176 source: https://doi.org/10.7892/boris.58528 | downloaded: 26.9.2021 Weather – July 2014, Vol. 69, No. 7 on the Twentieth Reanalysis Century 1944 compared withaclimatology based windsinJune ropical andsurface cyclones landingofJune1944:extrat- The D-Day Bieli, Michael SprengerMichael this group offorecasters hadto decide if called ‘suspended animation’ (Cox, 2002), waiting inastate ofwhattheircommander is referred to asensembleforecasting today. to Eisenhower –anearlyexampleofwhat ing thedebate andpresenting thefindings the Britishmeteorologist JamesStagglead- consensus intelephone conferences, with made separate forecasts andthensought US President) Eisenhower. Dwight They first Allied Supreme Commander, General(later the USAirForce, were assembled by the Dunstable group), theBritishAdmiralty, and British Meteorological (referred Office to as 2001). Three separate teams, from the Douglas, 1952; Stagg, 1971;Petterssen, weather forecasting sideaswell (seee.g. accompanied by onthe acombined effort shouldbe joint multinationalundertaking and landforces. wasdecidedthatthis It western Europe, would usealliedair, naval launched theinvasion ofGerman-occupied weather forecastimportant ever made. go aheadwould come down to themost for forecasts. theD-Day Their decisionto in three different teams were responsible circumstances, sixmeteorologists working Under thepressure ofthesehighlydramatic account onD-Day, 2013). seeAtkinson, fate ofthewarinEurope recent (for avery million soldierswere atstake, aswell asthe of weather. The lives of more thanhalfa totally dependentonanaccurate prediction was approaching, thewholeenterprise was invasion ofNormandyby thousandsofships Near theendof World War II,astheD-Day perspective meteorological Historical from overview a Introduction 2 1 Christoph Welker, Institute forAtmosphericandClimate CentreOeschger forClimatechange Science, ETHZurich,Switzerland ofBern,SwitzerlandUniversity Research andInstituteofGeography, While theAlliedcombat troops were Operation Overlord, theoperationthat 2 Nicolas Piaget 2 1 Melanie Melanie 2 and and ginally acceptable. The Germanarmy was the invasion weather on6June wasmar- of opportunity’ for theassault.Indeed, weather thatwould openasmall ‘window periodofimprovedthe forecast ofashort the morningof5Juneandwas basedon decision to invade on 6Junewasmadein tem over theAzores was stillprevailing. The flow associated withthehigh-pressure sys- fairly calm,becausetheweak anticyclonic the weather intheChannelregion was still However, atthetimethisdecisionwastaken, landings.to thepostponementofD-Day Stagg presented to GeneralEisenhower led (Ackerman andKnox, 2006). sistently indicated deteriorating weather together atDunstable,working justascon- Douglas’s intuition,bothmeteorologists hand, Petterssen’s upper-airanalyses and able weather on5June1944.Ontheother historical analoguesthatcalledfor accept- (Cox, 2002).Krick’s team consistently found increasing tensionandoccasional hostility were foughtoutinanatmosphere of Krick, Sverre Petterssen andtheAmerican Irvin forecasters, theNorwegian especiallybetween signals, critical differences amongtheweather long telephoneconferences over scrambled final consensus. three teams were strugglingto arrive ata became more and more anxious, andthe With approaching, D-Day thecommanders much ofaguidelinefor thepredictions. and analogueforecasts could notprovide April. Undertheseconditions, climatology with conditions more likethoseduring 1948; Ambrose, 1994;Fleming, 2004), exceptionally stormy (e.g. Montgomery, turned into a ‘forecaster’s nightmare’. was It May turnedto June, theplacidweather with persistent calmweather. However, as hoped for astable high’‘blocking associated for thehighlyvariableUKweather. scientificterritory,still unknown especially weather for several days andbeyond was in advance –atatimewhenforecasting conditions hadto bemadeatleast2days or not.Ontop ofthat,theforecast ofthese to goforward on5June1944 sion inhistory inva-seas) would permitthelargest military conditions (e.g. lightwindsandnon-stormy Finally, windy forecast thebleak, that Given thisbackground, Stagg andhisteam Behind thescenes, during

speed incomparison to theclimatology. wind andnear-surface frequency cyclone ness ’ ofthismonthisassessedinterms of ern France duringJune1944,the ‘stormi- conditions over theBritishIslesandnorth- uncommonly unsettled andstormy weather area. stressed Becausehistorical the reports ing seaswell inthelanding andvisibility we regard asthekeyvariablesdetermin- on windandatmosphericmoisture, which landingsisexaminedwithemphasis D-Day Office. The weather evolution duringthe providedtorical weather by charts theMet backto 1871,aswellwhich extends ashis- 2011), anovel globalatmosphericreanalysis Reanalysis (20CR;CompoCentury Our analyses are basedonthe Twentieth tion oftheweather conditions inJune1944. landings setsthestagefor ourinvestiga- The historical background oftheD-Day weather –goalsandquestions Reconstructing theD-Day of the atmosphere, and the uncertainty in of theatmosphere, andtheuncertainty produce anestimate ofthecomplete state Filterthe EnsembleKalman technique to numerical weather-prediction modelusing ensemble members)ofintegrations ofa term forecast from anensemble(with56 were combined with ashort- observations from 1871to present. Surface-pressure for insix-hourly theperiod intervals tainty global weather conditions andtheiruncer- use ofthe20CRdataset,whichcontains the weather inearlyJune1944,we made that are noteasilyobserved. To reconstruct andanalyses ofvariables of observations porally andspatiallyconsistent synthesis numerical models, they produce atem- ofmeasurementrating avariety datainto weather andclimate variability. Byincorpo- tool instudyingbecome animportant Retrospective analyses (orreanalyses) have The 20CRdataset cyclone identification The 20CRandobjective of theend World War II. Normandy beachesmarkedthebeginning caught by surpriseandthelandingson t al. et , that estimate. Monthly sea-surface temper- the months of June in the period 1871–2008 the landing by 1 day (Douglas, 1952). Due ature (SST) and sea-ice distributions from based on the 20CR ensemble data set. to the rapid movement of the cyclones, the the HadISST data set (Rayner et al., 2003) cyclones were expected to pass the Channel served as boundary conditions. For further region on the morning of 5 June.

Early June 1944 in 20CR landing of June 1944 The D-Day details on the 20CR refer to Compo et al. For the period in early June 1944 analysed (2011). Among others, the 20CR data set and historical weather charts in this section, the ensemble spread among was intended for the assessment of decadal- In this section, we provide only a brief the 56 SLP fields of the 20CR for the NA scale climate variability. Furthermore, the overview of the weather dynamics during and European sectors is generally small and 20CR can be used as a kind of ‘time machine’ D-Day and the preceding days, because ear- none of the fields shows a fundamentally to study historical weather events. lier studies have investigated the meteoro- different pattern; in particular, the ensemble In this study, we analysed different logical situation (e.g. Douglas, 1952; ECMWF, members hardly differ in the spatial extent atmospheric variables from the 20CR data 2004). The purpose is rather to show that of the Azores high. It is therefore sufficient set to assess on the one hand the synoptic the 20CR is able to capture the main fea- to analyse the ensemble-mean SLP fields weather situation, in particular the location tures of the D-Day weather, compared with in Figure 2, without studying all ensemble and development of cyclones in the North historical weather charts provided by the members individually. Weather – July 2014, Vol. 69, No. 7 Atlantic (NA) and European sectors during Met Office. On 4 June (0700 UTC) the centre of the east- June 1944; on the other hand, the 20CR was According to historical weather charts, ern cyclone, which had rapidly deepened, was used to describe surface winds and asso- two relatively weak surface cyclones were located northwest of Scotland (Figure 1(b)). ciated wind-generated waves that could located over the NA on 3 June 1944 At the same time, another cyclone was have seriously endangered the ships and (0700 UTC): one south of Iceland, head- already forming near Newfoundland. The paratroopers in the English Channel region. ing towards the British Isles, and another 20CR (Figure 2(b)) again agrees well with Furthermore, atmospheric moisture is used one east of Newfoundland (Figure 1(a)). At the historical chart. In the afternoon of 4 to estimate the probability of bad visibility, this time, the English Channel region was June, the cold front associated with the deep cloud coverage, or even precipitation. located between two frontal zones. The cyclone over Scotland was located over Great synoptic situation depicted in the histori- Britain (not shown). An increase in atmos- Surface cyclones and associated cal charts is captured by the 20CR, with a pheric moisture over the English Channel cyclone centre east of Newfoundland and region (Figure 3(d)) was associated with the high wind speeds in the 20CR the possibility of a second cyclone south of approaching cold front, and near-surface ensemble Iceland, although not visible as a closed SLP wind speed increased substantially over the To further examine the synoptic weather contour in Figure 2(a). Based on the weather course of the day (Figure 3(a)). situation during June 1944, the cyclone situation on the morning of 3 June, the At the break of dawn of 5 June, rain was identification algorithm developed by development of the weather that was to be falling from overcast skies and gale-force Wernli and Schwierz (2006) was applied to expected for 5 June was still too uncertain winds drove large waves to the beaches of the 20CR ensemble data set. In this algo- to justify a postponement of the landing. the Normandy. The gloomy forecast made rithm, a surface cyclone is identified as the However, the weather charts for 1800 UTC at Dunstable Meteorological Centre thus finite area that surrounds a local minimum (not shown) were unequivocal to postpone proved to be well justified. The cyclone in sea-level pressure (SLP) and is enclosed by the outermost closed SLP contour. Based on this scheme, we compute meas- ures of cyclone frequency and intensity. Relative cyclone frequency is obtained by dividing the number of six-hourly periods affected by a cyclone by the total number of six-hourly intervals in the analysis period. Furthermore, SLP minima of the identified cyclones are used as a measure of their intensity. To take the geographical SLP variability into account the cyclone field SLP minima are normalised by subtracting the local SLP climatology. For each 20CR ensemble member, monthly anomalies of these parameters are derived by subtracting the climatological mean for June months in 1871–2008 from the mean for June 1944. As a result, we obtained the storminess with respect to cyclone frequency (and cyclone intensity, respectively). Cyclones are typically associated with enhanced wind speeds along their fron- tal zones, which were a main concern of the D-Day forecasting team. Hence, to quantify the storminess of June 1944 with respect to wind, anomalous maximum near-surface wind speeds (i.e. wind at the pressure level 0.995 times the surface pres- sure, corresponding to a height of about Figure 1. Synoptic situation on (a) 3 June 1944 (0700 UTC), (b) 4 June (0700 UTC), (c) 5 June 30–40m above ground) are computed for (0700 UTC) and (d) 6 June (0700 UTC) according to historical weather charts by the Met Office. 177 weather type (10% for June 1944), which is typically associated with warm and dry weather conditions in summer, was signifi- cantly underrepresented with respect to the 1861–1943 mean value of 21%. Moreover, the sharp increase in the frequency of west- erly weather types that takes place in mid- June in the climatological mean (Barry and Chorley, 2003) apparently started approxi- mately half a month earlier in June 1944. Thus, this early shift in the prevailing flow regime might have led to disturbed weather The D-Day landing of June 1944 conditions over the British Isles. Figure 4 shows anomalies in relative cyclone frequency for June 1944 in the NA and European sectors, based on the 20CR ensemble data set. Although there is a region of anomalously high relative cyclone frequency (greater than 30%) centred Figure 2. Synoptic situation on (a) 3 June 1944 (0600 UTC), (b) 4 June (0600 UTC), (c) 5 June (0600 UTC) between northern Great Britain and south- and (d) 6 June (0600 UTC) according to the 20CR ensemble mean: SLP (in hPa; contour lines) and ern Scandinavia, the English Channel region near-surface potential temperature (in K; colour shade). Green dots indicate the locations of the was not directly affected by anomalously surface pressure and SLP observations that were assimilated into the 20CR at the respective time frequent cyclones. In a region over the NA

Weather – July 2014, Vol. 69, No. 7 69, No. Vol. – July 2014, Weather steps – based on version 2 of the International Surface Pressure Databank (see Compo et al., 2011). west of the British Isles, the relative cyclone The 20CR Great Britain grid cells (12°W–4°E, 48°N–60°N) are framed by a red box. The location of frequency was even reduced by approxi- the Normandy landing beaches is marked with a red star. mately 10–30%. The cyclones hitting the situated northwest of Scotland on 4 June In summary, (i) the synoptic situation British Isles in June 1944 were more intense was slowly moving eastward on 5 June in early June 1944 was characterised by than usual, with regard to the cyclones’ SLP and the cyclone centre was located slightly a succession of cyclones passing north of minima (not shown). Furthermore, during to the northeast of Scotland at 0700 UTC the English Channel, heavily influencing the June 1944 the maximum near-surface wind (Figure 1(c)). At the same time, the cyclone weather in that region, and (ii) the 20CR speed was anomalously high in many parts that had formed over Newfoundland on agrees well with historical weather charts of Great Britain, even though again not 4 June was further deepening on 5 June provided by the Met Office. directly in the region of the English Channel and moving eastward. However, its east- (not shown). ward movement was providing enough The storminess of June 1944 However, considering the climatologi- time of fair weather to launch the landings cal mean relative cyclone frequency of Montgomery (1948), Ambrose (1994) and on 6 June. Again, the 20CR captures well approximately 15–25% (Figure 4) in the Fleming (2004) stressed the uncommonly the main features of the synoptic situation aforementioned region of increased cyclone unsettled and stormy weather conditions (Figure 2(c)). frequency between northern Great Britain during the D-Day landing of June 1944 and The decision to launch the assault on 6 and southern Scandinavia, the pronounced during the preceding days; although, nei- June was made in the morning of 5 June. For positive anomaly of about 30% denotes that ther the study of Montgomery nor that of the German forces watching their defenses, this region was affected by cyclones during Ambrose is strictly speaking a meteorologi- it was difficult to foresee that this was the more than half of June 1944. Because the cal study. The aim of this section is to assess, moment the Allied armies had planned to surface winds associated with extratropi- on the basis of the 20CR ensemble data invade. Moreover, due to the high wind cal cyclones are highest along the frontal set, whether June 1944 was anomalously speeds the reconnaissance aircraft of the zones, the cyclone fields (defined in this stormy compared with the climatology. German air force stayed on the ground and study by the outermost closed isobar; see According to the classification system of the German forces thus failed to notice the section ‘Surface cyclones and associated high British Isles weather types by Lamb (1972), immediate danger of the situation. wind speeds in the 20CR Ensemble’) do not which distinguishes 27 different circulation On the morning of 6 June, the deep necessarily cover all areas with high winds. patterns based on surface synoptic weather cyclone north of Scotland, influencing the Bengtsson et al. (2009), who studied the charts, June 1944 was dominated by cyclonic weather in the English Channel region, spatial relation between cyclone and wind weather and westerly flow. This flow type is slowly moved further eastward and became fields for intense extratropical Northern typically associated with cool and cloudy less deep (Figures 1(d) and 2(d)). Compared Hemisphere winter cyclones at different life- conditions in summer, with changeable with the preceding 2 days, the near-surface cycle stages based on the European Centre weather and variable wind directions due to wind was weaker (Figure 3(a)) and the air- for Medium-Range Weather Forecasts 40 the passage of cyclones (Barry and Chorley, borne forces benefited from the improved Year Reanalysis, found that at the state of 2003). Compared with the monthly mean near-surface visibility along the Normandy maximum cyclone intensity the highest percentage frequencies of cyclonic and coast (Figure 3(b)). The 20CR indicates a wind speeds are, in general, found behind westerly conditions for June in 1861–1943, drying of the atmosphere over the Channel and to the right of the cyclone centre in the which are 13% and 15%, respectively, the on 6 June (Figure 3(d)). However, the 20CR area after the occluding cold front. Thus, the values for June 1944 were clearly higher further indicates the possibility of low increased number of cyclones that passed (23 and 27%)1. Also the purely anticyclonic clouds over the Normandy landing beaches to the north of the English Channel region (Figure 3(c)). Indeed, historical photographs 1 Numbers based on the complete series of in June 1944 very likely also affected the show clear evidence of overcast conditions Lamb’s weather-type classification, which is weather in the Normandy landing area and while the Allied forces were crossing the available online at http://www.cru.uea.ac.uk/cru/ required the forecasters’ constant alertness 178 Channel. data/lwt/ (last access: October 2013). and vigilance, especially because the exact The D-Day landing of June 1944 The D-Day

Figure 4. The 20CR ensemble-mean anomalous relative cyclone frequency for June 1944 (colour

shade). Anomalies are calculated by subtract- Weather – July 2014, Vol. 69, No. 7 ing the climatological mean for June months in 1871–2008 (dashed contours), for each 20CR ensemble member and grid cell. The 20CR Great Britain grid cells (12°W–4°E, 48°N–60°N) are framed by a red box. The location of the Normandy landing beaches is marked with a red star. Black grid cells show regions where the 20CR orography exceeds 1500 m asl.

in historical records. This information should be taken with a grain of salt, because public perception is often strongly influenced by single extreme weather events. However, it is well documented that the weather in the first week of June was very unsettled, with a ‘train’ of cyclones crossing the Atlantic. As a rough estimate, one third of June 1944 was characterised by well-founded influence of cyclonic activity over the northeast Atlantic, Great Britain and northern France. When using the 20CR, the current con- troversial discussion concerning the quality of the 20CR, in particular prior to 1950, is relevant. For example, Krueger et al. (2013; 2014) found, concerning storminess over Figure 3. (a) The 20CR ensemble distribution (box plots give minimum, lower quartile, median, the northeast Atlantic, considerable disa- upper quartile, and maximum; the red area denotes the ensemble mean) of six-hourly near-surface greement between the 20CR and observa- −1 wind speed (in ms ; lowest model level) for the 20CR grid cell closest to the Normandy landing tions prior to 1950. In contrast to Wang et al. beaches and the period 1–9 June 1944 (the day of the Normandy landing is marked in blue). The (2014), who showed that storminess in the dashed curve gives the corresponding temporal evolution of the 20CR ensemble-mean near-sur- NA and European sectors derived from the face wind direction (in degrees from north). (b) Near-surface relative humidity (RH; in %), (c) RH at 20CR agrees well with observations back to 850hPa (in %), and (d) atmospheric water vapour content (in kgm−2). about 1893. For analysing cyclones in the 20CR, the individual ensemble members should be pathways of the cyclones were not known Normandy. By assimilating only surface- used rather than the ensemble-mean analy- at that time. pressure observations and using monthly ses (Wang et al., 2012). Thus, we applied the Figure 5 shows that both cyclone fre- SST and sea-ice distributions as boundary objective cyclone identification and track- quency and maximum near-surface wind conditions, the 20CR contains a wide range ing scheme of Wernli and Schwierz (2006) speed over the British Isles were anoma- of atmospheric fields from the surface up to each of the 20CR ensemble members, lously high during June 1944 compared to the tropopause, some of which provided which further allowed us to produce an with the climatological average for all June valuable insight into the development of uncertainty estimation of our results. months in 1871–2008: regarding the 20CR the weather in the D-Day landing area. The main results of this study can be sum- ensemble mean, the cyclone frequency In this study, the 20CR was used to marised as follows: (maximum near-surface wind speed) in address two distinct aspects of the D-Day June 1944 was the seventh (ninth) highest weather. (i) How well is the synoptic and • The 20CR maps agree well with the in the 138-year period 1871–2008. mesoscale evolution at the beginning of historical weather charts by the Met June 1944 captured by the 20CR compared Office (Figures 1 and 2). The synoptic with historical weather analyses? (ii) How situation was characterised by a suc- Conclusions exceptional was the storminess of June cession of extratropical cyclones pass- The 20CR has proved to be a very effective 1944, expressed by cyclone frequency and ing north of the English Channel and ‘time machine’ to explore the weather in near-surface wind speed, compared with influencing the weather in that region, early June 1944, when World War II reached the 20CR climatology? Indeed, this month only permitting the landing in a short a turning point with the Allied invasion of is described as an unusually stormy month inter-frontal clearing. Slight differences 179 from Franklin’s Kite to El Niño. Wiley & Sons: New York, NY, 252 pp. Douglas CKM. 1952. Forecasting for the D-Day landings. Meteorol. Mag. 81: 161–171. ECMWF. 2014. Revisiting the meteorology of the D-Day period, using a modern anal- ysis and forecasting system. http://www. ecmwf.int/en/research/projects/era-clim/ d-day-analyses (accessed 13 June 2014). Fleming JR. 2004. Sverre Petterssen, the Bergen School, and the forecasts for D-Day. Proceedings of the Internation al Commission on the History of Meteorology, The D-Day landing of June 1944 Waterville. pp 75–83. Krueger O, Schenk F, Feser F et al. 2013. Inconsistencies between long-term trends in storminess derived from the 20CR reanalysis and observations. J. Clim. 26: 868–874. Krueger O, Feser F, Bärring L et al. 2014. Comment on ‘Trends and low frequency variability of extra-tropical cycl one activ- ity in the ensemble of twentieth century Figure 5. Spatial average for the 20CR Great Britain grid cells (12°W–4°E, 48°N–60°N) of anoma- reanalysis’ by Xiaolan L. Wang, Y. Feng, lous (a) relative cyclone frequency and (b) maximum near-surface wind speed (in ms−1) for June G. P. Compo, V. R. Swail, F. W. Zwiers, R. months in 1871–2008. Box plots indicate minimum, lower quartile, median, upper quartile, and J. Allan, and P. D. Sardeshmukh. Climate Weather – July 2014, Vol. 69, No. 7 69, No. Vol. – July 2014, Weather maximum of the 20CR ensemble distribution and the blue and red areas denote the ensemble Dynamics, 2012. Clim. Dyn. 42: 1127–1128. doi:10.1007/s00382-013-1814-9 mean. June 1944 is marked with a green circle and a dotted line. Lamb HH. 1972. British Isles Weather types and a register of daily sequence of exist between 20CR and historical maps to an interesting overview of the role of circulation patterns, 1861–1971. Geophys. with regard to position and intensity of the Met Office in preparation for D-Day Mem. 116: 85 pp. NA cyclones, but the hazardous weather (http://www.metoffice.gov.uk/archive/ Montgomery B. 1948. Normandy to the on 5 June 1944 and the clearing on 6 d-day-role-of-the-met-office). Support for Baltic: Twenty-First Army Group. Houghton June 1944 is well captured by the 20CR the Twentieth Century Reanalysis Project Mifflin: Boston, MA. (Figure 3). data set is provided by the US Department Petterssen S. 2001. In Weathering the Storm: • The climatological analysis of cyclone of Energy, Office of Science Innovative and Sverre Petterssen, the D-Day Forecast, and the activity based on the 20CR reveals for Novel Computational Impact on Theory Rise of Modern Meteorolo gy. Fleming JR (ed.). American Meteorological Society: Boston, June 1944 patches of increased cyclone and Experiment (DOE INCITE) programme, MA, 329 pp. frequency and wind maxima, as well as and Office of Biological and Environmental Rayner NA, Parker DE, Horton EB et al. reduced cyclone-field SLP minima, over Research (BER), and by the National Oceanic 2003. Global analyses of sea surface and northeast of Great Britain (Figure 4). and Atmospheric Administration Climate temperature, sea ice, and night marine Compared with the 20CR climatology, Program Office. Finally, we would like to a ir temperature since the late nineteenth century. J. Geophys. Res. 108: 4407. June 1944 ranks seventh with respect to thank two anonymous reviewers who doi:10.1029/2002JD002670 cyclone frequency over the British Isles, provided valuable comments and hence Shaw RH, Innes W. 1984. Some and ranks ninth concerning maximum contributed in substantially improving this Meteorological Aspects of the D-Day near-surface wind speed (Figure 5). article. Invasion of Europe, 6 June 1944. American Meteorolo gical Society: Boston, MA. Our study also illustrates that, despite the Stagg JM. 1971. Forecast for Overlord, June controlling influence humans try to exert, References 6, 1944. Littlehampton Book Services Ltd: world history is in the first instance often Addlestone, UK. Ackerman ST, Knox JA. 2 0 06. Meteorology: a game of dice rather than strategy. In the Understanding the Atmosphere, 2nd Edition. Wang XL, Feng Y, Compo GP et al. 2012. case of the D-Day forecasts, it was the role Brooks/Cole-Thomson Learning: Pacific Trends and low frequency variability of meteorology to at least understand the Grove, CA, 512 pp. of extra-tropical cyclone activity in th e ensemble of twentieth century reanalysis. rules of this game well enough to predict Ambrose S. 1994. D-Day J une 6, 1944: The Clim. Dyn. 40: 2775–2800. doi:10.1007/ Climactic Battle of World War II. Simon and its outcome some days in advance, that is, s00382-012-1450-9 Schuster: New York, NY. to give the military commanders the best Wang XL, Feng Y, Compo GP et al. Atkinson R. 2013. The road to D-Day. possible strategic starting point. Hence, to 2014. Is the storminess in the Twentieth http://www.foreignaffairs.com/arti- Century Reanalysis really inconsistent with put it in the words of the British meteor- cles/139455/rick-atkinson/the-road-to-d- observations? A reply to the comment ologist Reginald Sutcliffe: Meteorologically, day (accessed 10 June 2014). the occasion was remarkable not only for the by Krueger et al. (2014). Clim. Dyn. 42: Barry RG, Chorley RJ. 2003. Atmosphere, 1113–1125. doi:10.1007/s00382-013-1828-3 degree of dependence on the weather but also Weather, and Climate. Routledge Chapman Wernli H, Schwierz C. 2006. Surface & Hall: London, 472 pp. the planned dependence on weather forecasts cyclones in the ERA-40 dataset (Shaw and Innes, 1984). Bengtsson L, Hodges KI, Keenlyside N. (1958–2001). Part I: novel identification 2009. Will extratropical storms intensify in method and global climatology. J. Atmos. a warmer climate? J. Clim. 22: 2276–2301. Sci. 63: 2486–2507. Acknowledgements Compo GP, Whitaker JS, Sardeshmukh PD et al. 2011. The Twentieth Century We would like to thank Mark Beswick and Correspondence to: Christoph Welker Kate Willett from the Met Office for pro- Reanalysis project. Q. J. R. Meteorol. Soc. 4: 1–28. [email protected] viding historical weather charts (shown Cox JD. 2002. Storm Watchers: The © 2014 Royal Meteorological Society in Figure 1), the Meteorological Magazine Turbulent History of Weather Prediction 180 article by Douglas (1952) and for pointing doi:10.1002/wea.2339