sign in sign up
<<
Home , Boscastle, Cloudburst, Hendraburnick, Otterham

230 – August 2005, Vol. 60, No. 8 leading to flooding on 16 August 2004 August 16 on flooding to leading conditions the of analysis Meteorological : The ags Ms o nrh onal a wet was north of Most gauges. daily in available shown all from is derived 2004 Fig.1, August of half first the open moor. The distribution of anomalies for the of those approaching amounts receive catchment Valency the of parts upper the while coast, the of parts is other than wetter Boscastle moor. rainfall on the observed half than less receiving and of vicinity the in areas driest the with region, Cornwall north the across normal. than drier remain toground Boscastle area had about average rainfall, allowing the the that indicate available observations though region, the in above- average generally was rainfall July June, Following below-average rain from March to Antecedentconditions focused area. Boscastle the in conditions extreme which mechanisms the and small-scale rain the of and characteristics temporal spatial the , the generated that conditions be meteorological the will event, Evidence event. presented onthesoilconditions priorto the the of tions experimental high-resolution modelsimula- from insight additional with analyses, and observations operational using event, the to leading conditions meteorological the of of loss no analysis an presents was paper This life. there human event, the of ness sudden- and severity the given perhaps remarkably, and thankfully emergency though the services, by safety by to helped people were damaged of number substantial severely A flooding. was of coast Cornwall north the on Boscastle the of 2004, village August 16 of afternoon the On Exeter Office, Met ForecastingResearch MayBryony PeterClark Golding Brian vrg Ags rifl vre markedly varies rainfall August Average moisture deficit ( deficit moisture this period. this K wt a os suhwsel gradient south-westerly moist a with UK, south-west movinglow-pressure the westof areathe to over was dominated by a complex, slow- flow Synoptic-scale environmentSynoptic rm8–2 mm 80–220 from dropping Boscastle the around values of rainfall, range above-average the with tent lu (Smith satellite and rainfall radar including mation infor- meteorological using moisture soil of Probability evolution the Exchange the diagnoses Model) Distributed Surface incorporating Office Scheme (Met model in of reduction a excess diagnosed model The in mm. 100 catchment Valency the in values with variability spatial considerable eutn pten s osrie b the by raingauges. available of distribution constrained the is pattern so resulting variability, spatial was considerable There normal. than more 25% about receiving Boscastle with period, this during Fig. 1 Pr The Met Office’s Met The SMD between 1stand16thAugustbetween consis- ecipitation anomaly map for south-west England, 1–15 August 2004, relative to 1961–90 averages1961–90 relativeto 2004, August England,1–15 south-west for map anomaly ecipitation et al. et SMD SMD MOSES 05.Te eutn soil resulting The 2005). ) for Cornwall showed Cornwall for ) o4–8 mm 40–180 to - PDM land surface land SMD in eto wt this with vection there was no water in left it, the average rain tropopause.the reach the reach to equilibrium level, layerand another 15 minutes to boundary the from air for minutes 15 about taken have would it speed this At . the in observed 5ms and about probably was velocity vertical entrainment mean the cloud, the for of depth the over averaging allowing after ovcie vial Ptnil Energy ( Potential Available The km). (9.7 Convective hPa 250 at tropopause the hPa 450 only km). The highestcloudtops would beat (6.5 at was stop would most where cloud. level equilibrium the growingHowever, rapidly a produce would m. 900 about at lowestlayersthe in Above, base instability strong a from cloud tive convec- forming readily layers lower moist very with development, storm for primed (British Time)). The atmosphere was u vria vlct o aot 8ms m 18 about of velocity vertical mum iue sos h rdood sounding from at 1200 radiosonde issue). the this shows 2 Burt, Figure of 1 (Fig. Cornwall over CAPE If the total column of air were lifted until lifted were air of column total the If –1 ws bu 10Jkg J 170 about was ) , a value supported by the absence of absence the by supported value a , CAPE ol hv a maxi- a have would UTC –1 (1300 local time Udlt con- Undilute . –1 so , The Boscastle flood of 16 August 2004 August 16 of flood Boscastle The Weather – August 2005, WeatherAugustVol.No. – 8 60,

Fig. 2 Tephigram showing 1200 UTC radiosonde from Camborne (actually Fig. 3 300 hPa height (contours in gpm) and Potential Vorticity (colours in s–1) at released 1116 UTC) 1200 UTC 16 August 2004 rate would be given by the precipitable eastern Atlantic upper vortex (Fig. 3). The water divided by the lifting time. In this case, resulting potential vorticity maxima in Fig. 3 26 mm of precipitable water lifted to can be traced in the dark, dry bands of the dryness in 15 minutes corresponds to about Meteosat-8 upper tropospheric water 100 mm hr–1 rain rate. In real clouds, much of vapour image in Fig. 4 and were associated the rainfall usually evaporates into the with the surface troughs approaching surrounding air giving a rainfall ‘efficiency’of south-west England. The first of these less than 50%. Maximum 15-minute rain troughs, between Devon and Brittany (Fig. 1 rates observed by raingauges and the radar of Burt, this issue) with its apex over south- in this case were 80–100 mm hr–1, indicating ern Cornwall shown in Fig. 4, exhibited rota- an unusually high efficiency, while hourly tion in the satellite imagery, indicating accumulations of up to 60 mm indicate that upper tropospheric vortex development. this high efficiency was maintained over This is consistent with the backing of the jet multiple cloud lifecycles without a break. stream winds at tropopause level shown in The high tropospheric relative humidity the Camborne ascent (Fig. 2) and may have (80% or more below 700 mbar) contributed been the source of observed surface pres- to this high efficiency. sure falls of over 1 mbar recorded in Fig. 4 Meteosat-8 image in the upper tropospheric The near-surface wind in the Camborne Cornwall. The effect of these larger-scale water vapour band at 1230 UTC 16 August 2004 ascent (Fig.2) was south-south-westerly, processes on storm development was to veering to south-westerly 7.5 m s–1 (15 kn) at create an environment of weak uplift, main- and subsequent development of precipi- the top of the boundary layer. There was taining high humidity and hence support- tating cells along this convergence line weak, unidirectional shear from there up to ing the retention of cloud water in the between 1100 and 1135 UTC in radar cloud top – the wind remaining south-west- atmosphere, which contributed to the imagery. The mean speed of movement of erly, increasing to 17.5 m s–1 (35 kn) at unusually high rainfall efficiency. each cell was close to 10 m s–1, consistent 400 mbar. This structure ensured that any with the mid-level wind, while downstream downdraught would be down-wind of the Initiation and development of cell development resulted in an apparent initiation point. It did not favour develop- propagation speed closer to 15 m s–1. New ment of either multi-cell or storms, cells also formed upstream near the original which require directional shear of at least 20 The extreme rainfall accumulations location, so that each initial shower spread degrees between cloud base and the height observed in the Valency catchment above out into a line of storm cells, spaced at of origin of the downdraught (Pierce and Boscastle, resulted from prolonged very intervals of about 5 km, appearing as a Cooper 2000). The wind at the middle of the heavy rain over the four-hour period continuous line on radar imagery. The line storm layer (~500 mbar), was south-west 1200–1600 UTC. The operational rainfall was also evident on satellite imagery, 12.5 m s–1 (25 kn) consistent with the radar data showed that this was produced especially in its early stage before anvils observed movement of the storms (see by a sequence of convective storms that started to spread. Figure 6 shows the line of below). developed along the north coast of storms originating near at the The efficiency of rainfall production from Cornwall. Each storm element started as a head of the Camel estuary. the convective storms was enhanced by non-precipitating cumulus either near the The track of the rainfall cells varied slightly large scale uplift induced by synoptic-scale Fal Estuary or further north towards the during the 1200–1600 UTC period, but forcing in the eastern Atlantic. At 1200 UTC Camel Estuary. Rapid cloud development between the Camel Estuary and Bude, the on 16 August 2004, south-west England was started as each cell encountered con- variation was sufficiently small to ensure under the left exit region of a vergence near the north coast, in the vicinity that the heaviest rain fell on the same coast- maximum on the south-east flank of the of Wadebridge. Figure 5 shows the initiation facing catchments throughout the period – 231 232 Weather – August 2005, Vol. 60, No. 8 The Boscastle flood of 16 August 2004 right by an additional 25 km for clarity. See Fig. 7 for rainratekey.for Fig. clarity.7 See for km 25 additional an by right f aiu ri hs oe aot km 2 about north-west. moved has rain maximum of Boscastle, of axis the that shows inspection close though vicinity the in similar ably remark- look 7) (Fig. apart hours two taken distribution rainfall radar the of ‘snapshots’ 1130 for image satellite visible resolution high Meteosat-8 Fig.6 cells,1100–1135 storm 2nd and 1st the of evolution Initial Fig.5 hs hs o dvlpet produced development of phase This (a) UTC Fig. 7 Estimates of rainfall rate at 2 km resolution from the Cobbacombe radar at 1330 Cobbacomberadarat the from resolution km 2 rainfallrateat of Estimates Fig.7 water rapidly turned to ice visible crystals, in cloud remaining levels, these At °C. temperature –54 was the where km) (9.7 sufficiently mbar 250 were storms energetic to reach the tropopause fewat around a Bude, not did convergencecoastalline. the distort that downdraughts weak with cells storm packed closely of development the enable to enough shallow was it tion, precipita- heavy generate to enough strong related to the fact that while convection was been have to appears Boscastle the of vicinity in precipitation extreme the origin, microphysical its Whatever freezing. tact con- by rapidly grown have would crystals seed These Brittany. over initiated earlier storms of anvils outflow the from ice with seeded being were clouds ice the that initiatedicate in- images to satellite However, enough processes. cold just been only have would which °C, –20 to °C –15 around of temperature top cloud a implies This km). (6.5 mbar 450 at ascent Camborne the in level equilibrium the of vicinity the in imagery) satellite on (based tops cloud with depth mid-tropospheric to growth rapid t ltr tg, ute nrhes near north-east further stage, later a At UTC 16 August 2004. Each time is shifted is time Each 2004. August 16 (b) UTC (a) and 1530 and (a) South South Wales. into Channel Bristol the north across into and Devon east spread which cells storm of rows radar new generating 7(b)), and (Fig. imagery 8) (Fig. satellite the in visible is arcs such of succession A Bude. of north the to arc eastward an in bow to it causing line, convergence the a distorted which front gust a by in resulting downdraught, accompanied strengthened was precipitation This 7(b)). (Fig. line precipitation main the of north the to ground, the at intensity tion these storms was reflected in their precipita- of vigour greater The 8). (Fig. area Hartland the over shield cloud large a of growth the cetd y h Mt fie ult control quality Office Met been the by all accepted have observations daily conditions the and winds, surface favoured accurate raingauge measurements light With distribution Rainfall 1530 for image visible resolution high Meteosat-8 Fig.8 UTC 16 August 2004 August 16 UTC (b), 16 August 2004 August 16 (b), (p. 233 follows WeatherfollowsLog) 233 (p. The Boscastle flood of 16 August 2004 Weather – August 2005, Vol. 60, No. 8 233 followed by followed UTC , the gauge re- UTC relative relative to the check TBR from from the Tipping Bucket rain (Fig.10). (Fig.10). (See Burt, this issue, for UTC 1997) was re-run at higher resolution UTC half-hour half-hour periods of heavier rain centred on about 1235, 1315 and 1415 corded an uncorrected peak nearly rain 300 rate mm/hr. of Given that shortfall the of 20% the gauge would have occurred predominantly in the heaviest periods reached briefly have may rate rain maximum of rain, the true 400 mm/hr. more more detail on the timing and location the of heavy rain bursts.) period, During at the about last 1535 a more continuous period 1615 from 1455– Numerical model simulations The highest resolution Met Officetional forecast model opera- has a horizontal grid length of about 12 km represents and only features so larger accurately than about 60 km across. Convection has to be sented repre- by an estimate of the change in the flow resolved resulting from a population of convective clouds in equilibrium with the larger-scale forcing. Not surprisingly, this model produced a bland forecast, heavy convective rain forecast to occur any- with where in Devon and Cornwall. In order investigate to the processes extreme rainfall, the that Unified Model (Cullen led et al. to (1 km nested in 4 km nested in the standard 12 km) in a form which allowed individual to form, albeit resolved. not The 1 well km configuration was with run double (76 levels) compared with the the 12 and 4 km versions, vertical to resolve resolution better the layer boundary processes. Such requires a great deal high of computer time, so resolution the 1 km resolution model was run over an Time (UTC) ) from 2 km UTC record record is shown in TBR ) raingauge ) totals raingauge (spot values) UTC 12:00 12:15 12:30 12:45 13:00 13:15 13:30 13:45 14:00 14:15 14:30 14:45 15:00 15:15 15:30 15:45 16:00 16:15 16:30 Fig. 10 Fig. profile 1200–1630 rate rain uncorrected, Smoothed, gauge on 16 August 2004 0.00 50.00 The The Lesnewth 300.00 250.00 200.00 150.00 100.00 (mm/hr) Rain rate Fig.10, Fig.10, computed from the time taken for each millimetre to fall. This smoothed produced profile a compared to the 0.2to only recorded were which counts, mm raw 10-second precision. There was consider- both able at variability, short timescales of 5 to 10 minutes, associated with cells, individual and at longer timescales, with three tions from those observed by gauges not solely are due to sampling differences. The radar spatial pattern indicates heaviest that the total probably occurred a few kilometres to rainfall the south-west of near the A39. accumulation s) TBR recorded TBR istribution of on rainfall accumulated 16 August 2004 : 5-hour totals (1200–1700 observation which was about 20% low. Twenty-four Twenty-four hour raingauge totals from Continued Continued from p. 232 all available raingauges are shown as point values in Fig. 9, superimposed on the distri- bution of five-hour accumulations derived from the Cobbacombe radar. The patterns showing consistent, are considerable spatial variability with an elongated strip of high accumulations through (76.5 mm), Lesnewth (184.9 mm), Otterham (200.4 mm), Credacott Crowford Bridge (75.5 mm). (123 Of mm) these, only daily confirmed are Otterham and Lesnewth and gauges. After correction to match the daily check gauge, the Lesnewth Burt (this issue) discusses the issue of rain gauge accuracy further. Conditions were also favourable for accurate radar observations with rainfall a high freezing level and no anomalous propagation. Two cover radars the Boscastle area at a range of less than 100 km. recorded a slightly Of higher maximum storm these, accumulation than Predannack, but there is Cobbacombe very little overall bias evident between the two sets of data. In the areas coverage of common at 2 km resolution, recorded both radars nine accumulations exceeding 96 mm. pixels (36 sq km) with TBR process. process. Tipping Bucket Rain gauges ( maximum short period accumulations 82 of mm in 1 hour, 148 mm discusses in issue) 183(this Burt hours. 5 in mm 3 hours, and return periods for these some- the radar were Maximum accumulations. values from but due what to lower, the very high rainfall accumulation gradients, it is not possible to be sure that the differences in accumula- suffer suffer from under-reading at high rain rates Lesnewth 24-hour the in reflected is this and corrected Cobbacombe radar data radar Cobbacombe corrected and (colours) 24-hour (0900–0900 Fig. 9 D 9 Fig. 234 Weather – August 2005, Vol. 60, No. 8 The Boscastle flood of 16 August 2004 every 4km. every (c)valuestemperature sea land tofixed and fluxes orographysurface orographyflat with flat (b), Model.(a), Fullsimulation Fig. 12 Surface wind (variable length arrows) and convergences arrows)and in (colourlength scale(variable wind Surface Fig.12 (a) (c) Fig. 11 Accum (b) (a) raphy (b), flat orography with surface fluxes and land temperature fixed to sea values (c).valuestemperature sea land tofixed and fluxes orographysurface flat with raphy(b), that are highly predictable. highly are that factors by determined is location the that suggests forecast. This the of length the ing consider- especially remarkable, is location model. km ofthe The accuracy sented by a1 repre- realistically scale minimum the km, to 5 averaged observations radar to similar mm, 50 about of accumulations maximum The domain. model with precipitation, km intense simulated model 1 the of) (part in 0000 at analysis km 12 Cornwall. The model was initialised from the centred km, on area covering only 300 × 300 ain f rcptto fo 1200–1700 from precipitation of lation develop.featuresto resolution fine Figure 11(a)shows thepredicted accumu- ulated precipitation in mm from 1200–1700 from mm in precipitation ulated UTC

to give time for time give to UTC b (c) (b) ~.0 s (~0.001 Smsn 94. n hs cain such occasion, this On 1994). (Simpson coast the along front sea-breeze stationary quasi- offshorea produce can sea-breeze a and wind an between Interaction sea. and generated by differential heating of the land gradient pressure onshore an from results sea-breeze,the which is vergence One lines. con- such generate could that mechanisms ec a 1100 conver- at its gence and wind m 10 of distribution e ms m few (a motion vertical substantial generate to persistent in the forecast and strong enough highly was coast, north the to close shading iue 2a sos h corresponding the shows 12(a) Figure UTC 16 August 2004 from 1km grid length integrations of the Unified Model. Full simulation (a), flat orog- flat Model.(a), FullsimulationUnified the integrationsof length grid 1km from 2004 August 16 –1 –1 a cod ae Tee r two are There base. cloud at ) , hw a bns f orange of bands as shown ), UTC –1 Srn convergence Strong . ) at 1100 at ) UTC , 16 August 2004 from 1km grid length integrations of the Unified the integrationsof length grid 1km from 2004 August 16 , in generating the convergence line, re-runs line, convergence the generating in and uplift. convergence by marked is which jet, coastal the and land over flow backed the al. (Hunt sea the over flow ambient the than a is coastal jet that may be significantly result stronger The sea. smooth the to on face turning ofairflowing from arough landsur- and acceleration by caused is and coast, the to parallel is sea the over wind the when es aris- mechanism second Thebacking. tional fric- to due land the over offshore only and coast, the along being wind ambient the by complicated been have would interaction In order to clarify the processes involved processes the clarify to order In 2004) with a resultingbetween boundary . Wind arrows are plottedarrows areWind . et The Boscastle flood of 16 August 2004 Weather – August 2005, Vol. 60, No. 8 235 . , Met (2000) PB 1 3 in the Met EX Sea breeze and local The Andre J. Robert MOSES Q. J. Q. R. Soc., Meteorol. (Charles (Charles A. Lin, Rene (Ed) (2005) Boscastle and (1994) Meteorol. Appl.Meteorol. (2004) Coriolis effects in Forecasting Res. Tech. Rep. No. Res. Tech. Forecasting pp. pp. 2703–2731 . Cambridge University Press Met Office, UK. Available from Office Nimrod nowcasting system. Submitted to Office, Office, UK. Available from www. metoffice.com/research/nwp/ publications/papers/technical_reports/ 2005.html and Hunt, J. A., C. W. Rottman, R., J. Orr, Capon, R. mesoscale flows with sharp changes in surface conditions. 130, C. Pierce, E. and A. M. Cooper Comparison of the performance of 2 km resolution object-oriented model and Nimrod advection precipitation nowcast schemes. 350, www.metoffice.com/research/nwp/ publications/papers/technical_reports/ 2000.html Simpson, J. E. numerical numerical methods in atmospheric and ocean modelling. In: memorial volume Laprise and Harold Ritchie (Eds.)) Canadian Meteorological and Oceanographic Ottawa,Society, Canada, pp. 425–444 W. Golding, B. north post Cornwall flood event study – meteorological analysis of the conditions leading to flooding on 16 August 2004. Rep. No. 459 Res. Tech. Forecasting wind Smith, R. N. B., Blyth, E. W., M., Finch, J. Goodchild, S., Hall, R. L. and S. Madry, (2005) Soil state and surface hydrology diagnosis based on Correspondence to: Dr. B. W. Golding, Met to: B. Correspondence W. Dr. Office, Road, Exeter FitzRoy email: [email protected] © Crown copyright, 2005. doi : 10.1256/wea.71.05 , UTC , this issue, . –1 Weather (1997) An overview of (2005) Cloudburst upon and again from 1450 to 1615 UTC Burt, S. The Hendraburnick Boscastle Down: storm of 16 August 2004. pp. pp. 219–227 Mawson, M. T., Davies, Cullen, M. J. P., H., S. James, C. J. and A., Coulter, Malcolm A. numerical methods for the next generation UK NWP and model Mapping information is Crown Copyright The work described in this paper was References 6.5 km and spawned adjacent cells the front, along resulting in a continuous line of small but intense rain cells, which north-east moved with the mid-tropospheric wind. This line was most intense 1330 from 1230 to moving moving slightly seawards between the two periods. Over 200 mm of rain was recorded at the head of the Valency river catchment above Boscastle, and peak rates may briefly have reached 400 mm hr Acknowledgements The assistance of many colleagues from the Met Office and the Environment and of Agency, eye-witnesses, is gratefully acknow- ledged, in particular: Ross Melville, Roderick Smith, Andrew Bennett, Malcolm Kitchen, Andy Mark Beswick, Humphrey Cooper, Lean, Nigel Roberts, Sarah Watkin, Michael Saunby, Joanne Crawford, Mark Smees, Tim McInnes-Slegg. Rod and Quinn Roger Wood, 2005, reproduced under licence number 100008418. commissioned by the Agency. Environment “View from the The bridge”: dark sweep waters into smashing lower Boscastle, cars into buildings (courtesy Books) Grundy © Wayne Truran Summary Synoptic-scale developments over the east- ern Atlantic on 16 August 2004 created moist, a unstable environment with weak uplift over between Cornwall. frictionally-backed The winds over land interaction and a developing sea-breeze ated a stationary gener- sea-breeze front along the north coast. Cumulus clouds moving into it from the south, grew rapidly to about of the 1 km model setting were the land height to sea level over the performed, first south-west peninsula and then also setting the and fluxes moisture and heat surface the surface temperature over land typical to of the values sea. Figure 12(b) shows that the influence on changing precipitation the land of height just is rainfall pattern is simpler and shifted small. to the The north-west so that the heaviest rain falls in the sea, consistent with the from the loss hills allowing of a strengthened drag off- shore wind. Figure 12 shows that the con- vergence line is also more organised closer and to the coast in this case. the Removing enhanced surface fluxes (right panels) removes the Boscastle precipitation peak almost This completely. is expected, as sur- face heating is required to initiate convec- Fig.tion. However, that the coastal 12 shows convergence has also disappeared. Thus, it appears that the convergence line sea-breeze was a front whose determined by a position subtle balance was and , retardation the gradient between backing of the wind over land, and differ- ential heating.