The total fall from the 0.2 mm resolution tipping-bucket gauge over the rainfall day was 155.2 mm, a shortfall of almost 19% from the this Of mm. 184.9 of ‘checkgauge’value adjacent daily 0900–0900 the of (98% mm 152.0 mm, 155.2 total) fell between 1215 and 1619 Tamar) Tamar) was mostly minor, although rapid rise the in river levels during the evening was remarked upon as a ‘wall of water’places. in Fig. 7 Fig. Rainfall intensity (in millimetres per hour) over 5 at Cornwall minute Lesnewth, periods from 1215 to 1630 Table 2 Peak rainfall depth-duration measurements and estimates for the storm of 16 August 2004. The Lesnewth observations have been taken from the tipping-bucket record, adjusted to agree with the daily checkgauge value as described in the text. The Otterham and Hendraburnick Down values have been extrapolated/estimated respectively from the adjusted Lesnewth record.

Lesnewth Otterham Hendraburnick Down Duration SX 131 902 SX 169 916 SX 1588

Daily rainfall 184.9 mm 200.4 mm 250 mm ± 20% 0900–0900 UTC Measured Measured Estimated from radar rainfall accumulations The The Boscastle storm of 16 August 2004 4 hour rainfall 181.4 mm Extrapolated 197 mm Estimated 250 mm ± 20% 1220–1619 UTC 3 hour rainfall 148.4 mm Extrapolated 160–170 mm Estimated 220 mm ± 20 mm 1300–1600 UTC 2 hour rainfall 115.4 mm Extrapolated 125–130 mm Estimated 170 mm ± 15 mm 1410–1610 UTC 1 hour rainfall 85.7 mm Extrapolated 95–100 mm Estimated 125 mm ± 10 mm 1450–1550 UTC 15 minute rainfall 32.6 mm (130 mm/hr) Extrapolated 35–40 mm Estimated 50 mm ±10% 1530–1545 UTC (140-160 mm/hr) (~ 200 mm/hr) 5 minute rainfall 20.3 mm (240 mm/hr) > 250 mm/hr > 250 mm/hr Weather – August 2005, Vol. 60, 8 – No. Vol. August Weather 2005, 1530–1535 UTC

1 minute rainfall 6.3 mm at 1534 UTC > 500 mm/hr > 500 mm/hr (380 mm/hr)

distant, and (with considerably less occasions, five were long-duration (split across two rainfall days). Peak hourly certainty) to estimate the peak values close orographic falls, three were cyclonic rainfall amounts were estimated at 40 mm; there to the centre of the rainfall distribution events with little significant thundery was very little thundery activity. In terms of shown in Fig. 3. It must be emphasised that activity, and the remaining three were the extent of the area affected by over the Hendraburnick Down figures are only intense convective rainstorms. In only two 100 mm of rainfall (2800 km2) the storm reasoned estimates of the rainfall towards of these were the duration and intensity of remains without precedent to this day. the centre of the storm: they are not direct the rainfall comparable to the Otterham measurements and are therefore not value. directly comparable with existing depth- 11 October 1916: 208 mm at Kinlochquoich, western Scotland duration extremes. They should not be 12 November 1897: 204 mm at quoted without this important caveat. Another long-duration orographic fall in Seathwaite, Cumbria western Scotland on 11 October 1916 led to The Boscastle storm in context The publication of British Rainfall com- a daily fall of 208.3 mm at Kinlochquoich in menced in 1860, but it was not until 1897 the notoriously wet Glen Garry (British Since reasonably widespread reliable rainfall that the first reliable record of 200 mm in a Rainfall 1916, Douglas and Glasspoole records began in the British Isles about 140 day was obtained. On 12 November 1897, in 1947). More than 100 mm fell over some years ago, rainfall totals exceeding 200 mm a long-duration orographic fall, 204.0 mm 1800 km2 but less than 10 mm fell in parts of in 24 hours have been reliably recorded on was recorded at Seathwaite in the Lake eastern and southern Scotland. only 11 previous occasions (see for example, District (British Rainfall 1897). This was noted Bleasdale 1963 and Reynolds 1978). The as some 15 mm above the previous daily 200.4 mm measured at Otterham on highest fall on a record at the site back to 28 June 1917: 243 mm at Bruton, 16 August 2004 was the first ‘200’ in the 1844. An estimate of 295 mm in 24 hours Somerset British Isles for just over 30 years, and the was made for one of the monthly gauge first in England since 1957. A brief account A prolonged overnight cyclonic rainfall gave sites on The Stye (now known as Sty Head) over 200 mm in two locations in Somerset of all such events known to the author is at the head of Borrowdale. given below, in date order.* Of these 11 (Mill and Salter 1917). At Bruton (Sexey’s School) 242.8 mm was recorded (confirmed by the measurements at nearby King’s * All falls are listed to 0.1 mm for consistency 25-26 August 1912: 205 mm at School of 215.4 mm and at Pitcombe with previously published accounts (for all falls Brundall, Norwich prior to about 1970 this is a conversion from the Vicarage of 200.7 mm), while a few kilo- In one of the most notable British rainstorms original rainfall measurement in inches) although metres away at Aisholt (Timberscombe) for the intense convective storms the accuracy of of the twentieth century, a long-duration 213.1 mm fell. Some thunder occurred the measurement is likely to be at least two cyclonic rainfall event brought more than during the long night rain, but it does not orders of magnitude less than is implied by this 100 mm in about 24 hours to many areas in appear to have been a prominent feature; precision. Some of the falls are part-estimates, Norfolk and Suffolk (British Rainfall 1912). the highest 2 hour falls were estimated at no including (for example) the effects of hail Near the centre of the storm 205.5 mm fell at more than 50–60 mm. In all, about 2100 km2 224 bouncing out of the raingauge in one storm. Brundall, near Norwich, in about 22 hours received over 100 mm, 35 km2 over 200 mm. The Boscastle storm of 16 August 2004 Weather – August 2005, Vol. 60, No. 8 225 in 2 Flood 170 ± 15 mm phase of the Flood Flood Studies British Rainfall , Chapter 4) – ibid ( 2 hour estimated . . were were predicated on a maxi- 2 Camborne Camborne sounding): the 2 hour (NERC 1975 – Fig II 4.1). The maxi- UTC Flood Flood Studies Report Although the falls in this storm up to The most intense At least ten previous rainstorms within the within rainstorms previous ten least At towards towards the centre of the storm event. The latter figure is at least as high as mated the esti- maximum 2 hour expected rainfall in to the be area methodology used based in upon the the Report mum ‘storm efficiency factor’ as defined in the viz. the ratio of rainfall to precipitable water the in the amount representative air of column during the storm – implied for this storm is (155–185 mm rainfall between compared with a pre- about cipitable water content of 26 6 mm from the 1200 and 7 Boscastle Storm produced 115 mm rainfall at Lesnewth, probably Otterham, 125–130 mm and an at this storm, and 200perhaps 12 mm km or more over 16 August 2004: 200 mm at Otterham, Cornwall the in this paper is thus described event The British the in recorded be to fall mm 200 first be to and the first 30 years, just over Isles for recorded in England for almost More than 100 50 mm fell over about 83 years. km Relevant short-duration intense events rainfall placing the 200After mm fall at Otterham in context as one of the highest 24 hour rain- observed the falls yet within the British Isles, next step was to assess the position of the short-duration rainfall previous within event events. hourly the remarkable, were duration hour 2 fall of 86 mm at ceeded Lesnewth on has numerous been previous ex- (the occasions British Isles summarised sub-hour in records Burt are (2000) 30 – minutes 80 mm at Eskdalemuir, in Dumfries Galloway and on 26 June 97Antrim 1953 Co. Beg, Orra at minutes 45 in mm (Anon 1953); on 1 August 1980 (Woodley 110 mm 1981) and in Oxfordshire on 58 9 June 1910 ( minutes 1910) ). at Wheatley, on 17 January 1974. highest daily This rainfall total yet remains measured in the Scotland (Woodley 1982). maximum rainfall projections in the Studies Report mum ‘storm efficiency’ factor of 3.86. value This is substantially exceeded observed by Lesnewth value the (4.46) extrapolated and Otterham the value (4.83) have may falls hour 2 maximum and that implies been considerably underestimated in the Flood Studies Report British Isles have exceeded 115 mm in UTC 1955, British Meteorological ; approximately GMT British Rainfall showing showing the enormous drifts of , with a peak hourly fall of perhaps received received more than 100 mm and over over 200 mm. 1954, p. 46). 2 2 GMT 50 mm. There was some thundery activity during the storm, but this was not a signifi- cant feature of the event. 825 km In all, about 125 km 17 January 1974: 238 mm at Sloy, western Scotland Another long-duration orographic rainfall event (approximately 24 hours) in western in Adit Main Sloy at mm 238.4 gave Scotland the upper headwaters of Loch during the 24 hours Lomond commencing 0900 pp. 46–48). At Martinstown, approximately 190 in about 4½ hours commencing mm fell 1430 8 June 1957: 203 mm at Cornwall Camelford, On 8 June 1957 a daily fall of 180.1 mm was less than in Cornwall, at Camelford recorded 2004 August 16 the 10of centre the from km rainfall event. Of this in total, 138 2½ hr mm 1230–1500 fell During a prolonged cyclonic rainfall event, Martinstown in Dorset (about 5 km south- west of Dorchester) recorded a daily fall of 279.4 mm. This remains the highest 24 hour total yet reliably measured of totals daily by confirmed was fall The in Isles. the British 241.3 mm at 228.6 211.1Well), (Higher mm at Upwey mm Upwey (Friar at Upwey Waddon), (Elmwell) Wynford and House 200.7 mm ( at 18 July 1955: 279 mm at Martinstown, Dorset days days (153 mm December and was 110 mm on recorded 18 this event would December) have ranked as the largest on daily 17 total on record to Rainfall that time ( 100 mm fell in 1957). the The daily total was judged first to be low hour (Bleasdale owing to a considerable which amount fell during the storm of being lost from hail the raingauge, and the true daily total was estimated at 203 mm. This highest remains daily the rainfall total Cornwall. on A record careful for site survey conducted shortly after the event led to the conclusion that at least 225 mm had fallen over an area 4 1approximately to up by long km wide km just east of Camelford. Severe occurred flooding on the Camel river, bridges were destroyed or where badly damaged, four and there is a photograph in was an intense There itself. hail in Camelford rainfall this around gradient localised storm, 100 over to 2amounting than less in mm km horizontal distance. Magazine . 2 British Rainfall received received a fall of , and 200 mm over 2 1952, Bleasdale and 2 over over 200 mm. 2 daily fall was 129 mm at 1929). All of this fell in about , and 200 mm over just 5 km 2 observed . lies approximately 70 km 2 British Rainfall British Rainfall 100 mm, just 8 km 45 km 17–18 December 1954: 256 mm at Quoich, Loch western Scotland A prolonged 256.5 orographic mm in fall 22½ hours (Cruadhach) at brought on 17–18 December 1954. Had Loch Quoich the fall not been split across two rainfall The The famous ‘Lynmouth Storm’ of 15 August in 1952 fatalities 34 resulted in flood-related the north Devon villages Lynmouth of (Delderfield and 1953). Over heights of the Exmoor, a gauge at Barrow Longstone recorded 228.6 mm in day the ( rainfall Douglas 1952); of 135 mm is believed to have fallen in 5 hours this, approximately of intense thundery rainfall. Up to 275 mm centre the near fallen have to estimated was of this storm. More than 100 mm fell over approximately 400 km 15 August 1952: 229 mm at Barrow in Longstone north Devon – Storm’ the ‘Lynmouth 11 November 1929: 211 mm in South Wales Rhondda Valley, Secondary depressions and a very mild moist, warm sector gave heavy rainfall over orographic wide areas of western England and Wales on 11 highest daily November fall being 1929, 211.1mm the at Lluest Wen Reservoir in the upper Rhondda Valley ( 18 hours steady exceeded rainfall 15 mm/hr. which This highest daily rarely rainfall remains total yet measured in the Wales (Woodley 1982). In more the Valley, than 100 mm fell over approxi- Rhondda mately 55 km 18 August 1924 : 239 mm at Cannington, Somerset A series of very localised intense overnight thunderstorms with hail but remarkably ‘daily’ gave a fall of 238.8 mm at Cannington (Brymore House) on (Glasspoole 18 August 1925). accounts, 1924 it is From believed that than a 200 little eyewitness mm fell more in 5 hours The 0130–0630 h. band of intense narrow: rainfall only was 35 km very north-east of the centre of the August 2004 storm *. * On 10 August 1959 more than 100mm fell in very severe thunderstorms over Exmoor; the highest , Nutscale Reservoir although there was some evidence that the fall elsewhere in Porlock may have reached 250mm ( 1959–60, pp. 77–79). 2 hours: the five most intense storms on Otterham. As much as 200 mm ± 10% may reports, serious flooding has occurred record (where measurements based upon have fallen close to the centre of the storm before at Boscastle: for instance, in late- actual raingauge observations are available) event. At least five previous rainstorms with- October 1996 when ex-hurricane Lili de- are listed in Table 3*. All five storms were in the British Isles are known to have livered a combination of heavy rain and more intense than the estimated fall at reached or exceeded 165 mm in 3 hours: high tides which inundated the lower part Otterham, but it is likely that closer to the these are listed in Table 3. At the centre of of the village; in each of the Junes of 1963, centre of the storm only the Calderdale the storm the fall may have exceeded the 1958 and 1957 when sudden devel- storm of 19 May 1989 (193 mm in 2 hours at existing British Isles extreme value for this oped in circumstances similar to 16 August Walshaw Dean Reservoir) and the duration. 2004; and in July 1847 when a much broad- Hewenden Reservoir fall of 11 June 1956 Over a 4 hour period 181 mm fell at er region was affected (Eden 2004). The (155 mm in 109 minutes) surpassed it in Lesnewth and probably 197 mm at short, steep valleys of north Cornwall and intensity†. Otterham. Close to the centre the fall has north Devon are particularly vulnerable to The The Boscastle storm of 16 August 2004 Over a 3 hour period 148 mm fell at been estimated above as 250 mm ± 20% (i.e. localised downpours. They collect Lesnewth and probably 160–170 mm at possibly 300 mm). Almost all of this would water efficiently from the surrounding have fallen in 4 hours. The observed moors, channel it rapidly into swift-flowing * Clark (2005) provides evidence for a very Lesnewth value ranks as the fourth-highest streams which can rise and fall quickly, and localised fall of approximately 200 mm in little 4 hour measured rainfall on record within take it all out to sea in a few hours. Because more than an hour in the Cheviot Hills in northern England on 2 July 1893. This estimate the British Isles, exceeded only by the 2 hour of their almost instantaneous response to a is based largely upon streamflow evidence Calderdale storm of 19 May 1989, the sudden cloudburst, these valleys are known rather than raingauge observations and is 18 August 1924 Cannington storm (just over as ‘flashy catchments’and they produce true therefore not included in the table. 200 mm in 5 hours) and the 18 July 1955 ‘flash floods’. There have in all probability † Both of these West Yorkshire records – the sites Martinstown event (approximately 190 mm been many more major events in are only 10 km apart – have previously been sub- in about 4½ hours). The extrapolated 4 hour the village within the last few hundred jected to considerable scrutiny. The Walshaw Otterham value of 197 mm has probably years. Dean value was not accepted at the time by the never been surpassed in the British Isles Twice within the past half-century or so Met Office, although Collinge et al. (1990) sug- historical record. and within 100 km distance there has been Weather – August 2005, Vol. 60, 8 – No. Vol. August Weather 2005, gested that “… the reading is likely to be in severe flooding as a result of violent convec- excess of the true value, as a consequence of insplash, though not by much”: the Hewenden How rare an event was this tive storms, at Lynmouth in August 1952 Reservoir value – with a storm efficiency of 5.3 – storm? and just a few kilometres away at Camelford was deemed ‘not acceptable’ in the Flood Studies in June 1957. Intense, damaging convective Report (NERC 1975, p. 33) although both events Media coverage of the ‘Boscastle flood’tend- rainstorms are therefore a rare but hardly were listed in a recent Met Office study by Hand ed to be sensationalised and carried little or unknown feature of the local . It is et al. (2004). no historical perspective. Contrary to media impossible, and possibly meaningless, to

Table 3 The five largest rainstorms within the British Isles for 2 hour, 3 hour and 4 hour durations - approximately 1860 to date, excluding this event, and including only those storms for which documented raingauge observations exist. Figures in italic indicate an estimated total for this duration from a longer-period fall. Based upon original sources together with Jackson (1974, 1979), updated by the author.

Duration Amount (hours) Date Location and duration Reference(s)

2 19 May 1989 Walshaw Dean Reservoir 193 mm Acreman (1989), near Halifax, West Yorkshire Collinge et al. (1990) 11 June 1956 Hewenden Reservoir, West Yorkshire 155 mm British Rainfall 1956 in 109 min pp. 30–31 14 August 1975 Hampstead, north London 140 mm Keers and Wescott (1976) 6 June 1963 Southery, Norfolk 132 mm Jackson (1979) 5 September 1958 Knockholt, Kent 131 mm Rowsell (1960), Ludlam and Macklin (1960)

3 19 May 1989 Walshaw Dean Reservoir 193 mm in 2 hr Acreman (1989) near Halifax, West Yorkshire 7 October 1960 Horncastle, Lincolnshire 178 mm British Rainfall 1960, pp. 61–62 14 August 1975 Hampstead, north London 169 mm in 2.6 hr Keers and Wescott (1976) 170 mm in 3 hr 18 July 1955 Martinstown, Dorset 170 mm See 24 hour falls list 18 August 1924 Cannington, Somerset 165 mm See 24 hour falls list

4 19 May 1989 Walshaw Dean Reservoir 193 mm in 2 hr Acreman (1989) near Halifax, West Yorkshire 18 August 1924 Cannington, Somerset 185 mm See 24 hour falls list 18 July 1955 Martinstown, Dorset 185 mm See 24 hour falls list 7 October 1960 Horncastle, Lincolnshire 180 mm British Rainfall 1960, pp. 61–62 14 August 1975 Hampstead, north London 170 mm Keers and Wescott (1976) in 3 hr

226 The Boscastle storm of 16 August 2004 Weather – August 2005, Vol. 60, No. 8 227 33, Press , 29, (1960) 89, , Weather, British Boscastle HMSO Weather, pp. pp. 227–228 . (1976) The (1976) . Second Weather, pp. pp. 245–257 (1917) The The great (1917) 110, pp. pp. 15–31 (2005) (2005) p. p. 189 89, , 11, 37, Handbook of Meteorol. Mag. Meteorol. Costing Costing the earth: The Radio 4 documentary, Measurement Measurement of (1981) Exceptional rainfall (1925) The The unprecedented (1925) (1960) Storms of 5 1924, pp. 246–255 Flood Flood Studies Report. Volume (1978) Maximum pp. pp. 2–10 (1974) Largest two-hour falls (1981) Weather, (2004) 31, Meteorol. Mag., Meteorol. 1917, pp. 22–30, 45–47 (1975) Meteorol. Mag. Meteorol. ) Meteorol. Appl.,Meteorol. pp. pp. 168–175 NERC Environment Environment Agency Flood Flood study – key findings published. release issued 12 January 2005, available website. on EnvironmentAgency.gov.uk (At the time of going to press in May 2005, the final report had not been pub- lished) Glasspoole, J. rainfall at Cannington, August 18th 1924. British Rainfall Hand, W. H., Fox, N. I. and Collier, C. G. (2004) A study of twentieth-century extreme rainfall events in the with implications for forecast- ing. Jackson, M. of rain in the British Isles. 34, Keers, J. F. and Wescott, P. Hampstead Storm – 14 August 1975. Weather, pp. 71–73 The largest (1979) fall —— of rain possible in a few hours in Great Britain. Kirby, A. Boscastle Storm. broadcast at 2100h on 16 December 2004 Ludlam, F. H. and Macklin, W. C. The Horsham hailstorm of 5 September 1958. Met Office II - Studies Meteorological Reynolds, G. precipitation in Great Britain. pp. 5–11 and 5–14 Mill, H. R. and Salter, C. rain storm of June 28th, 1917. Rainfall Natural Environment Research Council ( Edition. Vol. 5 Vol. – Edition. Precipitation and Evaporation Meteorological Instruments Meteorological pp. 162–166 Rowsell, E. H. September 1958. pp. 252–257 Woodley, K. E. of 1 August 1980 over the North Antrim Plateau. (1982) Heaviest —— daily rainfall in Scotland. © Stephen Burt, 2005. doi: 10.1256/wea.26.05 , ., 1897, , 44, , p. , 1912, 45 , p. p. 156) and Meteorol. Mag. Meteorol. Weather pp. pp. 344–7 pp. pp. 353–67; The Lynmouth pp. pp. 11–42 J. Inst. Eng Water 45, , Weather British Rainfall (1990) Radar 82, 81, 73, 1916, pp. [60] to [62] , 1929, pp. 71–77 , ., Sunday Telegraph pp. pp. 354–365 pp. pp. 356–63 (1953) British Rainfall (1897) The The rainfall (1897) of (2000) A cluster of intense 45, pp. pp. 319–333 55, Weather 1910, pp. [16–17] and 1952, pp. 45–47 (1957) Rainfall at Camelford, , (1989) (1989) Extreme rainfall in The The Raleigh Exmouth Press, pp. pp. 92–97 29, (2004) Extreme precipitation (2005) The The cloudburst (2005) of 2 July 60, (2004) This This summer's (2004) frequent (1953) Thunderstorms Thunderstorms of (1953) June 26, Weather, Weather British Rainfall British Rainfall Meteorol. Mag. Meteorol. Meteorol. Mag Meteorol. pp. pp. 339–343 (with photographs pp. 45–55 Burt, Stephen —— (1952) Heavy —— falls on rainfall days. British Rainfall rainfall summer Berkshire, West events in 1999. Collinge, V. K., Archibald, E. J., Brown, K. R. and Lord, M. E. observations of the Halifax storm, 19 May 1989. Clark, C. 1893 over the Cheviot Hills, England. Weather, Doe, R. K. and run-off induced flash flooding at UK,Boscastle, Cornwall, 16 August 2004. J. Met (UK), Delderfield, E. R. flood disaster. Douglas, C. K. M. and Glasspoole, J. (1947) Meteorological conditions in heavy orographic rainfall in the British Isles. J. Q. R. Soc.Meteorol. Eden, P. heavy downpours. 22 August 2004 November November 12th 1897. numerous numerous photographs of storm damage on Exmoor between pp. 368–9 British Rainfall pp. [118] to [120] The rainfall (1910) of —— June 9th, 1910. British Rainfall [119–120] (1912) Great —— rain storm of August 25th–26th, 1912. pp. 28–47 The rainfall (1916) of —— October 11th 1916. The rainfall (1929) of —— November 11th 1929. Acreman, Acreman, M. Calderdale, 19 Calderdale, May 1989. pp. 438–46. See also comments by Nicholls, J. M. ( 86, between pp. 336–7) The distribution (1963) of —— exception- ally heavy falls of rain in the United Kingdom, 1863 to 1960. 17, Bleasdale, A. and Douglas, C. K. M. (1952) Storm over Exmoor on August 15, 1952. 156–7) Anon 1953. Bleasdale, A. on Cornwall, June 8, 1957. reply reply by M. Acreman, ( References Acknowledgements Roger Bailey from the Environment Agency Field Monitoring & Data Team, much provided Area, of the rainfall informa- Cornwall tion upon which this analysis is based, and Malcolm Kitchen from the Exeter Met provided Office the at radar rainfall used images in Fig. 6. Martin Bartlett Kidds from and the Graham National Meteorological Library in Exeter were exceptionally helpful and prompt in responding to requests for without which this analy- material reference sis would have been incomplete, and Philip Eden provided helpful comments and feed- back on my final draft version. I extend my thanks to all. Summary The devastation by runoff flooding occurred that in Boscastle and Haven and, to a lesser extent, in other parts Crackington of Cornwall and Devon on 16 August 2004 was due to an exceptionally intense storm, producing rain- almost 200 mm of rainfall within 4 hours at the wettest gauged and site quite possibly 250 mm or more in the same period close to storm. the The centre storm of that the was Hendraburnick centred Down upon ranks as one of most the extreme British rainfall events record, and the extrapolated fall of on 197 mm at Otterham a represents new 4 hour British Isles rainfall extreme. Equally were exceptional the intense rainfall gradients on the south-east flanks of the storm, amounting to 50 mm/km or more. attempt to assign accurate return periods to periods return accurate assign to attempt the extreme events of 16 August 2004, but the preliminary report of the Environment Agency into the flooding (Environment Agency 2005) in indicated that Boscastle the heaviest hourly rainfall period had of approximately 400 a years and the return 3 hour fall approximately should be 1300 noted, however, that years. the It return period of the assessed as resultant approximately 400 years) bears flooding (also only a coincidental relation to periods the of the rainfall which caused it. return