• Baldauf, M., , Siefert A. , , Förstner J. , , Majewski D. , , Raschendorfer M. , , and Reinhardt T. , 2011: Operational convective-scale numerical weather prediction with the COSMO model: Description and sensitivities. Mon. Wea. Rev., 139, 38873905, doi:10.1175/MWR-D-10-05013.1.

    • Search Google Scholar
    • Export Citation
  • Barrett, A. I., , Gray S. L. , , Kirshbaum D. J. , , Roberts N. M. , , Schultz D. M. , , and Fairman J. G. Jr., 2014: Synoptic versus orographic control on stationary convective banding. Quart. J. Roy. Meteor. Soc., 141, 11011113, doi:10.1002/qj.2409.

    • Search Google Scholar
    • Export Citation
  • Bell, H., , and Tobin G. , 2007: Efficient and effective? The 100-year flood in the communication and perception of flood risk. Environ. Hazards, 7, 302311, doi:10.1016/j.envhaz.2007.08.004.

    • Search Google Scholar
    • Export Citation
  • Bowler, N. E., , Arribas A. , , Mylne K. R. , , Robertson K. B. , , and Beare S. E. , 2008: The MOGREPS short-range ensemble prediction system. Quart. J. Roy. Meteor. Soc., 134, 703722, doi:10.1002/qj.234.

    • Search Google Scholar
    • Export Citation
  • Bowler, N. E., ,Arribas A. , , Beare S. E. , , Mylne K. R. , , and Shutts G. J. , 2009: The local ETKF and SKEB: Upgrades to the MOGREPS short-range ensemble prediction system. Quart. J. Roy. Meteor. Soc., 135, 767776, doi:10.1002/qj.394.

    • Search Google Scholar
    • Export Citation
  • Buizza, R., , Houtekamer P. L. , , Toth Z. , , Pellerin G. , , Wei M. Z. , , and Zhu Y. J. , 2005: A comparison of the ECMWF, MSC, and NCEP global ensemble prediction systems. Mon. Wea. Rev., 133, 10761097, doi:10.1175/MWR2905.1.

    • Search Google Scholar
    • Export Citation
  • Burningham, K., , Fielding J. , , and Thrush D. , 2008: It’ll never happen to me: Understanding public awareness of local flood risk. Disasters, 32, 216238, doi:10.1111/j.1467-7717.2007.01036.x.

    • Search Google Scholar
    • Export Citation
  • CEH, 2011: UK Hydrological Bulletin January–July 2011. Centre for Ecology and Hydrology, accessed 28 March 2016. [Available online at http://nrfa.ceh.ac.uk/node/110/.]

  • Clark, A. J., , Gallus W. A. Jr., , Xue M. , , and Kong F. , 2009: A comparison of precipitation forecast skill between small convection-permitting and large convection-parameterizing ensembles. Wea. Forecasting, 24, 11211140, doi:10.1175/2009WAF2222222.1.

    • Search Google Scholar
    • Export Citation
  • Clark, A. J., and et al. , 2011: Probabilistic precipitation forecast skill as a function of ensemble size and spatial scale in a convection-allowing ensemble. Mon. Wea. Rev., 139, 14101418, doi:10.1175/2010MWR3624.1.

    • Search Google Scholar
    • Export Citation
  • Clark, A. J., , Weiss S. J. , , and Kain J. S. , 2012: An overview of the 2010 hazardous weather testbed experimental forecast program spring experiment. Bull. Amer. Meteor. Soc., 93, 5574, doi:10.1175/BAMS-D-11-00040.1.

    • Search Google Scholar
    • Export Citation
  • Clark, M. R., 2011: An observational study of the exceptional ‘Ottery St Mary’ thunderstorm of 30 October 2008. Meteor. Appl., 18, 137154, doi:10.1002/met.187.

    • Search Google Scholar
    • Export Citation
  • Davies, T., , Cullen M. J. P. , , Malcolm A. J. , , Mawson M. H. , , Staniforth A. , , White A. A. , , and Wood N. , 2005: A new dynamical core for the Met Office’s global and regional modelling of the atmosphere. Quart. J. Roy. Meteor. Soc., 131, 17591782, doi:10.1256/qj.04.101.

    • Search Google Scholar
    • Export Citation
  • Dolan, J. G., , and Iadarola S. , 2008: Risk communication formats for low probability events: An exploratory study of patient references. BMC Med. Inf. Decis. Making, 8, 14, doi:10.1186/1472-6947-8-14.

    • Search Google Scholar
    • Export Citation
  • Done, J., , Davis C. A. , , and Weisman M. L. , 2004: The next generation of NWP: Explicit forecasts of convection using the Weather Research and Forecasting (WRF). Atmos. Sci. Lett., 5, 110117, doi:10.1002/asl.72.

    • Search Google Scholar
    • Export Citation
  • Duan, Y., and et al. , 2012: An overview of Beijing 2008 Olympics Research and Development Project (B08RDP). Bull. Amer. Meteor. Soc., 93, 381403, doi:10.1175/BAMS-D-11-00115.1.

    • Search Google Scholar
    • Export Citation
  • Edwards, A., , Elwyn G. , , and Mulley A. , 2002: Explaining risks: Turning numerical data into meaningful pictures. BMJ, 324, 827830, doi:10.1136/bmj.324.7341.827.

    • Search Google Scholar
    • Export Citation
  • Edwards, J., , and Slingo A. , 1996: Studies with a flexible new radiation code. Part I: Choosing a configuration for a large-scale model. Quart. J. Roy. Meteor. Soc., 122, 689719, doi:10.1002/qj.49712253107.

    • Search Google Scholar
    • Export Citation
  • Essery, R. L. H., , Best M. J. , , Betts R. A. , , Cox P. M. , , and Taylor C. M. , 2003: Explicit representation of subgrid heterogeneity in a GCM land surface scheme. J. Hydrometeor., 4, 530543, doi:10.1175/1525-7541(2003)004<0530:EROSHI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Falconer, R. H., , Cobby D. , , Smyth P. , , Astle G. , , Dent J. , , and Golding B. , 2009: Pluvial flooding: New approaches in flood warning, mapping and risk management. J. Flood Risk Manage., 2, 198208, doi:10.1111/j.1753-318X.2009.01034.x.

    • Search Google Scholar
    • Export Citation
  • Faulkner, H., , Parker D. , , Green C. , , and Beven K. , 2007: Developing a translational discourse to communicate uncertainty in flood risk between science and the practitioner. Ambio, 36, 692704, doi:10.1579/0044-7447(2007)36[692:DATDTC]2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • FFC, 2015: About us. Flood Forecasting Centre, accessed 28 March 2016. [Available online at http://www.ffc-environment-agency.metoffice.gov.uk/about/.]

  • Gebhardt, C., , Theis S. E. , , Paulat M. , , and Ben Bouallegue Z. , 2011: Uncertainties in COSMO-DE precipitation forecasts introduced by model perturbations and variation of lateral boundaries. Atmos. Res., 100, 168177, doi:10.1016/j.atmosres.2010.12.008.

    • Search Google Scholar
    • Export Citation
  • Germann, U., , and Zawadzki I. , 2004: Scale dependence of the predictability of precipitation from continental radar images. Part II: Probability forecasts. J. Appl. Meteor., 43, 7489, doi:10.1175/1520-0450(2004)043<0074:SDOTPO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Gigerenzer, G., , Hertwig R. , , van den Broek E. , , Fasolo B. , , and Katsikopoulos K. V. , 2005: “A 30% chance of rain tomorrow”: How does the public understand probabilistic weather forecasts? Risk Anal., 25, 623629, doi:10.1111/j.1539-6924.2005.00608.x.

    • Search Google Scholar
    • Export Citation
  • Gill, J., 2008: Communicating forecast uncertainty for service providers. WMO Bull., 57 (4), 237243.

  • Golding, B. W., , Clark P. , , and May B. , 2005: The Boscastle flood: Meteorological analysis of the conditions leading to flooding on 16 August 2004. Weather, 60, 230235, doi:10.1256/wea.71.05.

    • Search Google Scholar
    • Export Citation
  • Golding, B. W., and et al. , 2014: Forecasting capabilities for the London 2012 Olympics. Bull. Amer. Meteor. Soc., 95, 883896, doi:10.1175/BAMS-D-13-00102.1.

    • Search Google Scholar
    • Export Citation
  • Goldman, R., , Parker D. , , Eaton C. , , Borkan J. , , Gramling R. , , Cover R. , , and Ahern D. , 2006: Patients’ perceptions of cholesterol, cardiovascular disease risk, and risk communication strategies. Ann. Fam. Med., 4, 205212, doi:10.1370/afm.534.

    • Search Google Scholar
    • Export Citation
  • Gregory, D., , and Rowntree P. R. , 1990: A mass flux convection scheme with representation of cloud ensemble characteristics and stability-dependent closure. Mon. Wea. Rev., 118, 14831506, doi:10.1175/1520-0493(1990)118<1483:AMFCSW>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Guha-Sapir, D., , Vos F. , , Below R. , , and Ponserre S. , 2012: Annual disaster statistical review 2011: The numbers and trends. Centre for Research on the Epidemiology of Disasters, 42 pp. [Available online at http://cred.be/sites/default/files/2012.07.05.ADSR_2011.pdf.]

  • Hand, W. H., , Fox N. I. , , and Collier C. G. , 2004: A study of twentieth-century extreme rainfall events in the United Kingdom with implications for forecasting. Meteor. Appl., 11, 1531, doi:10.1017/S1350482703001117.

    • Search Google Scholar
    • Export Citation
  • Hanley, K. E., , Kirshbaum D. J. , , Belcher S. E. , , Roberts N. M. , , and Leoncini G. , 2011: Ensemble predictability of an isolated mountain thunderstorm in a high-resolution model. Quart. J. Roy. Meteor. Soc., 137, 21242137, doi:10.1002/qj.877.

    • Search Google Scholar
    • Export Citation
  • Hanley, K. E., , Kirshbaum D. J. , , Roberts N. M. , , and Leoncini G. , 2013: Sensitivities of a squall line over central Europe in a convective-scale ensemble. Mon. Wea. Rev., 141, 112133, doi:10.1175/MWR-D-12-00013.1.

    • Search Google Scholar
    • Export Citation
  • Harrison, D., , Georgiou S. , , Gaussiat N. , , and Curtiss A. , 2014: Long-term diagnostics of precipitation estimates and the development of radar hardware monitoring within a radar product data quality management system. Hydrol. Sci. J., 59, 12771292, doi:10.1080/02626667.2013.841316.

    • Search Google Scholar
    • Export Citation
  • Hirahara, Y., , Ishida J. , , and Ishimizu T. , 2012: Trial operation of the local forecast model at JMA. Development of and studies with regional and smaller-scale atmospheric models, regional ensemble, monthly and seasonal forecasting, WMO/WCRP/WGNE Blue Book 2011, World Climate Research Programme, 11–12. [Available online at http://www.wcrp-climate.org/WGNE/BlueBook/2011/chapters/BB_11_S5.pdf.]

  • Hohenegger, C., , Walser A. , , Langhans W. , , and Schaer C. , 2008: Cloud-resolving ensemble simulations of the August 2005 Alpine flood. Quart. J. Roy. Meteor. Soc., 134, 889904, doi:10.1002/qj.252.

    • Search Google Scholar
    • Export Citation
  • Ikeda, K., , Steiner M. , , Pinto J. , , and Alexander C. , 2013: Evaluation of cold-season precipitation forecasts generated by the hourly updating high-resolution rapid refresh model. Wea. Forecasting, 28, 921939, doi:10.1175/WAF-D-12-00085.1.

    • Search Google Scholar
    • Export Citation
  • Joslyn, S., , and Savelli S. , 2010: Communicating forecast uncertainty: Public perception of weather forecast uncertainty. Meteor. Appl., 17, 180195, doi:10.1002/met.190.

    • Search Google Scholar
    • Export Citation
  • Joslyn, S., , Pak K. , , Jones D. , , Pyles J. , , and Hunt E. , 2007: The effect of probabilistic information on threshold forecasts. Wea. Forecasting, 22, 804812, doi:10.1175/WAF1020.1.

    • Search Google Scholar
    • Export Citation
  • Joslyn, S., , Nadav-Greenberg L. , , and Nichols R. M. , 2009: Probability of precipitation: Assessment and enhancement of end-user understanding. Bull. Amer. Meteor. Soc., 90, 185193, doi:10.1175/2008BAMS2509.1.

    • Search Google Scholar
    • Export Citation
  • Kitchen, M., , Brown R. , , and Davies A. G. , 1994: Real-time correction of weather radar data for the effects of bright band, range and orographic growth in widespread precipitation. Quart. J. Roy. Meteor. Soc., 120, 12311254, doi:10.1002/qj.49712051906.

    • Search Google Scholar
    • Export Citation
  • Kloprogge, P., , Van der Sluijs J. , , and Wardekker A. , 2007: Uncertainty communication: Issues and good practice. Rep. NWS-E-2007-199, Utrecht University, 60 pp. [Available online at http://www.nusap.net/downloads/reports/uncertainty_communication.pdf.]

  • Knapp, P., , Raynor D. , , and Berry D. , 2004: Comparison of two methods of presenting risk information to patients about the side effects of medicines. Qual. Saf. Health Care, 13, 176180, doi:10.1136/qshc.2003.009076.

    • Search Google Scholar
    • Export Citation
  • Kong, F. Y., , Droegemeier K. K. , , and Hickmon N. L. , 2007: Multiresolution ensemble forecasts of an observed tornadic thunderstorm system. Part II: Storm-scale experiments. Mon. Wea. Rev., 135, 759782, doi:10.1175/MWR3323.1.

    • Search Google Scholar
    • Export Citation
  • Kunii, M., and et al. , 2011: Verifications and intercomparisons of mesoscale ensemble prediction systems in B08RDP. Tellus, 63A, 531549, doi:10.1111/j.1600-0870.2011.00512.x.

    • Search Google Scholar
    • Export Citation
  • Kurz-Milcke, E., , Gigerenzer G. , , and Martignn L. , 2008: Transparency in risk communication: Graphical and analogue tools. Ann. N. Y. Acad. Sci., 1128, 1828, doi:10.1196/annals.1399.004.

    • Search Google Scholar
    • Export Citation
  • Lean, H. W., , Clark P. A. , , Dixon M. , , Roberts N. M. , , Fitch A. , , Forbes R. , , and Halliwell C. , 2008: Characteristics of high-resolution versions of the Met Office Unified Model for forecasting convection over the United Kingdom. Mon. Wea. Rev., 136, 34083424, doi:10.1175/2008MWR2332.1.

    • Search Google Scholar
    • Export Citation
  • Lean, H. W., , Roberts N. M. , , Clark P. A. , , and Morcrette C. , 2009: The surprising role of orography in the initiation of an isolated thunderstorm in southern England. Mon. Wea. Rev., 137, 30263046, doi:10.1175/2009MWR2743.1.

    • Search Google Scholar
    • Export Citation
  • Leoncini, G., , Plant R. S. , , Gray S. L. , , and Clark P. A. , 2010: Perturbation growth at the convective scale for CSIP IOP18. Quart. J. Roy. Meteor. Soc., 136, 653670, doi:10.1002/qj.587.

    • Search Google Scholar
    • Export Citation
  • Lock, A. P., 2001: The numerical representation of entrainment in parameterizations of boundary layer turbulent mixing. Mon. Wea. Rev., 129, 11481163, doi:10.1175/1520-0493(2001)129<1148:TNROEI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Lock, A. P., , Brown A. R. , , Bush M. R. , , Martin G. M. , , and Smith R. N. B. , 2000: A new boundary layer mixing scheme. Part I: Scheme description and single-column model tests. Mon. Wea. Rev., 128, 31873199, doi:10.1175/1520-0493(2000)128<3187:ANBLMS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Lorenc, A. C., and et al. , 2000: The Met. Office global three-dimensional variational data assimilation scheme. Quart. J. Roy. Meteor. Soc., 126, 29913012, doi:10.1002/qj.49712657002.

    • Search Google Scholar
    • Export Citation
  • Molteni, F., , Buizza R. , , Palmer T. N. , , and Petroliagis T. , 1996: The new ECMWF ensemble prediction system: Methodology and validation. Quart. J. Roy. Meteor. Soc., 122, 73119, doi:10.1002/qj.49712252905.

    • Search Google Scholar
    • Export Citation
  • Moore, R. J., , Cole S. J. , , Dunn S. , , Ghimire S. , , Golding B. W. , , Pierce C. E. , , Roberts N. M. , , and Speight L. , 2015: Surface water flood forecasting for urban communities. CREW Rep. CRW2012_03, 33 pp. [Available online at http://www.crew.ac.uk/publications/surface-water-flood-forecasting-urban-communities.]

  • Morss, R. E., , Demuth J. L. , , and Lazo J. K. , 2008: Communicating uncertainty in weather forecasts: A survey of the U.S. public. Wea. Forecasting, 23, 974991, doi:10.1175/2008WAF2007088.1.

    • Search Google Scholar
    • Export Citation
  • Murphy, A. H., , Lichtenstein S. , , Fischhoff B. , , and Winkler R. L. , 1980: Misinterpretations of precipitation probability forecasts. Bull. Amer. Meteor. Soc., 61, 695701, doi:10.1175/1520-0477(1980)061<0695:MOPPF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Paling, J., 2003: Strategies to help patients understand risks. BMJ, 327, 745, doi:10.1136/bmj.327.7417.745.

  • Peachey, J., , Schultz D. M. , , Morss R. , , Roebber P. , , and Wood R. , 2013: How forecasts expressing uncertainty are perceived by UK students. Weather, 68, 176181, doi:10.1002/wea.2094.

    • Search Google Scholar
    • Export Citation
  • Peralta, C., , Ben Bouallègue Z. , , Theis S. E. , , Gebhardt C. , , and Buchhold M. , 2012: Accounting for initial condition uncertainties in COSMO-DE-EPS. J. Geophys. Res., 117, D07108, doi:10.1029/2011JD016581.

    • Search Google Scholar
    • Export Citation
  • Peters, P., , Bostrom A. , , and Cutter S. , 2008: Perspectives on visualizing uncertainty in natural hazards. Risk Assessment, Modeling and Decision Support, A. Bostrom, S. French, and S. Gottlieb, Eds., Risk, Governance and Society, Vol. 14, Springer, 295–318.

  • Pitt, M., 2009: Learning lessons from the 2007 floods. UK Cabinet Office Rep., 462 pp.

  • Price, D., and et al. , 2012: Operational use of a grid-based model for flood forecasting. Proc. Inst. Civ. Eng.: Water Manage., 165, 6577, doi:10.1680/wama.2012.165.2.65.

    • Search Google Scholar
    • Export Citation
  • Price, M., , Cameron R. , , and Butow P. , 2007: Communicating risk information: The influence of graphical display format on quantitative information perception—Accuracy, comprehension and preferences. Patient Educ. Couns., 69, 121128, doi:10.1016/j.pec.2007.08.006.

    • Search Google Scholar
    • Export Citation
  • Roberts, N. M., 2000: The relationship between water vapour imagery and thunderstorms. Met Office Forecasting Research Tech. Rep. 110, 40 pp.

  • Roberts, N. M., , and Lean H. W. , 2008: Scale-selective verification of rainfall accumulations from high-resolution forecasts of convective events. Mon. Wea. Rev., 136, 7897, doi:10.1175/2007MWR2123.1.

    • Search Google Scholar
    • Export Citation
  • Roulston, M. S., , and Kaplan T. R. , 2009: A laboratory-based study of understanding of uncertainty in 5-day site-specific temperature forecasts. Meteor. Appl., 16, 237244, doi:10.1002/met.113.

    • Search Google Scholar
    • Export Citation
  • Roulston, M. S., , Bolton G. E. , , Kleit A. N. , , and Sears-Collins A. L. , 2006: A laboratory study of the benefits of including uncertainty information in weather forecasts. Wea. Forecasting, 21, 116122, doi:10.1175/WAF887.1.

    • Search Google Scholar
    • Export Citation
  • Scheuren, J.-M., , le Polain de Waroux O. , , Below R. , , Guha-Sapir D. , , and Ponserre S. , 2008: Annual disaster statistical review: The numbers and trends 2007. Center for Research on the Epidemiology of Disasters Rep., 64 pp. [Available online at http://www.unisdr.org/files/2796_CREDAnnualStatisticalReview2007.pdf.]

  • Schirillo, J., , and Stone E. , 2005: The greater ability of graphical versus numerical displays to increase risk avoidance involves a common mechanism. Risk Anal., 25, 555566, doi:10.1111/j.1539-6924.2005.00624.x.

    • Search Google Scholar
    • Export Citation
  • Seity, Y., , Brousseau P. , , Malardel S. , , Hello G. , , Bénard P. , , Bouttier F. , , Lac C. , , and Masson V. , 2011: The AROME-France Convective-Scale Operational Model. Mon. Wea. Rev., 139, 976991, doi:10.1175/2010MWR3425.1.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., , Brooks H. E. , , Du J. , , Tracton M. S. , , and Rogers E. , 1999: Using ensembles for short-range forecasting. Mon. Wea. Rev., 127, 433446, doi:10.1175/1520-0493(1999)127<0433:UEFSRF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Stephens, E., , Mylne K. , , and Spiegelhalter D. , 2011: Using an online game to evaluate effective methods of communicating ensemble model output to different audiences. 2011 Fall Meeting, San Francisco, CA, Amer. Geophys. Union, Abstract ED33B-0776.

  • Steppeler, J., , Doms G. , , Schattler U. , , Bitzer H. W. , , Gassmann A. , , Damrath U. , , and Gregoric G. , 2003: Meso-gamma scale forecasts using the nonhydrostatic model LM. Meteor. Atmos. Phys., 82, 7596, doi:10.1007/s00703-001-0592-9.

    • Search Google Scholar
    • Export Citation
  • Stone, E., , Yates J. , , and Parker A. , 1997: Effects of numerical and graphical displays on professed risk-taking behaviour. J. Exp. Psychol., 3 (4), 243256.

    • Search Google Scholar
    • Export Citation
  • Tang, Y., , Lean H. W. , , and Bornemann J. , 2013: The benefits of the Met Office variable resolution NWP model for forecasting convection. Meteor. Appl., 20, 417426, doi:10.1002/met.1300.

    • Search Google Scholar
    • Export Citation
  • Tapsell, S., , Burton R. , , Oakes S. , , and Parker D. , 2005: The social performance of flood warning communications technologies. R&D Tech. Summary W5C-016/TS, Environment Agency, 3 pp. [Available online at https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/290685/scho0505bjdz-e-e.pdf.]

  • Tennant, W., 2015: Improving initial condition perturbations for MOGREPS-UK. Quart. J. Roy. Meteor. Soc., 141, 23242336, doi:10.1002/qj.2524.

    • Search Google Scholar
    • Export Citation
  • Tennant, W., , and Beare S. , 2014: New schemes to perturb sea-surface temperature and soil moisture content in MOGREPS. Quart. J. Roy. Meteor. Soc., 140, 11501160, doi:10.1002/qj.2202.

    • Search Google Scholar
    • Export Citation
  • Theis, S. E., , Hense A. , , and Damrath U. , 2005: Probabilistic precipitation forecasts from a deterministic model: A pragmatic approach. Meteor. Appl., 12, 257268, doi:10.1017/S1350482705001763.

    • Search Google Scholar
    • Export Citation
  • Thrush, D., , Burningham K. , , and Fielding J. , 2005: Flood warning for vulnerable groups: A qualitative study. R&D Tech. Rep. W5C-018/3, Environment Agency, 73 pp. [Available online at http://www.envirobase.info/PDF/3557_1.pdf.]

  • Tracton, M. S., , and Kalnay E. , 1993: Operational ensemble prediction at the National Meteorological Center: Practical aspects. Wea. Forecasting, 8, 379398, doi:10.1175/1520-0434(1993)008<0379:OEPATN>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Tracton, M. S., , Du J. , , Toth Z. , , and Juang H. , 1998: Short-Range Ensemble Forecasting (SREF) at NCEP/EMC. Proc. 12th Conf. on Numerical Weather Prediction, Phoenix, AZ, Amer. Meteor. Soc., 269–272.

  • Twigger-Ross, C., , Fernandez-Bilbao A. , , Tapsell S. , , Walker G. , , Watson N. , , Deeming H. , , and Kashefi E. , 2008: Improving flood warnings: Final report. Improving institutional and social responses to flooding, Science Rep. SC060019, Work Package 1a, Environment Agency, 76 pp.

  • Vié, B., , Nuissier O. , , and Ducrocq V. , 2011: Cloud-resolving ensemble simulations of Mediterranean heavy precipitating events: Uncertainty on initial conditions and lateral boundary conditions. Mon. Wea. Rev., 139, 403423, doi:10.1175/2010MWR3487.1.

    • Search Google Scholar
    • Export Citation
  • Vié, B., , Molinié G. , , Nuissier O. , , Vincendon B. , , Ducrocq V. , , Bouttier F. , , and Richard E. , 2012: Hydrometeorological evaluation of a convection-permitting ensemble prediction system for Mediterranean heavy precipitating events. Nat. Hazards Earth Syst. Sci., 12, 26312645, doi:10.5194/nhess-12-2631-2012.

    • Search Google Scholar
    • Export Citation
  • Walser, A., , Luthi D. , , and Schar C. , 2004: Predictability of precipitation in a cloud-resolving model. Mon. Wea. Rev., 132, 560577, doi:10.1175/1520-0493(2004)132<0560:POPIAC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Waters, E., , Weinstein N. , , Colditz G. , , and Emmons K. , 2006: Formats for improving risk communication in medical trade off decisions. J. Health Commun., 11, 167182, doi:10.1080/10810730500526695.

    • Search Google Scholar
    • Export Citation
  • Wedawatta, G., , Ingirige B. , , and Proverbs D. , 2014: Small businesses and flood impacts: The case of the 2009 flood event in Cockermouth. J. Flood Risk Manage., 7, 4253, doi:10.1111/jfr3.12031.

    • Search Google Scholar
    • Export Citation
  • Weusthoff, T., , Ament F. , , and Arpagaus M. , 2010: Assessing the benefits of convection-permitting models by neighborhood verification: Examples from MAP D-PHASE. Mon. Wea. Rev., 138, 34183433, doi:10.1175/2010MWR3380.1.

    • Search Google Scholar
    • Export Citation
  • Wikipedia, 2015: Hootsuite. Accessed 28 March 2016. [Available online at https://en.wikipedia.org/wiki/Hootsuite.]

  • Wilson, D., , and Ballard S. P. , 1999: A microphysically based precipitation scheme for the UK Meteorological Office Unified Model. Quart. J. Roy. Meteor. Soc., 125, 16071636, doi:10.1002/qj.49712555707.

    • Search Google Scholar
    • Export Citation
  • Yamagishi, I., 1997: When a 12.86% mortality is more dangerous than 24.14%: Implications for risk communication. Appl. Cognit. Psychol., 11, 495506, doi:10.1002/(SICI)1099-0720(199712)11:6<495::AID-ACP481>3.0.CO;2-J.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Rainfall accumulations from 0900 to 2100 UTC 8 Jul 2011 from the Met Office composite radar network.

  • View in gallery

    Rainfall accumulations from 0900 to 2100 UTC from (a) the 12 regional 18-km ensemble forecasts and (b) the 12 U.K. 2.2-km ensemble forecasts. Ensemble member numbers are marked in the upper left corner of each map.

  • View in gallery

    (a) Bar chart of maximum 3-h rainfall accumulations from the 12 ensemble members in the period 0900–2400 UTC 8 Jul 2011 within 10 (black) and 25 km (gray) of Edinburgh. (b) Ensemble forecasts of the probability of exceeding 40 mm of rain in any 3-h period between 0900 and 2400 UTC 8 Jul 2011: green and red pixels show probabilities estimated from the fraction of exceedances in each grid square. (c) Ensemble forecasts of the probability of exceeding 40 mm of rain in any 3-h period between 0900 and 2400 UTC 8 Jul 2011: colored shading indicates probabilities estimated from the mean fraction of exceedances within 25 km of each grid square using neighborhood processing. Crosses in (b) and (c) are radar-derived observations. The black square has width 25 km and is centered over Edinburgh.

  • View in gallery

    Map of locations referred to in the descriptions of the case studies.

  • View in gallery

    (a) The 0600–2100 UTC 5 Jul 2012 pointwise maximum of the hourly radar accumulations and (b) 1400–1500 UTC 5 Jul 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 5 Jul 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1400–1500 UTC 5 Jul 2012 from the 0300 UTC run. Locations of reported surface water flooding are indicated in (a) and (c) by double wavy lines.

  • View in gallery

    (a) The 0600–2100 UTC 29 Jul 2012 pointwise maximum of the hourly radar accumulations and (b) 1300–1400 UTC 29 Jul 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain during 0600–2100 UTC 29 Jul 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain during 1300–1400 UTC 29 Jul 2012 from the 0300 UTC run.

  • View in gallery

    (a) The 0600–2100 UTC 9 Jul 2012 pointwise maximum of the hourly radar accumulations and (b) 1300–1400 UTC 9 Jul 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 9 Jul 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1300–1400 UTC 9 Jul 2012 from the 0300 UTC run. Locations of reported surface water flooding are indicated in (a) and (c) by double wavy lines.

  • View in gallery

    (a) The 0600–2100 UTC 13 Jul 2012 pointwise maximum of the hourly radar accumulations and (b) 1700–1800 UTC 13 Jul 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 13 Jul 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1700–1800 UTC 13 Jul 2012 from the 0300 UTC run. Locations of reported surface water flooding are indicated in (a) and (c) by double wavy lines.

  • View in gallery

    (a) The 0600–2100 UTC 5 Aug 2012 pointwise maximum of the hourly radar accumulations and (b) 1400–1500 UTC 5 Aug 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 5 Aug 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1400–1500 UTC 5 Aug 2012 from the 0300 UTC run. Locations of reported surface water flooding are indicated in (a) and (c) by double wavy lines.

  • View in gallery

    (a) The 0600–2100 UTC 17 Aug 2012 pointwise maximum of the hourly radar accumulations and (b) 1100–1200 UTC 17 Aug 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 17 Aug 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1100–1200 UTC 17 Aug 2012 from the 0300 UTC run. Locations of reported surface water flooding are indicated in (a) and (c) by double wavy lines.

  • View in gallery

    (a) The 0600–2100 UTC 24 Aug 2012 pointwise maximum of the hourly radar accumulations and (b) 1500–1600 UTC 24 Aug 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 24 Aug 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1500–1600 UTC 24 Aug 2012 from the 0300 UTC run.

  • View in gallery

    Summary of trial results for the occurrence or not of intense and/or flood-producing rainfall somewhere in England and Wales, sometime during each day. (a) Calendar of daily intense rainfall/flood occurrence and forecast probability: black (white) cells are days when hourly accumulations of more than 16 mm were (were not) observed by radar. The figures are the maximum forecast probability of exceeding 16 mm h−1 sometime during the day. A dot after the forecast symbol indicates that surface water flooding was reported. Gray cells are missing data. (b) Reliability plot of forecast probability of intense rain vs observed frequency of intense rain (dotted line)/observed frequency of surface water flooding (dashed line). Results from a perfect forecasting system would lie on the 45° diagonal (full line). (c) Frequency of forecast probability ranges.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 175 175 17
PDF Downloads 93 93 7

MOGREPS-UK Convection-Permitting Ensemble Products for Surface Water Flood Forecasting: Rationale and First Results

View More View Less
  • 1 Met Office, Exeter, United Kingdom
  • | 2 Met Office@Reading, Reading University, Reading, United Kingdom
  • | 3 Met Office, Exeter, United Kingdom
© Get Permissions
Full access

ABSTRACT

Flooding is one of the costliest hazards in the United Kingdom. A large part of the annual flood damage is caused by surface water flooding that is a direct result of intense rainfall. Traditional catchment-based approaches to flood prediction are not applicable for surface water floods. However, given sufficiently accurate forecasts of rainfall intensity, with sufficient lead time, actions can be taken to reduce their impact. These actions require reliable information about severity and areas at risk that is clear and easily interpreted. The accuracy requirements, in particular, are very challenging, as they relate to prediction of intensities that occur only infrequently and that typically affect only small areas. In this paper, forecasts of intense rainfall from a new convection-permitting ensemble prediction system are evaluated using radar observations of intense rain and surface water flooding reports. An urban flooding case that occurred in Edinburgh in 2011 is first investigated and then a broader look is taken at performance through a 3-month period during the London Olympic and Paralympic Games in 2012. Conclusions are drawn about the value of the ensemble and the particular means of presenting the forecasts, and areas requiring further work are highlighted.

Current affiliation: Aspen Re, Zurich, Switzerland.

Corresponding author address: Brian Golding, Met Office, FitzRoy Road, Exeter EX1 3PB, United Kingdom. E-mail: brian.golding@metoffice.gov.uk

ABSTRACT

Flooding is one of the costliest hazards in the United Kingdom. A large part of the annual flood damage is caused by surface water flooding that is a direct result of intense rainfall. Traditional catchment-based approaches to flood prediction are not applicable for surface water floods. However, given sufficiently accurate forecasts of rainfall intensity, with sufficient lead time, actions can be taken to reduce their impact. These actions require reliable information about severity and areas at risk that is clear and easily interpreted. The accuracy requirements, in particular, are very challenging, as they relate to prediction of intensities that occur only infrequently and that typically affect only small areas. In this paper, forecasts of intense rainfall from a new convection-permitting ensemble prediction system are evaluated using radar observations of intense rain and surface water flooding reports. An urban flooding case that occurred in Edinburgh in 2011 is first investigated and then a broader look is taken at performance through a 3-month period during the London Olympic and Paralympic Games in 2012. Conclusions are drawn about the value of the ensemble and the particular means of presenting the forecasts, and areas requiring further work are highlighted.

Current affiliation: Aspen Re, Zurich, Switzerland.

Corresponding author address: Brian Golding, Met Office, FitzRoy Road, Exeter EX1 3PB, United Kingdom. E-mail: brian.golding@metoffice.gov.uk

1. Introduction

Flooding is one of the costliest natural hazards in the world today, with average annual losses being more than $60 billion and average fatalities being more than 15 000 per annum [figures for flooding, excluding coastal flooding by storm surges, are from the Centre for Research on the Epidemiology of Disasters (CRED) Emergency Events Database (EM-DAT) for the period 1993–2012]. Recent high-impact events have included the Thailand floods in August 2011, which cost $40 billion (Guha-Sapir et al. 2012) and, in the United Kingdom, the 2007 floods, which cost $8 billion (Scheuren et al. 2008; Pitt 2009). Predictions of flooding from major rivers have become much more accurate in recent years as a result of improved modeling of the weather and surface hydrology. However, substantial damage and distress is also caused by smaller weather systems that produce intense rainfall lasting just a few hours. Where this exceeds the carrying capacity of urban drainage systems (surface water and sewer flooding) or where it collects rapidly into torrents in steep terrain (flash flooding), it can cause considerable localized damage and loss of life. Recent examples from the United Kingdom include Morpeth in 2009 (Wedawatta et al. 2014), Boscastle in 2004 (Golding et al. 2005), and Ottery St Mary in 2009 (Clark 2011). Pitt (2009) estimated that more than half of the insurance costs of floods in the United Kingdom are due to surface water flooding. This problem is particularly acute in urban areas where sealing of the ground surface results in very rapid runoff of rainfall. Drainage networks are designed to carry excess rainfall away, but even when new, they are designed to deal with a limited volume of water. In old systems the interconnection of separately designed pipes, build-up of sediment and debris, and damage from ground movement all lead to reductions in carrying capacity. In England, Falconer et al. (2009) showed that widespread surface water flooding is generally associated with hourly to 6-hourly rainfall accumulations with a return period exceeding 3% probability per year.

a. The nature of extreme convective precipitation in the United Kingdom

Hand et al. (2004) classified extreme rainfall events in the United Kingdom from three origins: fronts associated with depressions, orographic uplift of warm conveyor belts, and convective storms. The first two are the dominant causes of long-duration extremes that cause major river flooding, while the last is the primary cause of surface water floods. Convective storms are characterized by strong vertical motions that lift large volumes of moisture from the boundary layer to high altitudes, where the moisture condenses and, in U.K. latitudes, freezes. The microphysical processes within this dense mixture of ice crystals, frozen raindrops, and cloud drops lead to the formation of graupel and hail, and may generate lightning. U.K. convection is strongly influenced by its maritime location. Boundary layer humidity tends to be high and strong convective inhibition is rarely observed. Extreme convective precipitation is frequently associated with surface airmass boundaries, such as fronts, and upper-level dynamical features, such as potential vorticity anomalies or filaments (Roberts 2000).

b. Forecasting convective precipitation

Historically, rainfall predictions from NWP models have not met the needs of real-time hydrological prediction. For large river systems, short-range predictability was obtained using upstream river and rainfall gauging, while the rainfall variability needed for prediction in smaller rivers and for surface water flooding was not resolvable by NWP. Where NWP predictions were used, they needed substantial calibration, as they could not capture the intense local rainfall peaks that cause floods. With the introduction of kilometer-scale limited-area NWP models (LAMs), that situation has changed, and it is now possible to provide useful predictions by directly feeding precipitation forecasts into hydrological models for small river systems. For surface water flooding, the scale of interest is even finer, and it is the details of local runoff and inundation that are of interest (Falconer et al. 2009). The development of surface water inundation forecasts depends on the potential capability of kilometer-scale NWP models. Such models differ from current global models in being able to represent directly the vertical air motions associated with convective rainstorms. Several studies have shown the benefits of this (Done et al. 2004; Lean et al. 2009; Clark et al. 2009). The location, timing, and character of the predicted rainstorms depend critically on larger-scale information imported from a global model through boundary conditions. The use of convection-permitting models improves the realism of simulations of quantitative precipitation forecasts (Weusthoff et al. 2010; Clark et al. 2012). Tang et al. (2013) described the 1.5-km convection-permitting configuration of the Met Office Unified Model (MetUM) now used as primary guidance for U.K. short-range weather forecasts. It has been very successful, particularly in its prediction of the realistic structure of rain systems that are either convectively generated or have convection embedded (Lean et al. 2008). Similar operational forecasting systems have been implemented by the weather services of France (Seity et al. 2011) using a 2.5-km configuration of the Applications of Research to Operations at Mesoscale (AROME) model, Germany (Steppeler et al. 2003; Baldauf et al. 2011) using a 2.8-km configuration of the Consortium for Small-Scale Modeling (COSMO) model, Japan (Hirahara et al. 2012) using a 2-km configuration of the Local Forecast Model (LFM), and the United States using the 3-km High-Resolution Rapid Refresh (HRRR) model (Ikeda et al. 2013).

c. Ensemble prediction systems

A single, deterministic forecast of the future state of the atmosphere is unlikely to match reality because of incompleteness of the initial state specification and because of biases in the representation of atmospheric processes. For some purposes, the fact that the forecast is usually near to reality will be sufficient. However, when lives and property are at stake, potential consequences of different possible weather outcomes need to be considered. An ensemble prediction system (EPS) produces a range of forecast scenarios that, taken together, can be used to assess the likelihood of particular hazardous weather situations occurring or of hazardous thresholds being surpassed. NWP ensembles were initially focused on medium-range prediction of surface pressure and 500-hPa height, relying on the predictability characteristics of low Rossby number midlatitude weather systems (Tracton and Kalnay 1993; Molteni et al. 1996). They have proved to be a successful way of dealing with the inherent uncertainty of weather and climate forecasts (Buizza et al. 2005) and can now be considered a mature capability.

Requirements for tackling forecast uncertainty in short-range forecasts of precipitation are very different from those that formed the basis of medium-range EPSs. Bowler et al. (2008) outlined the differences: in resolution, initial perturbations effective at short lead times, and representation of model uncertainties that impact on surface variables. High resolution implies nesting, which introduces the complication of matching boundary perturbations. NCEP’s Short-Range Ensemble Forecast (SREF) was the pioneering short-range ensemble, based on research combining two LAMs with boundary conditions taken from the NCEP medium-range global ensemble (Tracton et al. 1998; Stensrud et al. 1999). An international intercomparison of mesoscale EPSs was held in the World Weather Research Programme (WWRP) Beijing Olympics 2008 Research and Development Project (Kunii et al. 2011; Duan et al. 2012). With the implementation of kilometer-scale LAMs, research has turned to configuring EPSs at this scale. The processes involved in convective cloud systems are a challenge to initial perturbation specification with much greater growth rates anticipated, while many more model processes need to be explored as potential sources of uncertainty, yet the large scales continue to provide the environment for convective development and so remain important sources of uncertainty. Clark et al. (2009, 2011), using data from the 2007 NOAA Hazardous Weather Testbed Spring Experiment, discussed the factors influencing the growth of ensemble spread. Leoncini et al. (2010) performed case studies of an intense convecting storm using randomly perturbed potential temperature fields with the aim of identifying physical processes that lead to perturbation growth on the convective sale. Hanley et al. (2011, 2013) and Barrett et al. (2014) use ensemble case studies to show the relationship between the predictability of convective rainfall and large-scale uncertainties. Several papers have investigated sensitivity to initial and boundary conditions, including Walser et al. (2004), Kong et al. (2007), Hohenegger et al. (2008), Vié et al. (2011, 2012), Peralta et al. (2012), and Gebhardt et al. (2011). These broadly support the hypothesis that, on average, initial conditions dominate up to 6 h ahead while boundary condition and model perturbations are more important by 12 h ahead.

d. Communication of weather information

The results presented in this paper depend on the choice of user presentation formats as well as the prediction methodology. Probabilistic information is less familiar to recipients than deterministic information and requires different approaches to postprocessing and presentation. The problems faced by meteorologists in communicating the risk of flooding from intense rainfall are similar to those studied by hydrologists dealing with larger-scale flood risk (Twigger-Ross et al. 2008; Thrush et al. 2005; Tapsell et al. 2005; Burningham et al. 2008), by the medical profession in communicating the side effects of treatments (Paling 2003; Goldman et al. 2006), and by emergency managers dealing with disasters (Kloprogge et al. 2007). Extensive research in these fields has shown that the interpretation placed on a warning is dependent on a wide variety of factors, ranging from the ability of the recipient to understand the words used (e.g., “technical” words, such as probability or precipitation), their meaning (e.g., does a 30% chance of rain mean it will rain in 30% of places, for 30% of the time, or that there is 30% confidence of any rain), relating the warning to experience (e.g., a previous flood at this location), trust (e.g., who issued the warning?, what do I think of their track record?), and preparedness (e.g., what can I do about it?). In recent years an extensive literature has developed on the use and interpretation of probability forecasts. Morss et al. (2008) and Joslyn and Savelli (2010) showed that people interpret deterministic forecasts as uncertain, but that the inferred uncertainty may not be consistent with expert opinion, and that recipients would prefer that it was made explicit. Controlled experiments reported by Roulston et al. (2006), Joslyn et al. (2007), Roulston and Kaplan (2009), Stephens et al. (2011), and Peachey et al. (2013) support the hypothesis that appropriately presented probabilistic forecasts lead to better decisions. However, there are also many examples (Murphy et al. 1980; Joslyn et al. 2009; Gigerenzer et al. 2005; Morss et al. 2008; Joslyn and Savelli 2010; Bell and Tobin 2007) suggesting that people often misunderstand the nature of probability forecasts. Gigerenzer et al. (2005) found improved understanding when the definition of probability was stated explicitly: for example, “on days like today, rain occurs at X on 3 occasions in 10.” Studies in medicine (Edwards et al. 2002; Yamagishi 1997; Waters et al. 2006; Dolan and Iadarola 2008; Knapp et al. 2004; Kurz-Milcke et al. 2008) have found that multiple representations of the same evidence may be needed to successfully convey information, especially for small probabilities. Further, Gill (2008), Schirillo and Stone (2005), Faulkner et al. (2007), Edwards et al. (2002), Price et al. (2007), Peters et al. (2008), and Stone et al. (1997) all emphasize the benefits of using combinations of text, numbers, and graphics to communicate risk information.

In this paper we present the rationale and results of a trial of rainfall probability products from a new kilometer-scale EPS covering the United Kingdom. The trial was part of a broader experiment (Golding et al. 2014) carried out during the 2012 Olympic and Paralympic Games. The trial described in this paper used a simple rainfall threshold expressed in language designed to make the connection with people’s experience as direct as possible. Raw outputs of the system have also been coupled to a 1-km grid rainfall–runoff model of England and Wales (Price et al. 2012). Use of this coupled system as a proof of concept in support of the Glasgow Commonwealth Games in 2014 is the subject of a separate paper (Moore et al. 2015). Section 2 describes the structure of the ensemble prediction system. Section 3 presents results from a single case study of surface water flooding in Edinburgh, Scotland, which was used to guide the design of the trial. In section 4 we assess the probability outputs for several heavy rainfall events during the 2012 trial period. In section 5 we discuss some of the lessons from the trial, and in section 6 we draw some overall conclusions.

2. The forecasting approach

The NWP model used for this study was the MetUM, which solves the equations of atmospheric motion using the integration scheme of Davies et al. (2005). Representations of other physical processes include radiation (Edwards and Slingo 1996), mixing in the lower part of the atmosphere (Lock et al. 2000; Lock 2001), clouds and precipitation (Wilson and Ballard 1999), and (in the global and regional configurations) convection (Gregory and Rowntree 1990). The Met Office Surface Exchange Scheme (MOSES) is used to model the exchanges of heat, moisture, and momentum between the surface and atmosphere (Essery et al. 2003). Performance at convection-permitting resolutions has been investigated by Lean et al. (2008) and Tang et al. (2013). In this configuration, clouds are represented as interacting mixtures of cloud drops, snow, ice crystals, graupel, and rain.

The Met Office Global and Regional Ensemble Prediction System (MOGREPS) used in this research consisted of a three-level cascade: MOGREPS-G (Tennant and Beare 2014), with a global 60-km grid; MOGREPS-R, with a regional 18-km grid; and MOGREPS-UK (Tennant 2015), with a local 2.2-km grid (note that, subsequent to this research, MOGREPS-UK is now nested directly inside a 33-km grid MOGREPS-G). MOGREPS-G was initialized using the Met Office four-dimensional variational data assimilation (4D-Var) system (Lorenc et al. 2000) and used the local ensemble transform Kalman filter (LETKF) perturbation technique of Bowler et al. (2009) to generate its initial perturbations. MOGREPS-R used a regional 4D-Var initialization with perturbations downscaled from MOGREPS-G. MOGREPS-UK used both initial state and perturbations downscaled from MOGREPS-R. The use of downscaled perturbations avoids disturbances due to mismatched boundary conditions.

The Edinburgh case study described in section 3 was run using the MOGREPS-UK 2.2-km ensemble in offline, experimental mode. It was initialized from the 3-h forecast of the 1500 UTC operational MOGREPS-R ensemble run, the day before the flood. A control and 11 perturbed members were run.

The 2012 trial described in section 4 was run using the MOGREPS-UK 2.2-km ensemble running a 6-hourly cycle as part of the real-time operational schedule and using initial conditions from the 3-h MOGREPS-R forecast, but without operational use of the products. A control and 11 perturbed members were run in each cycle.

Bearing in mind the lack of high-resolution data assimilation, forecasts were only presented for lead times of 6 h or more to avoid the spinup period for convective-scale motions. Forecast products were generated from the ensemble outputs according to an assessment of the information that might be of most use to the intended audience. The Edinburgh case study was aimed at professional users in flood management. For this audience, a rainfall accumulation threshold with specific relevance to surface water flooding was selected. As will be discussed in section 3, presentation of the probabilities of intense rainfall evolved in response to the findings of this case. The choice of a presentation that represented occurrence of intense rainfall within a radius of any point was guided by the areal responsibilities of the managers. When taking these lessons into the 2012 trial, choices needed to be adjusted for a public audience. In this case, the intense rain products formed part of a portfolio of probability products designed to answer common questions that had been recorded during a variety of public engagement exercises [e.g., during the work described in Stephens et al. (2011)] and provided by other public authorities (e.g., during four roadshows on public requirements for surface water flood alerts in early 2012; Environment Agency 2012, unpublished report). These indicated that recipients would be largely concerned with what happened in their street, but over a range of time scales from an hour up to the period of daylight.

Verification of the precipitation forecasts is carried out using observations from the U.K. network of C-band radars. Over most of England and Wales the density of radars ensures that any location lies within 100 km of a radar (at which range, the median beam height is about 1 km above ground level), minimizing the impact of the bright band and low-level modification of the rain rate (e.g., from evaporation). Following signal processing for clutter removal, the radar returns are processed for spurious echo removal, range correction, and vertical profile correction (including brightband removal) and are calibrated using time series comparison with nearby rain gauges (Kitchen et al. 1994; Harrison et al. 2014). The resulting single-site precipitation estimates are composited onto a 1-km U.K. grid, using the lowest beam at each grid location. The 5-min corrected radar rate estimates are aggregated to hourly 2-km gridded values for comparison with the model predictions. These 2-km gridded rainfall estimates are compared routinely with rain gauges lying within 100 km of each radar. The root-mean-square difference during summer 2012 was 1.02 mm in 68 247 nonzero hourly rainfall comparisons. A significant part of that difference is attributable to differences in spatial and temporal sampling.

While rainfall verification is important in checking whether the rainfall forecast is accurate, it does not address the issue of why a member of the public would want to know about rainfall of this intensity, which is mainly for surface water flooding leading to impassable roads and flooded homes. The joint Flood Forecasting Centre (FFC) of the Environment Agency (EA) and the Met Office (FFC 2015) is responsible for the issue of flood alerts for England and Wales. One of their responsibilities is routine real-time verification, including for surface water flooding. Flood data are gathered by FFC staff from a variety of sources: primarily the Internet, using Google to search for news items mentioning flooding, but also social media using Hootsuite (Wikipedia 2015) to retrieve tweets that mention flooding and associated words, and internal (EA) reporting. Data gathering is the responsibility of the duty hydrometeorologist but is delegated to other members of the team as the operational workload increases during an event. Responsibility is not just to gather information but to assess its veracity (e.g., exaggerated local press report or perhaps a burst water main) and classify the observation with a severity and flood source. The system has developed over time. Summer 2012 fell midway through this process of maturing.

3. Preliminary case study

Following a very dry spring of 2011, soil moisture deficits in Scotland had recovered somewhat during a moderately wet June (CEH 2011). A brief anticyclonic spell in early July was replaced from 6 July by a cutoff low that moved slowly northeastward with regions of convective instability circulating around it. On 8 July 2011, parts of Edinburgh experienced flooding from heavy downpours associated with slow-moving thunderstorms. Daily gauges in the area reported up to 43.6 mm. Radiosonde ascents from close to the region of interest (not shown) indicated the possibility of slow-moving heavy storms, supported by model profiles. Figure 1 shows the accumulated rainfall for the daytime period observed by the radar network.

Fig. 1.
Fig. 1.

Rainfall accumulations from 0900 to 2100 UTC 8 Jul 2011 from the Met Office composite radar network.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

The global ensemble from 1200 UTC 7 July was used to initialize a regional ensemble from 1500 UTC and then the high-resolution U.K. ensemble from 1800 UTC, providing a forecast that would have been available for planning decisions prior to the day of the flood. The control run of the regional ensemble described the large-scale meteorological conditions well. However, both the control run and the individual ensemble members simulated the precipitation rather poorly (Fig. 2a)—precipitation was too widespread over land and the highest accumulations were significantly underestimated. Furthermore, too much rain was forecast on the upwind (southeast) slope of the mountains. This behavior is typical of a model that uses a parameterization scheme to represent convective clouds.

Fig. 2.
Fig. 2.

Rainfall accumulations from 0900 to 2100 UTC from (a) the 12 regional 18-km ensemble forecasts and (b) the 12 U.K. 2.2-km ensemble forecasts. Ensemble member numbers are marked in the upper left corner of each map.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

The 2.2-km ensemble members were more realistic (Fig. 2b). Rainfall totals are much closer to observations and the patterns more realistic. The upwind side of mountains had less precipitation in some members, including the control. Most of the forecasts show widespread showers being blown onto the east coast and intensifying over high ground. However, the relative intensities over the North Sea and over land vary from member 1, with no rain over the sea, to member 6, with widespread rain over the sea. Comparing Fig. 2b with Fig. 1, it is evident that a majority of the ensemble members have storms over the high ground that exceed the intensity of any observed storm in that area. On the other hand, the storms forecast in the lowland area around and to the west of Edinburgh have accumulations that are similar to or rather less than those observed. The difference is most likely linked to differences in the model responses to strong topographic forcing over the mountains as opposed to the weaker forcing farther south.

Looking at the Edinburgh area (see Fig. 3b for its location), Fig. 3a shows the maximum 3-h rainfall within 10 and 25 km of the city from each ensemble member. The upper dashed line marks the “40 mm in 3 h” threshold for “extreme” rainfall alerts to emergency responders for surface water flooding. The graph shows that 4 out of 12 members exceeded the alert threshold for a flood-generating storm within 25 km, with two near misses exceeding a lower threshold of 30 mm. Within 10 km, there were just two near misses. In this case, a 25-km-wide averaging domain appears to be needed to provide an actionable signal.

Fig. 3.
Fig. 3.

(a) Bar chart of maximum 3-h rainfall accumulations from the 12 ensemble members in the period 0900–2400 UTC 8 Jul 2011 within 10 (black) and 25 km (gray) of Edinburgh. (b) Ensemble forecasts of the probability of exceeding 40 mm of rain in any 3-h period between 0900 and 2400 UTC 8 Jul 2011: green and red pixels show probabilities estimated from the fraction of exceedances in each grid square. (c) Ensemble forecasts of the probability of exceeding 40 mm of rain in any 3-h period between 0900 and 2400 UTC 8 Jul 2011: colored shading indicates probabilities estimated from the mean fraction of exceedances within 25 km of each grid square using neighborhood processing. Crosses in (b) and (c) are radar-derived observations. The black square has width 25 km and is centered over Edinburgh.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

More widely across Scotland, Fig. 3b shows just a scattering of grid squares with one or more members (i.e., ≥8% probability) exceeding the 40 mm in 3 h threshold, suggesting discrete, disconnected locations at risk of intense rainfall. This is clearly misleading—the precise locations are unpredictable and we would expect probability to vary smoothly—and is due to the ensemble being too small. Inspection shows that only about 1% of pixels in Fig. 3b have nonzero probability. Given that all pixels should have nonzero probability, it would appear that at least 100 times as many ensemble members are needed, giving a number that is impractically large with current computers. Instead, we use postprocessing to generate a smooth probability field that more realistically indicates the likelihood of extreme rainfall (Fig. 3b). First, for each pixel and ensemble member, we note whether a nearby pixel, within 25 km, exceeded the threshold (40 mm in 3 h). If so, the pixel was set to 1 to indicate that an event happened nearby; otherwise it was set to 0. This spreads the high probability pixels to a circle of radius 12 pixels on the 2-km grid. The probabilities displayed are then the fraction of those pixels with a value of 1 within 25 km of each pixel. In other words, it gives the probability of at least one extreme occurrence within 25 km of a location given a spatial uncertainty of 25 km in the forecasts. This is a variation on neighborhood approaches described by Germann and Zawadzki (2004) and Theis et al. (2005) for forecasting and by Roberts and Lean (2008) for verification. To the extent that the residual uncertainty is spatial, it effectively increases the ensemble size by a factor of 169. The resulting probability distribution in Fig. 3c compares well with the locations of observed 3-hourly rainfall totals exceeding 30 and 40 mm. Highest probabilities are forecast over the mountains to the north of Edinburgh, where several storms of near to the threshold intensity were observed. Farther south, over Edinburgh, a probability of 30%–40% of exceeding the threshold within 25 km is predicted.

The fact that the most intense storm did not occur in the region of highest probability is not necessarily inconsistent. Indeed, given suitable conditions of convective instability and inhibition, it is to be expected that convective storms will be less probable, but more intense, in an area of weaker surface forcing. On the other hand, the overprediction of rainfall, noted earlier in the area of strong surface forcing over the mountains, has resulted in erroneously raised probabilities in that area relative to those farther south. Nevertheless, taking account of the vulnerability of the urban fabric to rainfall of this intensity, the forecast risk (taken as the product of probability and vulnerability) was clearly highest in Edinburgh, and a probability of 30%–40% would certainly be sufficient to alert emergency responders to the risk of surface water flooding impacts in the city.

Results from this case study support the principle of using a convective-scale ensemble to predict a range of scenarios that could be expected to include storms similar in magnitude and location to those observed. However, if the output of an affordable ensemble is to provide useful information to surface water flood managers, it is clear that great care will be needed in postprocessing and presenting the results.

4. The 2012 trial

During the Olympic and Paralympic Games in July–September 2012, an ensemble configuration similar to that used for the Edinburgh case study was trialed in real time. Probabilistic products were used by forecasters and made available to the public on a research website set up for the purpose (Golding et al. 2014). Several rainfall rate thresholds were used (0.2, 4, 8, and 16 mm h−1) described as light, heavy, very heavy, and intense rainfall. The upper rain-rate threshold was selected on the basis that it approximated to the thresholds of 30 mm in 1 h, 40 mm in 3 h, and 50 mm in 6 h found by Falconer et al. (2009) to produce widespread surface water flooding in urban areas of lowland England. They represent average 30-yr return periods in this area. We presented maps of the probability of exceeding each threshold at some time in a period, either in 1 h or in the entire daytime period (taken as 0600–2100 UTC). Thus, we provided information on the likelihood of rain occurring that would be intense enough to cause localized flooding, both at some time during the following day, to enable forward decisions to be made about staffing and equipment availability, and at some time during each hour of the current day, to enable more focused decisions on readiness of response teams. Shades of gray were chosen for the maps so as to provide the same visual relation of increasing density to increasing probability, regardless of the color vision of the user. Explanatory material was provided on the website to help members of the public interpret the maps.

In designing the trial, a key decision was the form of the postprocessing to be undertaken on the ensemble probabilities, bearing in mind the conclusions of the Edinburgh case study, the fact that the audience was the public rather than flood forecasters, and the wider range of variables and thresholds that were being presented. In the light of previous interaction with the public and with emergency managers who deal with the public, it was concluded that maps of rainfall probabilities for an area around each location would be confusing and that the meaning of site-specific probabilities was more likely to be understood by members of the public. At the same time, the length scale of the neighborhood processing was reduced from 25 to 15 km in the light of preliminary (unpublished) verification results.

Initial implementation of the ensemble took place in early June 2012, but full product generation and archiving commenced on 5 July. Nine days of data were missed in the first 2 weeks because of archive malfunctions, and a further 6 days are missing through the rest of the trial, which completed on 30 September. A total of 77 days of data are available for analysis.

Following a dry start to the year, summer 2012 was notably wet in the United Kingdom. From June to early August, Atlantic depressions repeatedly approached the United Kingdom from the west, becoming slow moving with active, often intense, convection over the United Kingdom, leading to frequent localized surface water flooding. From the middle of August, the depressions became less frequent and the precipitation more stratiform. September was initially dry, but heavy rainfall returned later.

Initial inspection of the radar observations and forecast probabilities indicated that a direct comparison might not provide a good match to the perceptions of forecast users. It was evident from comparing the area over which the observed radar rate exceeded 16 mm h−1 sometime during an hour with the area over which the hourly observed radar accumulation exceeded 16 mm that the former greatly overstated the extent of observed disruption from flooding. We chose to focus on the surface water flood impact, and hence to use the 16-mm threshold hourly accumulation area as the verifying observation, accepting that this implies a recalibration of the MOGREPS-UK products.

Although forecasts were generated and presented for the whole United Kingdom, the quality of the radar observations is much reduced over the complex terrain of Scotland, and we only have a record of surface flooding for England and Wales. We have therefore chosen to restrict the analysis presented here to England and Wales.

From the 77 cases for which both radar data and ensemble forecasts were available, we selected seven cases that illustrate characteristic strengths and weaknesses of the forecasts. For each case, we present the maximum hourly rainfall accumulation though the 15-h period 0600–2100 UTC, the selected hourly rainfall accumulation when the rainfall was at its most intense (usually shortly after midday), the forecast probability of intense rain sometime during the 15-h period 0600–2100 UTC of the next day from the 1500 UTC forecast run, and the forecast probability of intense rain during the selected hour from the 0300 UTC run on the same day. The forecasts were selected, taking account of information gained from the four roadshows on public requirements for surface water flood alerts held in early 2012 (Environment Agency 2012, unpublished report). These indicated a need for an evening alert, to be used in deciding on the following day’s activities, updated with more detail in the early morning for planning prior to their start. Consistent with this, the 0300 UTC forecast usually provides more reliable detail at smaller space and time scales than the previous day’s 1500 UTC forecast run. Locations referred to in the text are shown in Fig. 4.

Fig. 4.
Fig. 4.

Map of locations referred to in the descriptions of the case studies.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

a. Case 1: 5 July 2012

Widespread heavy showers associated with a deep cutoff low centered to the southwest of the United Kingdom were mostly focused along an old cold front lying north-northwest–south-southeast from western Scotland to eastern England (Fig. 5a). Hourly radar accumulations show that several storms produced hourly totals in excess of 16 mm, with isolated hourly accumulations of over 32 mm. The first storm developed over the hills of northern England at 1000 UTC and was succeeded from noon by areas of storms in southwestern Scotland and eastern England (Fig. 5b). From 1700 UTC a new area of storms developed in northwestern England, lasting into the evening. Flooding was reported in northwestern and northeastern England and also to the north of London (see Table 1). The forecast probability of intense rainfall from 1500 UTC for 0600–2100 UTC (Fig. 5c) captured extremely well the general areas at risk of these large hourly rain accumulations, though with no areas exceeding a 60% probability. The hourly probability maps depend much more on temporal forecast accuracy and are therefore more susceptible to small-scale forecast errors. Hourly probabilities from the 0300 UTC run (Fig. 5d) missed an early storm. Otherwise, the general patterns were quite well picked out at the 10%–20% probability level, though with a tendency for probabilities to increase 1–2 h after the storms actually reached the threshold intensity.

Fig. 5.
Fig. 5.

(a) The 0600–2100 UTC 5 Jul 2012 pointwise maximum of the hourly radar accumulations and (b) 1400–1500 UTC 5 Jul 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 5 Jul 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1400–1500 UTC 5 Jul 2012 from the 0300 UTC run. Locations of reported surface water flooding are indicated in (a) and (c) by double wavy lines.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

Table 1.

Flooding reports on 5 Jul 2012.

Table 1.

b. Case 2: 29 July 2012

A vigorous depression was centered over northeastern Scotland with showers covering the whole of the United Kingdom (Fig. 6a). From 1100 UTC, heavy showers developed over central and eastern England, associated with passage of a trough. Some of these showers developed into bands that propagated downwind. The heaviest showers (≥32 mm) were in East Anglia from 1300 to 1500 UTC (Fig. 6b). Despite the intensity of the observed showers, there was no flooding reported on this day (Flood Forecasting Centre 2012, unpublished report). The 1500 UTC forecast from the previous afternoon (Fig. 6c) correctly picked out the southern Hampshire to Essex track as the most intense, with a core probability of over 60% in the London area. It also had areas of 40% probability over northern Norfolk, matching the observed precipitation there, but additional 40% areas in northwestern England were not matched by observed storms. The 2100 UTC forecast had an even better match (not shown). The 0300 UTC forecast (not shown) was slightly less well oriented over southern England, but still picked up the southern Hampshire maximum at 60% and the area of intensification over East Anglia at 60%. All of the forecasts highlighted early development over the north coast of Cornwall and Devon during the morning, but it was the 0300 UTC run that correctly timed this as it moved east and then developed into the southern Hampshire to Essex band, with peak hourly probabilities of 40% over southern Hampshire from 1300 to 1400 UTC (Fig. 6d).

Fig. 6.
Fig. 6.

(a) The 0600–2100 UTC 29 Jul 2012 pointwise maximum of the hourly radar accumulations and (b) 1300–1400 UTC 29 Jul 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain during 0600–2100 UTC 29 Jul 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain during 1300–1400 UTC 29 Jul 2012 from the 0300 UTC run.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

c. Case 3: 9 July 2012

From 1200 to 2000 UTC, an isolated band of small intense showers developed along a quasi-stationary warm front lying across the northerly airstream associated with a depression over southern Scandinavia. The most intense showers were in Lancashire and local totals of ≥32 mm were recorded from 1300 to 1400 UTC (Fig. 7b). Serious surface water flooding was reported in parts of Yorkshire. Flooding was also reported in Wales and Dorset, but this was most likely due to fluvial flooding from earlier rainfall (see Table 2). The 1500 UTC forecast (Fig. 7c) had no area of probability exceeding 5% for this day, and only the 0300 UTC forecast (not shown) gave any indication of this development in the daily probabilities (at a probability of 10%). The hourly probabilities from the 0300 UTC run (Fig. 7d) indicated just a 5% likelihood in the vicinity of the observed location.

Fig. 7.
Fig. 7.

(a) The 0600–2100 UTC 9 Jul 2012 pointwise maximum of the hourly radar accumulations and (b) 1300–1400 UTC 9 Jul 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 9 Jul 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1300–1400 UTC 9 Jul 2012 from the 0300 UTC run. Locations of reported surface water flooding are indicated in (a) and (c) by double wavy lines.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

Table 2.

Flooding reports on 9 Jul 2012.

Table 2.

d. Case 4: 13 July 2012

A bent back occlusion in a broad trough was analyzed through the Midlands and Wales from a weak depression crossing southeastern England. A band of rain through northern Wales and the Midlands produced rain throughout the day without any intense cells. Around noon, a line of storms started to develop in southern Wales and then another in mid-Wales, possibly on the sea breeze. The mid-Wales storms quickly became intense, with peak hourly accumulations of more than 32 mm, and took on the character of a squall line (Figs. 8a,b). Flooding was reported widely across Cheshire, mid-Wales, and East Anglia (Table 3). The 15-h probabilities from the 1500 UTC run were low (Fig. 8c), generally having peak values of 20% across southern Wales, the southern Midlands, central southern England, and East Anglia. The 0300 UTC run (not shown) picked out the correct area of southern Wales for a local maximum probability, but without the extension northward into northern Wales, and still of only 20%. It also had another area of 40% probability in central southern England, where the showers did not intensify to the same extent. The hourly maps (Fig. 8d) show only the weakest of indications, at 5%–10% of the developments in Wales. The 0300 UTC run had the best attempt at the mid-Wales storms in the hourly maps for 1800–1900 UTC, but only at the 5% probability level. There was more consistency for the southern England band, which, however, was only observed to develop 8-mm hourly accumulations.

Fig. 8.
Fig. 8.

(a) The 0600–2100 UTC 13 Jul 2012 pointwise maximum of the hourly radar accumulations and (b) 1700–1800 UTC 13 Jul 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 13 Jul 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1700–1800 UTC 13 Jul 2012 from the 0300 UTC run. Locations of reported surface water flooding are indicated in (a) and (c) by double wavy lines.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

Table 3.

Flooding reports on 13 Jul 2012.

Table 3.

e. Case 5: 5 August 2012

A depression initially centered over the western part of the English Channel moved northeast during the day. Initially, storms developed in southern Scotland, with hourly totals in excess of 32 mm, associated with a trough moving north (Figs. 9a,b). Later storms on the tail of this trough over northeastern England also had ≥32 mm hourly totals. Farther south, storms developed in central southern and southwestern England that then propagated northeast, reaching ≥16 mm, with an isolated storm reaching ≥32 mm over Norfolk. An intense storm (≥32 mm) also developed in Cheshire near the Welsh border and moved northeast before decaying. A mix of surface and fluvial flooding was reported widely across southwestern, northwestern, and northeastern England; southern Wales; and Cheshire (see Table 4). The day probabilities from the 1500 UTC run (Fig. 9c) all captured the widespread nature of the storms, and especially the maximum (60%) in the corridor from southwestern England toward the northeast. All runs also had 40% probabilities in northern England and the northern Wales border, the latter location being the only forecast area where intense storms did not break out. Given the sensitivity of the Olympics in London, all runs suggested that the main area of risk would be to the west of London, with only a 20% probability over the city itself. Hourly probabilities correctly identified the enhanced risk over the northern Pennines in the early morning. In the south, the eastward propagation of enhanced probability from southwestern England is well captured, with peak probabilities of ≥20% in the 0300 UTC run (Fig. 9d).

Fig. 9.
Fig. 9.

(a) The 0600–2100 UTC 5 Aug 2012 pointwise maximum of the hourly radar accumulations and (b) 1400–1500 UTC 5 Aug 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 5 Aug 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1400–1500 UTC 5 Aug 2012 from the 0300 UTC run. Locations of reported surface water flooding are indicated in (a) and (c) by double wavy lines.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

Table 4.

Flooding reports on 5 Aug 2012.

Table 4.

f. Case 6: 17 August 2012

A waving quasi-stationary front extended from the south of Ireland to northeastern England, with heavy bursts of rain giving maximum hourly accumulations of more than 16 mm over the mountains of southern Wales (Figs. 10a,b). Flooding was reported in Cornwall, Devon, and in several parts of southern Wales, mostly from surface water (Table 5). The affected areas were well portrayed by the daily probabilities peaking at 90% in the 1500 UTC run (Fig. 10c) and 95% in the 2100 UTC run (not shown). The hourly probabilities from the 0300 UTC run (Fig. 10d) are much higher than in previous cases because of the widespread nature of the rainfall, reaching 40% and with the 20% line encompassing all affected areas.

Fig. 10.
Fig. 10.

(a) The 0600–2100 UTC 17 Aug 2012 pointwise maximum of the hourly radar accumulations and (b) 1100–1200 UTC 17 Aug 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 17 Aug 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1100–1200 UTC 17 Aug 2012 from the 0300 UTC run. Locations of reported surface water flooding are indicated in (a) and (c) by double wavy lines.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

Table 5.

Flooding reports on 17 Aug 2012.

Table 5.

g. Case 7: 24 August 2012

A depression was centered to the southwest of Ireland with an occluded front that moved east across southern England and Wales during the day. Apart from some late afternoon showers in northern Wales, which just reached 8-mm hourly accumulation, the main area of rain associated with the front reached Cornwall around 0400 UTC and then moved east along the south coast of England, leaving Kent at about 1900 UTC (Figs. 11a,b). Hourly accumulations in excess of 8 mm were observed out to sea, but not over land. No surface water flooding was reported on this day (Flood Forecasting Centre 2012, unpublished report). The forecasts all correctly identified the position and timing of the frontal rain area (Fig. 11c). The forecast probabilities (5%–10%) in northwestern Wales are slightly misplaced but give an indication of risk there. In contrast, probabilities of intense rainfall sometime during the day exceeded 60% along the south coast counties as far east as the Isle of Wight in all runs. In the hourly probability charts (Fig. 11d), all runs gave a core of 20% in the center of the rain area as it moved east.

Fig. 11.
Fig. 11.

(a) The 0600–2100 UTC 24 Aug 2012 pointwise maximum of the hourly radar accumulations and (b) 1500–1600 UTC 24 Aug 2012 hourly radar accumulation. (c) The 15-h probabilities of intense rain sometime during 0600–2100 UTC 24 Aug 2012 from the 1500 UTC run; (d) the 1-h probabilities of intense rain sometime during 1500–1600 UTC 24 Aug 2012 from the 0300 UTC run.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

Similar analyses (not shown) were performed on the remaining 70 cases. Figure 12a records summary results from the previous day’s 1500 UTC forecasts in calendar form. Each cell shows the maximum forecast probability of intense rainfall occurring somewhere in England and Wales during the period 0600–2100 UTC on that day. The 20 days on which an hourly accumulation in excess of 16 mm was observed sometime between 0600 and 2100 UTC and somewhere in England and Wales are filled in black. The 21 days on which surface water flooding was reported somewhere in England and Wales are marked with a dot. The maximum forecast probability over England and Wales is extracted for each forecast and placed in bins (0%–5%, 5%–20%, 20%–40%, 40%–60%, 60%–80%, 80%–95%, and 95%–100%). Figure 12b shows the frequency of each forecast probability bin, and Fig. 12c shows the forecast reliability for intense rainfall days and/or surface water flooding days. A reliable forecasting system will produce results lying on the 45° diagonal (the full line). The results shown in Fig. 12c are reasonably close to this, both for the days with intense rainfall (dotted line) and for the days with reported surface water flooding (dashed line). We may deduce that at broad spatial and time scales, the forecasts are reliable.

Fig. 12.
Fig. 12.

Summary of trial results for the occurrence or not of intense and/or flood-producing rainfall somewhere in England and Wales, sometime during each day. (a) Calendar of daily intense rainfall/flood occurrence and forecast probability: black (white) cells are days when hourly accumulations of more than 16 mm were (were not) observed by radar. The figures are the maximum forecast probability of exceeding 16 mm h−1 sometime during the day. A dot after the forecast symbol indicates that surface water flooding was reported. Gray cells are missing data. (b) Reliability plot of forecast probability of intense rain vs observed frequency of intense rain (dotted line)/observed frequency of surface water flooding (dashed line). Results from a perfect forecasting system would lie on the 45° diagonal (full line). (c) Frequency of forecast probability ranges.

Citation: Journal of Hydrometeorology 17, 5; 10.1175/JHM-D-15-0083.1

5. Discussion

Cases 1, 2, and 5 show widespread convective storms associated with a cyclonic weather regime and weak fronts. The forecasts show similar behavior, identifying the areas most at risk with higher probabilities, which generally become more accurately placed in subsequent forecasts. The probability level is also generally higher the more intense the observed rainfall.

Cases 3 and 4 show isolated storms along a frontal boundary. Here the forecasts also give a clear signal, but at a very low probability level, and in case 4, a misplaced location. The isolated nature of the storms has contributed to the low probabilities in both the ensemble and neighborhood processing.

Cases 6 and 7 show convective maxima embedded in stratiform frontal precipitation. Here the forecasts are less successful at locating the maxima and give extensive areas of high probability on occasions when the threshold was either not reached, or only locally over high terrain. Here the forecast rainfall has been too heavy and its extensive nature has increased the neighborhood fractional occurrence in each member, resulting in raised probabilities.

In all cases, the match of the highest probabilities to the observed areas of intense rainfall generally improved as the forecast length decreased. However, this did not happen uniformly in every area of intense rainfall. Given that the results here are from a downscaled ensemble, they illustrate the sensitive nature of the large-scale drivers of convective storms in the United Kingdom. Since we cannot expect convective-scale data assimilation to correct small synoptic-scale forecast errors at lead times beyond 12 h ahead, these results make a strong argument for combining successive runs in the probability postprocessing. They also emphasize the importance of further improvements to synoptic-scale data assimilation.

Overall, we see that there is a good match between areas of maximum probability and those areas where intense rainfall occurs. This may be partly a reflection of the nature of the rainfall during this period—extended periods of more isolated or more widespread precipitation might give different signals. Further work is needed to determine how to provide more useful guidance to users, whether by guidance on associating thresholds with weather types or through more sophisticated postprocessing. Such guidance or postprocessing will assume greater importance when applied to streamflow or surface water inundation probabilities, when the nature of the meteorological drivers will be less directly evident.

6. Conclusions

The individual case studies in sections 3 and 4 confirmed that the Met Office Unified Model, consistent with findings of other kilometer-scale modeling studies, is able to represent intense rainfall events that give rise to surface water flooding, causing damage and disruption. It reproduces the main characteristics of intense summer convection in the United Kingdom, including peak hourly accumulations at kilometer scale. Further model and data assimilation development can be expected to refine the spatial and temporal accuracy of the forecasts.

The probability forecasts for 0600–2100 UTC of the following day, summarized in Fig. 12, successfully identified both the days on which intense rain was observed and the days on which surface water flooding was observed with a high degree of reliability. In general, there was a good match between the areas of higher probability and the observations. Cases with large stratiform areas of rain, such as cases 6 and 7, appeared to be overforecast by the ensemble, while those, such as cases 3 and 4, with isolated convective storms appeared to be underforecast. The averaging effect of the neighborhood postprocessing contributes to this, and it is clear that further work remains to be done on the method of presentation of this information.

The individual hourly predictions were not assessed quantitatively, but generally showed some skill in the additional detail, consistent with the shorter lead time. The match between locations and times of peak probability and observed intense rainfall was mostly good. However, in general, the hourly probabilities were very low. While this is probably consistent with the true uncertainty in forecasts of such extreme intensities, it raises difficult challenges in communication that require further work, especially if such information is to be provided in public warnings.

The presentational changes introduced between the pilot case study and the 2012 trial produced a set of products that were easier to explain to a public audience and that retained smaller-scale information, while maintaining a coherent picture of the variations in probability. However, they resulted in very small probabilities of exceeding the threshold in a specific hour, which undoubtedly reduced the likelihood of appropriate decisions being taken. Both the success of the changes and the remaining weaknesses emphasize the need to work with end users in designing operational products of this sort.

For simplicity, a downscaling approach to ensemble generation was used in the case study and trial. As a result, only forecasts of more than 6 h ahead were presented. In the future, incorporation of high-resolution data assimilation and perturbations will enable use of the ensemble forecasts in the first 6 h, when most warnings are issued. This should also permit more spatial precision to be retained in the postprocessing.

The work presented in this paper forms part of an ongoing program of work aimed at improving operational surface water flood alert services in the United Kingdom. The results were of sufficient quality to implement the ensemble in operational use and to couple it to the grid-to-grid distributed rainfall–runoff model. Future work is aimed at calibrating the probabilistic runoff predictions to produce forecasts of flood inundation and impact that can be used more directly in the preparation of alerts.

Acknowledgments

Implementation of the Met Office’s convection-permitting ensemble prediction system is the culmination of work on convective-scale prediction over more than a decade and involving a huge number of scientists. We are grateful to all those involved in setting up the trial for the Olympic and Paralympic Games period, especially Clive Pierce, Stephen Moseley, and Andrew Bennet. Stephen Moseley produced the case study graphics used in this paper. The study of the 2011 Edinburgh case was originally carried out for the Scottish Environmental Protection Agency. We are very grateful to the joint Met Office/Environment Agency Flood Forecasting Centre staff for making available the flood impact evidence that was collected during the period of the 2012 trial. The final version of the paper has benefitted greatly from comments made by the reviewers, for which we are grateful.

REFERENCES

  • Baldauf, M., , Siefert A. , , Förstner J. , , Majewski D. , , Raschendorfer M. , , and Reinhardt T. , 2011: Operational convective-scale numerical weather prediction with the COSMO model: Description and sensitivities. Mon. Wea. Rev., 139, 38873905, doi:10.1175/MWR-D-10-05013.1.

    • Search Google Scholar
    • Export Citation
  • Barrett, A. I., , Gray S. L. , , Kirshbaum D. J. , , Roberts N. M. , , Schultz D. M. , , and Fairman J. G. Jr., 2014: Synoptic versus orographic control on stationary convective banding. Quart. J. Roy. Meteor. Soc., 141, 11011113, doi:10.1002/qj.2409.

    • Search Google Scholar
    • Export Citation
  • Bell, H., , and Tobin G. , 2007: Efficient and effective? The 100-year flood in the communication and perception of flood risk. Environ. Hazards, 7, 302311, doi:10.1016/j.envhaz.2007.08.004.

    • Search Google Scholar
    • Export Citation
  • Bowler, N. E., , Arribas A. , , Mylne K. R. , , Robertson K. B. , , and Beare S. E. , 2008: The MOGREPS short-range ensemble prediction system. Quart. J. Roy. Meteor. Soc., 134, 703722, doi:10.1002/qj.234.

    • Search Google Scholar
    • Export Citation
  • Bowler, N. E., ,Arribas A. , , Beare S. E. , , Mylne K. R. , , and Shutts G. J. , 2009: The local ETKF and SKEB: Upgrades to the MOGREPS short-range ensemble prediction system. Quart. J. Roy. Meteor. Soc., 135, 767776, doi:10.1002/qj.394.

    • Search Google Scholar
    • Export Citation
  • Buizza, R., , Houtekamer P. L. , , Toth Z. , , Pellerin G. , , Wei M. Z. , , and Zhu Y. J. , 2005: A comparison of the ECMWF, MSC, and NCEP global ensemble prediction systems. Mon. Wea. Rev., 133, 10761097, doi:10.1175/MWR2905.1.

    • Search Google Scholar
    • Export Citation
  • Burningham, K., , Fielding J. , , and Thrush D. , 2008: It’ll never happen to me: Understanding public awareness of local flood risk. Disasters, 32, 216238, doi:10.1111/j.1467-7717.2007.01036.x.

    • Search Google Scholar
    • Export Citation
  • CEH, 2011: UK Hydrological Bulletin January–July 2011. Centre for Ecology and Hydrology, accessed 28 March 2016. [Available online at http://nrfa.ceh.ac.uk/node/110/.]

  • Clark, A. J., , Gallus W. A. Jr., , Xue M. , , and Kong F. , 2009: A comparison of precipitation forecast skill between small convection-permitting and large convection-parameterizing ensembles. Wea. Forecasting, 24, 11211140, doi:10.1175/2009WAF2222222.1.

    • Search Google Scholar
    • Export Citation
  • Clark, A. J., and et al. , 2011: Probabilistic precipitation forecast skill as a function of ensemble size and spatial scale in a convection-allowing ensemble. Mon. Wea. Rev., 139, 14101418, doi:10.1175/2010MWR3624.1.

    • Search Google Scholar
    • Export Citation
  • Clark, A. J., , Weiss S. J. , , and Kain J. S. , 2012: An overview of the 2010 hazardous weather testbed experimental forecast program spring experiment. Bull. Amer. Meteor. Soc., 93, 5574, doi:10.1175/BAMS-D-11-00040.1.

    • Search Google Scholar
    • Export Citation
  • Clark, M. R., 2011: An observational study of the exceptional ‘Ottery St Mary’ thunderstorm of 30 October 2008. Meteor. Appl., 18, 137154, doi:10.1002/met.187.

    • Search Google Scholar
    • Export Citation
  • Davies, T., , Cullen M. J. P. , , Malcolm A. J. , , Mawson M. H. , , Staniforth A. , , White A. A. , , and Wood N. , 2005: A new dynamical core for the Met Office’s global and regional modelling of the atmosphere. Quart. J. Roy. Meteor. Soc., 131, 17591782, doi:10.1256/qj.04.101.

    • Search Google Scholar
    • Export Citation
  • Dolan, J. G., , and Iadarola S. , 2008: Risk communication formats for low probability events: An exploratory study of patient references. BMC Med. Inf. Decis. Making, 8, 14, doi:10.1186/1472-6947-8-14.

    • Search Google Scholar
    • Export Citation
  • Done, J., , Davis C. A. , , and Weisman M. L. , 2004: The next generation of NWP: Explicit forecasts of convection using the Weather Research and Forecasting (WRF). Atmos. Sci. Lett., 5, 110117, doi:10.1002/asl.72.

    • Search Google Scholar
    • Export Citation
  • Duan, Y., and et al. , 2012: An overview of Beijing 2008 Olympics Research and Development Project (B08RDP). Bull. Amer. Meteor. Soc., 93, 381403, doi:10.1175/BAMS-D-11-00115.1.

    • Search Google Scholar
    • Export Citation
  • Edwards, A., , Elwyn G. , , and Mulley A. , 2002: Explaining risks: Turning numerical data into meaningful pictures. BMJ, 324, 827830, doi:10.1136/bmj.324.7341.827.

    • Search Google Scholar
    • Export Citation
  • Edwards, J., , and Slingo A. , 1996: Studies with a flexible new radiation code. Part I: Choosing a configuration for a large-scale model. Quart. J. Roy. Meteor. Soc., 122, 689719, doi:10.1002/qj.49712253107.

    • Search Google Scholar
    • Export Citation
  • Essery, R. L. H., , Best M. J. , , Betts R. A. , , Cox P. M. , , and Taylor C. M. , 2003: Explicit representation of subgrid heterogeneity in a GCM land surface scheme. J. Hydrometeor., 4, 530543, doi:10.1175/1525-7541(2003)004<0530:EROSHI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Falconer, R. H., , Cobby D. , , Smyth P. , , Astle G. , , Dent J. , , and Golding B. , 2009: Pluvial flooding: New approaches in flood warning, mapping and risk management. J. Flood Risk Manage., 2, 198208, doi:10.1111/j.1753-318X.2009.01034.x.

    • Search Google Scholar
    • Export Citation
  • Faulkner, H., , Parker D. , , Green C. , , and Beven K. , 2007: Developing a translational discourse to communicate uncertainty in flood risk between science and the practitioner. Ambio, 36, 692704, doi:10.1579/0044-7447(2007)36[692:DATDTC]2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • FFC, 2015: About us. Flood Forecasting Centre, accessed 28 March 2016. [Available online at http://www.ffc-environment-agency.metoffice.gov.uk/about/.]

  • Gebhardt, C., , Theis S. E. , , Paulat M. , , and Ben Bouallegue Z. , 2011: Uncertainties in COSMO-DE precipitation forecasts introduced by model perturbations and variation of lateral boundaries. Atmos. Res., 100, 168177, doi:10.1016/j.atmosres.2010.12.008.

    • Search Google Scholar
    • Export Citation
  • Germann, U., , and Zawadzki I. , 2004: Scale dependence of the predictability of precipitation from continental radar images. Part II: Probability forecasts. J. Appl. Meteor., 43, 7489, doi:10.1175/1520-0450(2004)043<0074:SDOTPO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Gigerenzer, G., , Hertwig R. , , van den Broek E. , , Fasolo B. , , and Katsikopoulos K. V. , 2005: “A 30% chance of rain tomorrow”: How does the public understand probabilistic weather forecasts? Risk Anal., 25, 623629, doi:10.1111/j.1539-6924.2005.00608.x.

    • Search Google Scholar
    • Export Citation
  • Gill, J., 2008: Communicating forecast uncertainty for service providers. WMO Bull., 57 (4), 237243.

  • Golding, B. W., , Clark P. , , and May B. , 2005: The Boscastle flood: Meteorological analysis of the conditions leading to flooding on 16 August 2004. Weather, 60, 230235, doi:10.1256/wea.71.05.

    • Search Google Scholar
    • Export Citation
  • Golding, B. W., and et al. , 2014: Forecasting capabilities for the London 2012 Olympics. Bull. Amer. Meteor. Soc., 95, 883896, doi:10.1175/BAMS-D-13-00102.1.

    • Search Google Scholar
    • Export Citation
  • Goldman, R., , Parker D. , , Eaton C. , , Borkan J. , , Gramling R. , , Cover R. , , and Ahern D. , 2006: Patients’ perceptions of cholesterol, cardiovascular disease risk, and risk communication strategies. Ann. Fam. Med., 4, 205212, doi:10.1370/afm.534.

    • Search Google Scholar
    • Export Citation
  • Gregory, D., , and Rowntree P. R. , 1990: A mass flux convection scheme with representation of cloud ensemble characteristics and stability-dependent closure. Mon. Wea. Rev., 118, 14831506, doi:10.1175/1520-0493(1990)118<1483:AMFCSW>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Guha-Sapir, D., , Vos F. , , Below R. , , and Ponserre S. , 2012: Annual disaster statistical review 2011: The numbers and trends. Centre for Research on the Epidemiology of Disasters, 42 pp. [Available online at http://cred.be/sites/default/files/2012.07.05.ADSR_2011.pdf.]

  • Hand, W. H., , Fox N. I. , , and Collier C. G. , 2004: A study of twentieth-century extreme rainfall events in the United Kingdom with implications for forecasting. Meteor. Appl., 11, 1531, doi:10.1017/S1350482703001117.

    • Search Google Scholar
    • Export Citation
  • Hanley, K. E., , Kirshbaum D. J. , , Belcher S. E. , , Roberts N. M. , , and Leoncini G. , 2011: Ensemble predictability of an isolated mountain thunderstorm in a high-resolution model. Quart. J. Roy. Meteor. Soc., 137, 21242137, doi:10.1002/qj.877.

    • Search Google Scholar
    • Export Citation
  • Hanley, K. E., , Kirshbaum D. J. , , Roberts N. M. , , and Leoncini G. , 2013: Sensitivities of a squall line over central Europe in a convective-scale ensemble. Mon. Wea. Rev., 141, 112133, doi:10.1175/MWR-D-12-00013.1.

    • Search Google Scholar
    • Export Citation
  • Harrison, D., , Georgiou S. , , Gaussiat N. , , and Curtiss A. , 2014: Long-term diagnostics of precipitation estimates and the development of radar hardware monitoring within a radar product data quality management system. Hydrol. Sci. J., 59, 12771292, doi:10.1080/02626667.2013.841316.

    • Search Google Scholar
    • Export Citation
  • Hirahara, Y., , Ishida J. , , and Ishimizu T. , 2012: Trial operation of the local forecast model at JMA. Development of and studies with regional and smaller-scale atmospheric models, regional ensemble, monthly and seasonal forecasting, WMO/WCRP/WGNE Blue Book 2011, World Climate Research Programme, 11–12. [Available online at http://www.wcrp-climate.org/WGNE/BlueBook/2011/chapters/BB_11_S5.pdf.]

  • Hohenegger, C., , Walser A. , , Langhans W. , , and Schaer C. , 2008: Cloud-resolving ensemble simulations of the August 2005 Alpine flood. Quart. J. Roy. Meteor. Soc., 134, 889904, doi:10.1002/qj.252.

    • Search Google Scholar
    • Export Citation
  • Ikeda, K., , Steiner M. , , Pinto J. , , and Alexander C. , 2013: Evaluation of cold-season precipitation forecasts generated by the hourly updating high-resolution rapid refresh model. Wea. Forecasting, 28, 921939, doi:10.1175/WAF-D-12-00085.1.

    • Search Google Scholar
    • Export Citation
  • Joslyn, S., , and Savelli S. , 2010: Communicating forecast uncertainty: Public perception of weather forecast uncertainty. Meteor. Appl., 17, 180195, doi:10.1002/met.190.

    • Search Google Scholar
    • Export Citation
  • Joslyn, S., , Pak K. , , Jones D. , , Pyles J. , , and Hunt E. , 2007: The effect of probabilistic information on threshold forecasts. Wea. Forecasting, 22, 804812, doi:10.1175/WAF1020.1.

    • Search Google Scholar
    • Export Citation
  • Joslyn, S., , Nadav-Greenberg L. , , and Nichols R. M. , 2009: Probability of precipitation: Assessment and enhancement of end-user understanding. Bull. Amer. Meteor. Soc., 90, 185193, doi:10.1175/2008BAMS2509.1.

    • Search Google Scholar
    • Export Citation
  • Kitchen, M., , Brown R. , , and Davies A. G. , 1994: Real-time correction of weather radar data for the effects of bright band, range and orographic growth in widespread precipitation. Quart. J. Roy. Meteor. Soc., 120, 12311254, doi:10.1002/qj.49712051906.

    • Search Google Scholar
    • Export Citation
  • Kloprogge, P., , Van der Sluijs J. , , and Wardekker A. , 2007: Uncertainty communication: Issues and good practice. Rep. NWS-E-2007-199, Utrecht University, 60 pp. [Available online at http://www.nusap.net/downloads/reports/uncertainty_communication.pdf.]

  • Knapp, P., , Raynor D. , , and Berry D. , 2004: Comparison of two methods of presenting risk information to patients about the side effects of medicines. Qual. Saf. Health Care, 13, 176180, doi:10.1136/qshc.2003.009076.

    • Search Google Scholar
    • Export Citation
  • Kong, F. Y., , Droegemeier K. K. , , and Hickmon N. L. , 2007: Multiresolution ensemble forecasts of an observed tornadic thunderstorm system. Part II: Storm-scale experiments. Mon. Wea. Rev., 135, 759782, doi:10.1175/MWR3323.1.

    • Search Google Scholar
    • Export Citation
  • Kunii, M., and et al. , 2011: Verifications and intercomparisons of mesoscale ensemble prediction systems in B08RDP. Tellus, 63A, 531549, doi:10.1111/j.1600-0870.2011.00512.x.

    • Search Google Scholar
    • Export Citation
  • Kurz-Milcke, E., , Gigerenzer G. , , and Martignn L. , 2008: Transparency in risk communication: Graphical and analogue tools. Ann. N. Y. Acad. Sci., 1128, 1828, doi:10.1196/annals.1399.004.

    • Search Google Scholar
    • Export Citation
  • Lean, H. W., , Clark P. A. , , Dixon M. , , Roberts N. M. , , Fitch A. , , Forbes R. , , and Halliwell C. , 2008: Characteristics of high-resolution versions of the Met Office Unified Model for forecasting convection over the United Kingdom. Mon. Wea. Rev., 136, 34083424, doi:10.1175/2008MWR2332.1.

    • Search Google Scholar
    • Export Citation
  • Lean, H. W., , Roberts N. M. , , Clark P. A. , , and Morcrette C. , 2009: The surprising role of orography in the initiation of an isolated thunderstorm in southern England. Mon. Wea. Rev., 137, 30263046, doi:10.1175/2009MWR2743.1.

    • Search Google Scholar
    • Export Citation
  • Leoncini, G., , Plant R. S. , , Gray S. L. , , and Clark P. A. , 2010: Perturbation growth at the convective scale for CSIP IOP18. Quart. J. Roy. Meteor. Soc., 136, 653670, doi:10.1002/qj.587.

    • Search Google Scholar
    • Export Citation
  • Lock, A. P., 2001: The numerical representation of entrainment in parameterizations of boundary layer turbulent mixing. Mon. Wea. Rev., 129, 11481163, doi:10.1175/1520-0493(2001)129<1148:TNROEI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Lock, A. P., , Brown A. R. , , Bush M. R. , , Martin G. M. , , and Smith R. N. B. , 2000: A new boundary layer mixing scheme. Part I: Scheme description and single-column model tests. Mon. Wea. Rev., 128, 31873199, doi:10.1175/1520-0493(2000)128<3187:ANBLMS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Lorenc, A. C., and et al. , 2000: The Met. Office global three-dimensional variational data assimilation scheme. Quart. J. Roy. Meteor. Soc., 126, 29913012, doi:10.1002/qj.49712657002.

    • Search Google Scholar
    • Export Citation
  • Molteni, F., , Buizza R. , , Palmer T. N. , , and Petroliagis T. , 1996: The new ECMWF ensemble prediction system: Methodology and validation. Quart. J. Roy. Meteor. Soc., 122, 73119, doi:10.1002/qj.49712252905.

    • Search Google Scholar
    • Export Citation
  • Moore, R. J., , Cole S. J. , , Dunn S. , , Ghimire S. , , Golding B. W. , , Pierce C. E. , , Roberts N. M. , , and Speight L. , 2015: Surface water flood forecasting for urban communities. CREW Rep. CRW2012_03, 33 pp. [Available online at http://www.crew.ac.uk/publications/surface-water-flood-forecasting-urban-communities.]

  • Morss, R. E., , Demuth J. L. , , and Lazo J. K. , 2008: Communicating uncertainty in weather forecasts: A survey of the U.S. public. Wea. Forecasting, 23, 974991, doi:10.1175/2008WAF2007088.1.

    • Search Google Scholar
    • Export Citation
  • Murphy, A. H., , Lichtenstein S. , , Fischhoff B. , , and Winkler R. L. , 1980: Misinterpretations of precipitation probability forecasts. Bull. Amer. Meteor. Soc., 61, 695701, doi:10.1175/1520-0477(1980)061<0695:MOPPF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Paling, J., 2003: Strategies to help patients understand risks. BMJ, 327, 745, doi:10.1136/bmj.327.7417.745.

  • Peachey, J., , Schultz D. M. , , Morss R. , , Roebber P. , , and Wood R. , 2013: How forecasts expressing uncertainty are perceived by UK students. Weather, 68, 176181, doi:10.1002/wea.2094.

    • Search Google Scholar
    • Export Citation
  • Peralta, C., , Ben Bouallègue Z. , , Theis S. E. , , Gebhardt C. , , and Buchhold M. , 2012: Accounting for initial condition uncertainties in COSMO-DE-EPS. J. Geophys. Res., 117, D07108, doi:10.1029/2011JD016581.

    • Search Google Scholar
    • Export Citation
  • Peters, P., , Bostrom A. , , and Cutter S. , 2008: Perspectives on visualizing uncertainty in natural hazards. Risk Assessment, Modeling and Decision Support, A. Bostrom, S. French, and S. Gottlieb, Eds., Risk, Governance and Society, Vol. 14, Springer, 295–318.

  • Pitt, M., 2009: Learning lessons from the 2007 floods. UK Cabinet Office Rep., 462 pp.

  • Price, D., and et al. , 2012: Operational use of a grid-based model for flood forecasting. Proc. Inst. Civ. Eng.: Water Manage., 165, 6577, doi:10.1680/wama.2012.165.2.65.

    • Search Google Scholar
    • Export Citation
  • Price, M., , Cameron R. , , and Butow P. , 2007: Communicating risk information: The influence of graphical display format on quantitative information perception—Accuracy, comprehension and preferences. Patient Educ. Couns., 69, 121128, doi:10.1016/j.pec.2007.08.006.

    • Search Google Scholar
    • Export Citation
  • Roberts, N. M., 2000: The relationship between water vapour imagery and thunderstorms. Met Office Forecasting Research Tech. Rep. 110, 40 pp.

  • Roberts, N. M., , and Lean H. W. , 2008: Scale-selective verification of rainfall accumulations from high-resolution forecasts of convective events. Mon. Wea. Rev., 136, 7897, doi:10.1175/2007MWR2123.1.

    • Search Google Scholar
    • Export Citation
  • Roulston, M. S., , and Kaplan T. R. , 2009: A laboratory-based study of understanding of uncertainty in 5-day site-specific temperature forecasts. Meteor. Appl., 16, 237244, doi:10.1002/met.113.

    • Search Google Scholar
    • Export Citation
  • Roulston, M. S., , Bolton G. E. , , Kleit A. N. , , and Sears-Collins A. L. , 2006: A laboratory study of the benefits of including uncertainty information in weather forecasts. Wea. Forecasting, 21, 116122, doi:10.1175/WAF887.1.

    • Search Google Scholar
    • Export Citation
  • Scheuren, J.-M., , le Polain de Waroux O. , , Below R. , , Guha-Sapir D. , , and Ponserre S. , 2008: Annual disaster statistical review: The numbers and trends 2007. Center for Research on the Epidemiology of Disasters Rep., 64 pp. [Available online at http://www.unisdr.org/files/2796_CREDAnnualStatisticalReview2007.pdf.]

  • Schirillo, J., , and Stone E. , 2005: The greater ability of graphical versus numerical displays to increase risk avoidance involves a common mechanism. Risk Anal., 25, 555566, doi:10.1111/j.1539-6924.2005.00624.x.

    • Search Google Scholar
    • Export Citation
  • Seity, Y., , Brousseau P. , , Malardel S. , , Hello G. , , Bénard P. , , Bouttier F. , , Lac C. , , and Masson V. , 2011: The AROME-France Convective-Scale Operational Model. Mon. Wea. Rev., 139, 976991, doi:10.1175/2010MWR3425.1.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., , Brooks H. E. , , Du J. , , Tracton M. S. , , and Rogers E. , 1999: Using ensembles for short-range forecasting. Mon. Wea. Rev., 127, 433446, doi:10.1175/1520-0493(1999)127<0433:UEFSRF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Stephens, E., , Mylne K. , , and Spiegelhalter D. , 2011: Using an online game to evaluate effective methods of communicating ensemble model output to different audiences. 2011 Fall Meeting, San Francisco, CA, Amer. Geophys. Union, Abstract ED33B-0776.

  • Steppeler, J., , Doms G. , , Schattler U. , , Bitzer H. W. , , Gassmann A. , , Damrath U. , , and Gregoric G. , 2003: Meso-gamma scale forecasts using the nonhydrostatic model LM. Meteor. Atmos. Phys., 82, 7596, doi:10.1007/s00703-001-0592-9.

    • Search Google Scholar
    • Export Citation
  • Stone, E., , Yates J. , , and Parker A. , 1997: Effects of numerical and graphical displays on professed risk-taking behaviour. J. Exp. Psychol., 3 (4), 243256.

    • Search Google Scholar
    • Export Citation
  • Tang, Y., , Lean H. W. , , and Bornemann J. , 2013: The benefits of the Met Office variable resolution NWP model for forecasting convection. Meteor. Appl., 20, 417426, doi:10.1002/met.1300.

    • Search Google Scholar
    • Export Citation
  • Tapsell, S., , Burton R. , , Oakes S. , , and Parker D. , 2005: The social performance of flood warning communications technologies. R&D Tech. Summary W5C-016/TS, Environment Agency, 3 pp. [Available online at https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/290685/scho0505bjdz-e-e.pdf.]

  • Tennant, W., 2015: Improving initial condition perturbations for MOGREPS-UK. Quart. J. Roy. Meteor. Soc., 141, 23242336, doi:10.1002/qj.2524.

    • Search Google Scholar
    • Export Citation
  • Tennant, W., , and Beare S. , 2014: New schemes to perturb sea-surface temperature and soil moisture content in MOGREPS. Quart. J. Roy. Meteor. Soc., 140, 11501160, doi:10.1002/qj.2202.

    • Search Google Scholar
    • Export Citation
  • Theis, S. E., , Hense A. , , and Damrath U. , 2005: Probabilistic precipitation forecasts from a deterministic model: A pragmatic approach. Meteor. Appl., 12, 257268, doi:10.1017/S1350482705001763.

    • Search Google Scholar
    • Export Citation
  • Thrush, D., , Burningham K. , , and Fielding J. , 2005: Flood warning for vulnerable groups: A qualitative study. R&D Tech. Rep. W5C-018/3, Environment Agency, 73 pp. [Available online at http://www.envirobase.info/PDF/3557_1.pdf.]

  • Tracton, M. S., , and Kalnay E. , 1993: Operational ensemble prediction at the National Meteorological Center: Practical aspects. Wea. Forecasting, 8, 379398, doi:10.1175/1520-0434(1993)008<0379:OEPATN>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Tracton, M. S., , Du J. , , Toth Z. , , and Juang H. , 1998: Short-Range Ensemble Forecasting (SREF) at NCEP/EMC. Proc. 12th Conf. on Numerical Weather Prediction, Phoenix, AZ, Amer. Meteor. Soc., 269–272.

  • Twigger-Ross, C., , Fernandez-Bilbao A. , , Tapsell S. , , Walker G. , , Watson N. , , Deeming H. , , and Kashefi E. , 2008: Improving flood warnings: Final report. Improving institutional and social responses to flooding, Science Rep. SC060019, Work Package 1a, Environment Agency, 76 pp.

  • Vié, B., , Nuissier O. , , and Ducrocq V. , 2011: Cloud-resolving ensemble simulations of Mediterranean heavy precipitating events: Uncertainty on initial conditions and lateral boundary conditions. Mon. Wea. Rev., 139, 403423, doi:10.1175/2010MWR3487.1.

    • Search Google Scholar
    • Export Citation
  • Vié, B., , Molinié G. , , Nuissier O. , , Vincendon B. , , Ducrocq V. , , Bouttier F. , , and Richard E. , 2012: Hydrometeorological evaluation of a convection-permitting ensemble prediction system for Mediterranean heavy precipitating events. Nat. Hazards Earth Syst. Sci., 12, 26312645, doi:10.5194/nhess-12-2631-2012.

    • Search Google Scholar
    • Export Citation
  • Walser, A., , Luthi D. , , and Schar C. , 2004: Predictability of precipitation in a cloud-resolving model. Mon. Wea. Rev., 132, 560577, doi:10.1175/1520-0493(2004)132<0560:POPIAC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Waters, E., , Weinstein N. , , Colditz G. , , and Emmons K. , 2006: Formats for improving risk communication in medical trade off decisions. J. Health Commun., 11, 167182, doi:10.1080/10810730500526695.

    • Search Google Scholar
    • Export Citation
  • Wedawatta, G., , Ingirige B. , , and Proverbs D. , 2014: Small businesses and flood impacts: The case of the 2009 flood event in Cockermouth. J. Flood Risk Manage., 7, 4253, doi:10.1111/jfr3.12031.

    • Search Google Scholar
    • Export Citation
  • Weusthoff, T., , Ament F. , , and Arpagaus M. , 2010: Assessing the benefits of convection-permitting models by neighborhood verification: Examples from MAP D-PHASE. Mon. Wea. Rev., 138, 34183433, doi:10.1175/2010MWR3380.1.

    • Search Google Scholar
    • Export Citation
  • Wikipedia, 2015: Hootsuite. Accessed 28 March 2016. [Available online at https://en.wikipedia.org/wiki/Hootsuite.]

  • Wilson, D., , and Ballard S. P. , 1999: A microphysically based precipitation scheme for the UK Meteorological Office Unified Model. Quart. J. Roy. Meteor. Soc., 125, 16071636, doi:10.1002/qj.49712555707.

    • Search Google Scholar
    • Export Citation
  • Yamagishi, I., 1997: When a 12.86% mortality is more dangerous than 24.14%: Implications for risk communication. Appl. Cognit. Psychol., 11, 495506, doi:10.1002/(SICI)1099-0720(199712)11:6<495::AID-ACP481>3.0.CO;2-J.

    • Search Google Scholar
    • Export Citation
Save