• Aksoy, A., , D. Dowell, , and C. Snyder, 2009: A multicase comparative assessment of the ensemble Kalman filter for assimilation of radar observations. Part I: Storm-scale analyses. Mon. Wea. Rev., 137, 18051824, doi:10.1175/2008MWR2691.1.

    • Search Google Scholar
    • Export Citation
  • Aksoy, A., , D. Dowell, , and C. Snyder, 2010: A multicase comparative assessment of the ensemble Kalman filter for assimilation of radar observations. Part II: Short-range ensemble forecasts. Mon. Wea. Rev., 138, 12731292, doi:10.1175/2009MWR3086.1.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., 2001: An ensemble adjustment filter for data assimilation. Mon. Wea. Rev., 129, 28842903, doi:10.1175/1520-0493(2001)129<2884:AEAKFF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., 2007: An adaptive covariance inflation error correction algorithm for ensemble filters. Tellus, 59A, 210224, doi:10.1111/j.1600-0870.2006.00216.x.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., 2009: Spatially and temporally varying adaptive covariance inflation for ensemble filters. Tellus, 61A, 7283, doi:10.1111/j.1600-0870.2008.00361.x.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., , and N. Collins, 2007: Scalable implementations of ensemble filter algorithms for data assimilation. J. Atmos. Oceanic Technol., 24, 14521463, doi:10.1175/JTECH2049.1.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., , T. Hoar, , K. Raeder, , H. Liu, , N. Collins, , R. Torn, , and A. Avellano, 2009: The Data Assimilation Research Testbed: A community facility. Bull. Amer. Meteor. Soc., 90, 12831296, doi:10.1175/2009BAMS2618.1.

    • Search Google Scholar
    • Export Citation
  • Carrio, G. G., , W. R. Cotton, , D. Zupanski, , and M. Zupanski, 2008: Development of an aerosol retrieval method: Description and preliminary tests. J. Appl. Meteor. Climatol., 47, 27602776, doi:10.1175/2008JAMC1729.1.

    • Search Google Scholar
    • Export Citation
  • Chaboureau, J.-P., , and J.-P. Pinty, 2006: Validation of a cirrus parameterization with Meteosat second generation observations. Geophys. Res. Lett., 33, L03815, doi:10.1029/2005GL024725.

    • Search Google Scholar
    • Export Citation
  • Chang, F.-L., , P. Minnis, , B. Lin, , M. Khaiyer, , R. Palikonda, , and D. Spangenberg, 2010: A modified method for inferring cloud top height using GOES 12 imager 10.7 and 13.3 μm data. J. Geophys. Res., 115, D06208, doi:10.1029/2009JD012304.

    • Search Google Scholar
    • Export Citation
  • Chen, Y., , F. Weng, , Y. Han, , and Q. Liu, 2008: Validation of the Community Radiative Transfer Model by using CloudSat data. J. Geophys. Res.,113, D00A03, doi:10.1029/2007JD009561.

  • Crum, T. D., , and R. L. Alberty, 1993: The WSR-88D and the WSR-88D Operational Support Facility. Bull. Amer. Meteor. Soc., 74, 16691687, doi:10.1175/1520-0477(1993)074<1669:TWATWO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Dawson, D. T., II, , L. J. Wicker, , E. R. Mansell, , and R. L. Tanamachi, 2012: Impact of the environmental low-level wind profile on ensemble forecasts of the 4 May 2007 Greensburg, Kansas, tornadic storm and associated mesocyclones. Mon. Wea. Rev., 140, 696716, doi:10.1175/MWR-D-11-00008.1.

    • Search Google Scholar
    • Export Citation
  • Derber, J. C., , and W.-S. Wu, 1998: The use of TOVS cloud-cleared radiances in the NCEP SSI analysis system. Mon. Wea. Rev.,126, 2287–2299, doi:10.1175/1520-0493(1998)126<2287:TUOTCC>2.0.CO;2.

  • Dowell, D. C., , and L. J. Wicker, 2009: Additive noise for storm-scale ensemble forecasting and data assimilation. J. Atmos. Oceanic Technol., 26, 911927, doi:10.1175/2008JTECHA1156.1.

    • Search Google Scholar
    • Export Citation
  • Dowell, D. C., , F. Zhang, , L. J. Wicker, , C. Snyder, , and N. A. Crook, 2004: Wind and temperature retrievals in the 17 May 1981 Arcadia, Oklahoma, supercell: Ensemble Kalman filter experiments. Mon. Wea. Rev., 132, 19822005, doi:10.1175/1520-0493(2004)132<1982:WATRIT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Dowell, D. C., , L. J. Wicker, , and C. Snyder, 2011: Ensemble Kalman filter assimilation of radar observations of the 8 May 2003 Oklahoma City supercell: Influences of reflectivity observations on storm-scale analyses. Mon. Wea. Rev., 139, 272294, doi:10.1175/2010MWR3438.1.

    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46, 30773107, doi:10.1175/1520-0469(1989)046<3077:NSOCOD>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Eilts, M. D., , and S. D. Smith, 1990: Efficient dealiasing of Doppler velocities using local environment constraints. J. Atmos. Oceanic Technol., 7, 118128, doi:10.1175/1520-0426(1990)007<0118:EDODVU>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Ek, M. B., , K. E. Mitchell, , Y. Lin, , E. Rogers, , P. Grunmann, , V. Koren, , G. Gayno, , and J. D. Tarpley, 2003: Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model. J. Geophys. Res., 108, 8851, doi:10.1029/2002JD003296.

    • Search Google Scholar
    • Export Citation
  • Errico, R. M., 2000: The dynamical balance of leading singular vectors in a primitive-equation model. Quart. J. Roy. Meteor. Soc.,126, 1601–1618, doi:10.1002/qj.49712656603.

  • Fujita, T., , D. J. Stensrud, , and D. C. Dowell, 2007: Surface data assimilation using an ensemble Kalman filter approach with initial condition and model physics uncertainties. Mon. Wea. Rev., 135, 18461868, doi:10.1175/MWR3391.1.

    • Search Google Scholar
    • Export Citation
  • Gao, J., , and D. J. Stensrud, 2012: Assimilation of reflectivity data in a convective-scale, cycled 3DVAR framework with hydrometeor classification. J. Atmos. Sci., 69, 10541065, doi:10.1175/JAS-D-11-0162.1.

    • Search Google Scholar
    • Export Citation
  • Gao, J., , M. Xue, , K. Brewster, , and K. K. Droegemeier, 2004: A three-dimensional variational data assimilation method with recursive filter for single-Doppler radar. J. Atmos. Oceanic Technol., 21, 457469, doi:10.1175/1520-0426(2004)021<0457:ATVDAM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Gaspari, G., , and S. E. Cohn, 1999: Construction of correlation functions in two and three dimensions. Quart. J. Roy. Meteor. Soc., 125, 723757, doi:10.1002/qj.49712555417.

    • Search Google Scholar
    • Export Citation
  • Grasso, L. D., , and T. Greenwald, 2004: Analysis of 10.7 μm brightness temperatures of a simulated thunderstorm with two-moment microphysics. Mon. Wea. Rev., 132, 815825, doi:10.1175/1520-0493(2004)132<0815:AOMBTO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Grell, G. A., , and D. Devenyi, 2002: A generalized approach to parameterizing convection combining ensemble and data assimilation techniques. Geophys. Res. Lett.,29, 1693, doi:10.1029/2002GL015311.

  • Hong, S.-Y., , Y. Noh, , and J. Dudhia, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 23182341, doi:10.1175/MWR3199.1.

    • Search Google Scholar
    • Export Citation
  • Iacono, M. J., , J. S. Delamere, , E. J. Mlawer, , M. W. Shephard, , S. A. Clough, , and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, doi:10.1029/2008JD009944.

  • Janjić, Z. I., 1994: The step-mountain eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev.,122, 927–945, doi:10.1175/1520-0493(1994)122<0927:TSMECM>2.0.CO;2.

  • Jones, T. A., , J. Otkin, , D. J. Stensrud, , and K. Knopfmeier, 2013a: Assimilation of satellite infrared radiances and Doppler radar observations during a cool season Observing System Simulation Experiment. Mon. Wea. Rev., 141, 32733299, doi:10.1175/MWR-D-12-00267.1.

    • Search Google Scholar
    • Export Citation
  • Jones, T. A., , D. J. Stensrud, , P. Minnis, , and R. Palikonda, 2013b: Evaluation of a forward operator to assimilate cloud water path into WRF-DART. Mon. Wea. Rev., 141, 22722289, doi:10.1175/MWR-D-12-00238.1.

    • Search Google Scholar
    • Export Citation
  • Jones, T. A., , J. Otkin, , D. J. Stensrud, , and K. Knopfmeier, 2014: Forecast evaluation of an Observing System Simulation Experiment assimilating both radar and satellite data. Mon. Wea. Rev., 142, 107124, doi:10.1175/MWR-D-13-00151.1.

    • Search Google Scholar
    • Export Citation
  • Jung, Y., , M. Xue, , and M. Tong, 2012: Ensemble Kalman filter analyses of the 29–30 May 2004 Oklahoma tornadic thunderstorm using one- and two-moment bulk microphysics schemes, with verification against polarimetric radar data. Mon. Wea. Rev., 140, 14571475, doi:10.1175/MWR-D-11-00032.1.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., , and J. M. Fritsch, 1993: Convective parameterization for mesoscale models: The Kain–Fritsch scheme. The Representation of Cumulus Convection in Numerical Models, Meteor. Monogr., No. 24, Amer. Meteor. Soc., 165–170.

  • Kain, J. S., , S. R. Dembek, , S. J. Weiss, , J. L. Case, , J. J. Levit, , and R. A. Sobash, 2010: Extracting unique information from high-resolution forecast models: Monitoring selected fields and phenomena every time step. Wea. Forecasting, 25, 15361542, doi:10.1175/2010WAF2222430.1.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., and et al. , 2013: A feasibility study for probabilistic convection initiation forecasts based on explicit numerical guidance. Bull. Amer. Meteor. Soc., 94, 12131225, doi:10.1175/BAMS-D-11-00264.1.

    • Search Google Scholar
    • Export Citation
  • Kalman, R. E., 1960: A new approach to linear filtering and prediction problems. J. Basic Eng., 82, 3545, doi:10.1115/1.3662552.

  • Keil, C., , A. Tafferner, , H. Mannstein, , and U. Schättler, 2003: Evaluating high-resolution model forecasts of European winter storms by use of satellite and radar observations. Wea. Forecasting, 18, 732747, doi:10.1175/1520-0434(2003)018<0732:EHMFOE>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Lakshmanan, V., , K. Hondl, , G. Stumpf, , and T. Smith, 2003: Quality control of WSR-88D data. 31st Radar Conf., Seattle, WA, Amer. Meteor. Soc., 7.1. [Available online at https://ams.confex.com/ams/pdfpapers/62597.pdf.]

  • Lin, Y.-L., , R. D. Farley, , and H. D. Orville, 1983: Bulk parameterization of the snow field in a cloud model. J. Climate Appl. Meteor.,22, 1065–1092, doi:10.1175/1520-0450(1983)022<1065:BPOTSF>2.0.CO;2.

  • Lu, H., , and Q. Xu, 2009: Trade-offs between observation accuracy and resolutions in configuring phased-array radar velocity scans for ensemble-based storm-scale data assimilation. J. Appl. Meteor. Climatol., 48, 12301244, doi:10.1175/2008JAMC2009.1.

    • Search Google Scholar
    • Export Citation
  • Majcen, M., , P. Markowski, , Y. Richardson, , D. Dowell, , and J. Wurman, 2008: Multipass objective analyses of Doppler radar data. J. Atmos. Oceanic Technol., 25, 18451858, doi:10.1175/2008JTECHA1089.1.

    • Search Google Scholar
    • Export Citation
  • Mansell, E. R., 2008: EnKF analysis and forecast predictability of a tornadic supercell storm. 24th Conf. on Severe Local Storms, Savannah, GA, Amer. Meteor. Soc., P5.2. [Available online at https://ams.confex.com/ams/24SLS/techprogram/paper_142059.htm.]

  • McNally, A. P., , P. D. Watts, , J. A. Smith, , R. Engelen, , G. A. Kelly, , J. N. Thepaut, , and M. Matricardi, 2006: The assimilation of AIRS radiance data at ECMWF. Quart. J. Roy. Meteor. Soc.,132, 935–957, doi:10.1256/qj.04.171.

  • McPherson, R. A., and et al. , 2007: Statewide monitoring of the mesoscale environment: A technical update on the Oklahoma Mesonet. J. Atmos. Oceanic Technol., 24, 301321, doi:10.1175/JTECH1976.1.

    • Search Google Scholar
    • Export Citation
  • Mecikalski, J. R., , P. Minnis, , and R. Palikonda, 2013: Use of satellite derived cloud properties to quantify growing cumulus beneath cirrus clouds. Atmos. Res., 120–121, 192201, doi:10.1016/j.atmosres.2012.08.017.

    • Search Google Scholar
    • Export Citation
  • Menzel, W. P., , and J. F. Purdom, 1994: Introducing GOES-I: The first of a new generation of geostationary operational environmental satellites. Bull. Amer. Meteor. Soc., 75, 757781, doi:10.1175/1520-0477(1994)075<0757:IGITFO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Migliorini, S., 2012: On the equivalence between radiance and retrieval assimilation. Mon. Wea. Rev.,140, 258–265, doi:10.1175/MWR-D-10-05047.1.

  • Milbrandt, J. A., , and M. K. Yau, 2005a: A multimoment bulk microphysics parameterization. Part I: Analysis of the role of the spectral shape parameter. J. Atmos. Sci., 62, 30513064, doi:10.1175/JAS3534.1.

    • Search Google Scholar
    • Export Citation
  • Milbrandt, J. A., , and M. K. Yau, 2005b: A multimoment bulk microphysics parameterization. Part II: A proposed three-moment closure and scheme description. J. Atmos. Sci., 62, 30653081, doi:10.1175/JAS3535.1.

    • Search Google Scholar
    • Export Citation
  • Minnis, P., , J. Huang, , B. Lin, , Y. Yi, , R. F. Arduini, , T.-F. Fan, , J. K. Ayers, , and G. G. Mace, 2007: Ice cloud properties in ice-over-water cloud systems using Tropical Rainfall Measuring Mission (TRMM) visible and infrared scanner and TRMM Microwave Imager data. J. Geophys. Res., 112, D06206, doi:10.1029/2006JD007626.

    • Search Google Scholar
    • Export Citation
  • Minnis, P., and et al. , 2008a: Near-real time cloud retrievals from operational and research meteorological satellites. Remote Sensing of Clouds and the Atmosphere XIII, R. H. Picard et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 7107), 710703, doi:10.1117/12.800344.

  • Minnis, P., and et al. , 2008b: Cloud detection in non-polar regions for CERES using TRMM VIRS and Terra and Aqua MODIS data. IEEE Trans. Geosci. Remote Sens., 46, 38573884, doi:10.1109/TGRS.2008.2001351.

    • Search Google Scholar
    • Export Citation
  • Minnis, P., and et al. , 2011: CERES Edition-2 cloud property retrievals using TRMM VIRS and Terra and Aqua MODIS data, Part I: Algorithms. IEEE Trans. Geosci. Remote Sens., 49, 43744400, doi:10.1109/TGRS.2011.2144601.

    • Search Google Scholar
    • Export Citation
  • Mlawer, E. J., , S. J. Taubman, , P. D. Brown, , M. J. Iacono, , and S. A. Clough, 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102, 16 66316 682, doi:10.1029/97JD00237.

    • Search Google Scholar
    • Export Citation
  • Nakanishi, M., , and H. Niino, 2006: An improved Mellor–Yamada level-3 model: Its numerical stability and application to a regional prediction of advection fog. Bound.-Layer Meteor., 119, 397407, doi:10.1007/s10546-005-9030-8.

    • Search Google Scholar
    • Export Citation
  • Otkin, J. A., 2010: Clear and cloudy sky infrared brightness temperature assimilation using an ensemble Kalman filter. J. Geophys. Res., 115, D19207, doi:10.1029/2009JD013759.

    • Search Google Scholar
    • Export Citation
  • Otkin, J. A., , and T. Greenwald, 2008: Comparison of WRF model-simulated MODIS-derived cloud data. Mon. Wea. Rev., 136, 19571970, doi:10.1175/2007MWR2293.1.

    • Search Google Scholar
    • Export Citation
  • Pincus, R., , R. J. P. Hofmann, , J. L. Anderson, , K. Raeder, , N. Collins, , and J. S. Whitaker, 2011: Can fully accounting for clouds in data assimilation improve short-term forecast by global models? Mon. Wea. Rev., 139, 946956, doi:10.1175/2010MWR3412.1.

    • Search Google Scholar
    • Export Citation
  • Polkinghorne, R., , and T. Vukicevic, 2011: Data assimilation of cloud affected radiances in a cloud resolving model. Mon. Wea. Rev., 139, 755773, doi:10.1175/2010MWR3360.1.

    • Search Google Scholar
    • Export Citation
  • Polkinghorne, R., , T. Vukicevic, , and F. Evans, 2010: Validation of cloud resolving model background data for data assimilation. Mon. Wea. Rev., 138, 781795, doi:10.1175/2009MWR3012.1.

    • Search Google Scholar
    • Export Citation
  • Schmit, T. J., , E. M. Prins, , A. J. Schreiner, , and J. J. Gurka, 2001: Introducing the GOES-M Imager. Natl. Wea. Dig., 25, 2837. [Available online at www.nwas.org/digest/papers/2001/Vol25No34/Pg28-Schmit.pdf.]

    • Search Google Scholar
    • Export Citation
  • Skamarock, W. C., and et al. , 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp., doi:10.5065/D68S4MVH.

  • Snyder, C., , and F. Zhang, 2003: Assimilation of simulated Doppler radar observations with an ensemble Kalman filter. Mon. Wea. Rev., 131, 16631677, doi:10.1175//2555.1.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., , J.-W. Bao, , and T. T. Warner, 2000: Using initial condition and model physics perturbations in short-range ensemble simulations of mesoscale convective systems. Mon. Wea. Rev.,128, 2077–2107, doi:10.1175/1520-0493(2000)128<2077:UICAMP>2.0.CO;2.

  • Stensrud, D. J., and et al. , 2009: Convective-scale warn-on-forecast system: A vision for 2020. Bull. Amer. Meteor. Soc., 90, 14871499, doi:10.1175/2009BAMS2795.1.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., and et al. , 2013: Progress and challenges with warn-on-forecast. Atmos. Res., 123, 216, doi:10.1016/j.atmosres.2012.04.004.

    • Search Google Scholar
    • Export Citation
  • Thompson, G., , R. M. Rasmussen, , and K. Manning, 2004: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part I: Description and sensitivity analysis. Mon. Wea. Rev., 132, 519542, doi:10.1175/1520-0493(2004)132<0519:EFOWPU>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Thompson, G., , P. R. Field, , R. M. Rasmussen, , and W. R. Hall, 2008: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part II: Implementation of a new snow parameterization. Mon. Wea. Rev., 136, 50955115, doi:10.1175/2008MWR2387.1.

    • Search Google Scholar
    • Export Citation
  • Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117, 17791800, doi:10.1175/1520-0493(1989)117<1779:ACMFSF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Tong, M., , and M. Xue, 2005: Ensemble Kalman filter assimilation of Doppler radar data with a compressible nonhydrostatic model: OSS experiments. Mon. Wea. Rev., 133, 17891807, doi:10.1175/MWR2898.1.

    • Search Google Scholar
    • Export Citation
  • Vukicevic, T., , T. Greenwald, , M. Zupanski, , D. Zupanski, , T. Vonder Haar, , and A. Jones, 2004: Mesoscale cloud state estimation from visible and infrared satellite radiances. Mon. Wea. Rev., 132, 30663077, doi:10.1175/MWR2837.1.

    • Search Google Scholar
    • Export Citation
  • Vukicevic, T., , M. Sengupta, , A. S. Jones, , and T. Vonder Haar, 2006: Cloud-resolving satellite data assimilation: Information content of IR window observations and uncertainties in estimation. J. Atmos. Sci., 63, 901919, doi:10.1175/JAS3639.1.

    • Search Google Scholar
    • Export Citation
  • Wang, C., , and X. Huang, 2014: Parallax correction in the analysis of multiple satellite data sets. IEEE Geosci. Remote Sens. Lett., 11, 965969, doi:10.1109/LGRS.2013.2283573.

    • Search Google Scholar
    • Export Citation
  • Wheatley, D. M., , and D. J. Stensrud, 2010: The impact of assimilating surface pressure observations on severe weather events in a WRF mesoscale ensemble system. Mon. Wea. Rev.,138, 1673–1694, doi:10.1175/2009MWR3042.1.

  • Wilks, D. S., 2006: Statistical Methods in the Atmospheric Sciences. 2nd ed. Academic Press, 627 pp.

  • Yussouf, N., , and D. J. Stensrud, 2010: Impact of phased-array radar observations over a short assimilation period: Observing system simulation experiments using an ensemble Kalman filter. Mon. Wea. Rev., 138, 517538, doi:10.1175/2009MWR2925.1.

    • Search Google Scholar
    • Export Citation
  • Yussouf, N., , E. R. Mansell, , L. J. Wicker, , D. M. Wheatley, , and D. J. Stensrud, 2013: The ensemble Kalman filter analyses and forecasts of the 8 May 2003 Oklahoma City tornadic supercell storm using single- and double-moment microphysics schemes. Mon. Wea. Rev., 141, 33883412, doi:10.1175/MWR-D-12-00237.1.

    • Search Google Scholar
    • Export Citation
  • Zhang, F., , C. Snyder, , and J. Sun, 2004: Impacts of initial estimate and observation availability on convective-scale data assimilation with an ensemble Kalman filter. Mon. Wea. Rev., 132, 12381253, doi:10.1175/1520-0493(2004)132<1238:IOIEAO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Zhang, S. Q., , M. Zupanski, , A. Y. Hou, , X. Lin, , and S. H. Cheung, 2013: Assimilation of precipitation-affected radiances in a cloud-resolving WRF ensemble data assimilation system. Mon. Wea. Rev., 141, 754772, doi:10.1175/MWR-D-12-00055.1.

    • Search Google Scholar
    • Export Citation
  • Zupanski, D., , S. Q. Zhang, , M. Zupanski, , A. Y. Hou, , and S. H. Cheung, 2011: A prototype WRF-based ensemble data assimilation system for dynamically downscaling satellite precipitation observations. J. Hydrometeor., 12, 118134, doi:10.1175/2010JHM1271.1.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    (a) The 1800 UTC KOUN radiosonde showing vertical profiles of temperature, dewpoint, wind speed, and direction. (b) Corresponding KOUN hodograph at 1800 UTC. (c) Ensemble mean 2-m surface temperature at 1800 UTC over west-central OK. (d) Ensemble mean 2-m relative humidity at 1800 UTC with a 40% contour indicating approximate location of the dryline at this time.

  • View in gallery

    GOES-13 6.7-μm water vapor channel with WSR-88D reflectivity overlaid at (a) 1900, (b) 1930, (c) 2000, and (d) 2030 UTC 24 May 2011. Satellite data have been parallax corrected. Reflectivity data are within a black outline, outside of which satellite brightness temperatures are shown.

  • View in gallery

    Map showing the background 15-km model domain (labeled “1”), inner 3-km domain (labeled “2”), and 3-km experiment domain (labeled “3”). Locations for the three WSR-88Ds used in this research (KFDR, KTLX, KVNX) are also shown.

  • View in gallery

    Observation diagnostics including bias, RMSI, CR, and number of observations assimilated for (a),(b) radar reflectivity; (c),(d) radial velocity; (e),(f) LWP; and (g),(h) IWP generated from the PATHRAD experiment.

  • View in gallery

    (a) GOES-13 CWP retrievals, and ensemble mean simulated CWP from (b) CNTL, (c) PATH, (d) RADP, (e) RAD0, and (f) PATHRAD experiments analyzed at 1915 UTC. Black contours in (b)–(f) indicate regions of where GOES-13 retrievals are CWP > 1.0 kg m−2.

  • View in gallery

    (a) WSR-88D reflectivity at 4 km, and ensemble mean simulated reflectivity from (b) CNTL, (c) PATH, (d) RADP, (e) RAD0, and (f) PATHRAD experiments analyzed at 1915 UTC. Black contours in (b)–(f) where WSR-88D reflectivity is >35 dBZ.

  • View in gallery

    As in Fig. 5, but at 1930 UTC.

  • View in gallery

    As in Fig. 6, but at 1930 UTC.

  • View in gallery

    As in Fig. 5, but at 2000 UTC. Lines through southern storm indicate location of cross sections shown in Fig. 11.

  • View in gallery

    As in Fig. 6, but at 2000 UTC.

  • View in gallery

    Cross section of ensemble mean hydrometeor mixing ratios for (a) PATH, (b) RADP, (c) RAD0, and (d) PATHRAD at 2000 UTC. Filled contours represent qc mixing ratio, line contours represent qi + qs mixing ratio, black hatched areas represent the region where qg > 1.0 g kg−1, and red hatched areas represent the region where qr > 1.0 g kg−1.

  • View in gallery

    (a)–(d) Probability of simulated 4-km reflectivity from each experiment >45 dBZ for the 90-min forecast period between 1930 and 2100 UTC. Hatched area indicates the region of observed WSR-88D reflectivity >45 dBZ.

  • View in gallery

    Skill scores including (a) POD, (b) FAR, and (c) HSS for 4-km ensemble mean reflectivity for the 90-min forecast period between 1930 and 2100 UTC. Skill scores are computed using an observed radar reflectivity threshold of 35 dBZ. (d) POD, (e) FAR, and (f) HSS for ensemble mean CWP for the 90-min forecast period between 1930 and 2100 UTC computed using an observed CWP threshold of 1.0 kg m−2.

  • View in gallery

    SWDOWN RMSE calculated between ensemble mean forecasts and OK Mesonet observations from 87 sites for (a) the 1930–2100 UTC forecast period and (b) the 2000–2130 UTC forecast period.

  • View in gallery

    (a)‒(d) Probability of simulated 4-km reflectivity from each experiment >45 dBZ for the 90-min forecast period between 2000 and 2130 UTC. Hatched area indicates the region of observed WSR-88D reflectivity >45 dBZ.

  • View in gallery

    As in Fig. 13, but for 90-min forecasts initiated at 2000 UTC.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 77 77 16
PDF Downloads 55 55 5

Simultaneous Radar and Satellite Data Storm-Scale Assimilation Using an Ensemble Kalman Filter Approach for 24 May 2011

View More View Less
  • 1 Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, Norman, Oklahoma
  • | 2 NOAA/OAR/National Severe Storms Laboratory, Norman, Oklahoma
  • | 3 NASA Langley Research Center, Hampton, Virginia
  • | 4 Science Systems and Applications, Inc., Hampton, Virginia
© Get Permissions
Full access

Abstract

Assimilating high-resolution radar reflectivity and radial velocity into convection-permitting numerical weather prediction models has proven to be an important tool for improving forecast skill of convection. The use of satellite data for the application is much less well understood, only recently receiving significant attention. Since both radar and satellite data provide independent information, combing these two sources of data in a robust manner potentially represents the future of high-resolution data assimilation. This research combines Geostationary Operational Environmental Satellite 13 (GOES-13) cloud water path (CWP) retrievals with Weather Surveillance Radar-1988 Doppler (WSR-88D) reflectivity and radial velocity to examine the impacts of assimilating each for a severe weather event occurring in Oklahoma on 24 May 2011. Data are assimilated into a 3-km model using an ensemble adjustment Kalman filter approach with 36 members over a 2-h assimilation window between 1800 and 2000 UTC. Forecasts are then generated for 90 min at 5-min intervals starting at 1930 and 2000 UTC. Results show that both satellite and radar data are able to initiate convection, but that assimilating both spins up a storm much faster. Assimilating CWP also performs well at suppressing spurious precipitation and cloud cover in the model as well as capturing the anvil characteristics of developed storms. Radar data are most effective at resolving the 3D characteristics of the core convection. Assimilating both satellite and radar data generally resulted in the best model analysis and most skillful forecast for this event.

Corresponding author address: Dr. Thomas A. Jones, Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, and NOAA/OAR/National Severe Storms Laboratory, 120 David L. Boren Blvd., Norman, OK 73072. E-mail: thomas.jones@noaa.gov

Abstract

Assimilating high-resolution radar reflectivity and radial velocity into convection-permitting numerical weather prediction models has proven to be an important tool for improving forecast skill of convection. The use of satellite data for the application is much less well understood, only recently receiving significant attention. Since both radar and satellite data provide independent information, combing these two sources of data in a robust manner potentially represents the future of high-resolution data assimilation. This research combines Geostationary Operational Environmental Satellite 13 (GOES-13) cloud water path (CWP) retrievals with Weather Surveillance Radar-1988 Doppler (WSR-88D) reflectivity and radial velocity to examine the impacts of assimilating each for a severe weather event occurring in Oklahoma on 24 May 2011. Data are assimilated into a 3-km model using an ensemble adjustment Kalman filter approach with 36 members over a 2-h assimilation window between 1800 and 2000 UTC. Forecasts are then generated for 90 min at 5-min intervals starting at 1930 and 2000 UTC. Results show that both satellite and radar data are able to initiate convection, but that assimilating both spins up a storm much faster. Assimilating CWP also performs well at suppressing spurious precipitation and cloud cover in the model as well as capturing the anvil characteristics of developed storms. Radar data are most effective at resolving the 3D characteristics of the core convection. Assimilating both satellite and radar data generally resulted in the best model analysis and most skillful forecast for this event.

Corresponding author address: Dr. Thomas A. Jones, Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, and NOAA/OAR/National Severe Storms Laboratory, 120 David L. Boren Blvd., Norman, OK 73072. E-mail: thomas.jones@noaa.gov

1. Introduction

Assimilating high-resolution remote sensing data into convection-permitting numerical weather prediction (NWP) models has led to vast improvements in forecasting convective systems. The two observation platforms that generate high-resolution data include radars (e.g., Snyder and Zhang 2003) and satellites (e.g., Pincus et al. 2011). Radar data in particular have proven to be an indispensable tool for correctly analyzing convection within NWP models (e.g., Kain et al. 2010). In the United States, the Weather Surveillance Radar-1988 Dopplers (WSR-88Ds; Crum and Alberty 1993) retrieve radar reflectivity and radial velocity in regions of ongoing precipitation. Many studies have analyzed the impacts of assimilating these data and found that each type provides information leading to improvements of convective-scale analyses and forecasts (e.g., Dowell et al. 2004; Gao et al. 2004; Aksoy et al. 2009, 2010; Dawson et al. 2012; Gao and Stensrud 2012; Yussouf et al. 2013). Assimilating radial velocity has generally been found to be the easier of the two; thus, it was the focus of several early studies (Snyder and Zhang 2003; Zhang et al. 2004). More recently, assimilation of radar reflectivity has become an important topic of research (Dowell et al. 2004; Tong and Xue 2005; Aksoy et al. 2009, 2010; Yussouf and Stensrud 2010; Dowell et al. 2011; Yussouf et al. 2013). Radar reflectivity is directly linked to the cloud hydrometeor properties of precipitating clouds and, when assimilated into an NWP with advanced cloud microphysics, can generate a very accurate analysis of convection within the model (Tong and Xue 2005). However, the relationship between reflectivity and cloud hydrometeor species is highly complex and very dependent on the cloud microphysics scheme selected within the model (Tong and Xue 2005; Yussouf et al. 2013). One common theme of most radar data assimilation studies is a focus on creating an accurate analysis of mature convection with the goal of improving forecasts of tornadic circulations. This represents one of the key goals of the Warn-on-Forecast (WoF) project (Stensrud et al. 2009, 2013), and much work is still needed on this topic. However, capturing the early development of storms and the near-storm environment is equally important to fulfill the WoF goals, but has received somewhat less attention.

One important disadvantage of radar data assimilation is that it does not capture the nonprecipitation phase of cloud development during convective initiation. Operational radars such as the WSR-88D are not sensitive to the small cloud hydrometeors associated with nonprecipitating clouds and thus produce little in the way of useful data that can be assimilated into an NWP model under these conditions. Forecasting the time and location of convection initiation has proven to be a significant challenge (Kain et al. 2013), and determining a way to assimilate information relating to convection initiation is receiving greater interest in the research community (e.g., Mecikalski et al. 2013).

Assimilating satellite observations provides a potential solution to the convective initiation and near-storm environment difficulties that currently exist (Vukicevic et al. 2004, 2006; Polkinghorne et al. 2010; Polkinghorne and Vukicevic 2011; Zupanski et al. 2011; Jones et al. 2013b; Zhang et al. 2013). Two different approaches to satellite data assimilation are currently used. The first is the direct assimilation of satellite infrared and microwave radiances. Model state variables are generally transformed into simulated satellite radiances via a radiative transfer model (RTM) built into the forward operator (Vukicevic et al. 2004, 2006; McNally et al. 2006; Chen et al. 2008; Otkin 2010; Polkinghorne et al. 2010; Polkinghorne and Vukicevic 2011; Zupanski et al. 2011). Assimilating radiances directly often has been the favored approach in operations and research studies since it avoids uncertainties and discrepancies in the various retrieval algorithms that differ from satellite to satellite (Derber and Wu 1998; Errico 2000). This technique generally performs best in clear-sky regions, but it can be very resource intensive depending on the number and resolution of channels being assimilated (Migliorini 2012). For storm-scale applications, uncertainties in surface emissivity and needed bias corrections add a layer of difficulty compared to assimilating these data on larger scales. Adding cloudy radiances only increases the uncertainties and adds the complication of potential differences in cloud microphysics assumptions between the model and the RTM (e.g., Zupanski et al. 2011). Using observing system simulation experiments (OSSEs), Otkin (2010) and Jones et al. (2013a, 2014) have studies assimilating cloud radiances for mesoscale applications with a degree of success. Jones et al. (2013a, 2014) used an OSSE to determine the impacts of assimilating radar reflectivity and radial velocity combined with Geostationary Operational Environmental Satellite (GOES) Advanced Baseline Imager (ABI) 6.95-μm (water vapor channel) brightness temperatures TB for a winter weather case on 24 December 2009. Assimilating both radar and satellite observations generally produced the most accurate analysis and short-term forecasts, though the improvement over radar data assimilation alone was not large, especially at low levels.

The second method used to assimilate satellite data is through the use of derived products, known as retrievals, such as temperature and humidity profiles in clear sky, and cloud properties in cloudy regions. This represents a more resource friendly and easier to interpret approach while providing nearly the same information content (Migliorini 2012). Jones et al. (2013b) assimilated cloud water path (CWP) retrievals (Minnis et al. 2008b, 2011) from the GOES-13 satellite for a severe weather event occurring on 10 May 2010. Results indicated that assimilating CWP retrievals at 15-min intervals over a 3-h period generated a better representation of convection and suppressed development of spurious convection within the model compared to an experiment that did not assimilate CWP retrievals. Validation against Oklahoma (OK) Mesonet surface observations showed that assimilating CWP reduced errors in surface radiation and temperature indicating an improved near-storm environment. While this initial attempt at CWP assimilation is promising, the results are only indicative of a single case.

It is clear that both radar and satellite data bring independent and useful information to convection-permitting models; thus, their combined assimilation should prove advantageous. This research combines the CWP assimilation methods used by Jones et al. (2013b) with techniques used for radar data assimilation similar to those developed by Dowell et al. (2011) and Yussouf et al. (2013) to examine the impacts of assimilating both in a convection-permitting NWP model for a severe weather event occurring on 24 May 2011. This event was characterized by the development of several supercell thunderstorms in west-central OK that eventually moved into central OK, producing several significant tornadoes. Emphasis will be placed on using satellite and radar data to correctly analyze the time and location of convective initiation (CI) and the development phase of the convection. Once mature convection has developed, data assimilation ceases with forecasts generated thereafter. Using an ensemble adjustment Kalman filter (EaKF) approach (Kalman 1960; Anderson 2001; Anderson and Collins 2007), several experiments are performed to test the relative impacts of assimilating radar and satellite data. The EaKF approach uses the statistics from an ensemble of nonlinear forecasts to estimate the background error covariances. When cycled, this allows for flow-dependent covariances to be generated, which provide a significant advantage over constant background errors used in many three-dimensional variational methods. This approach also provides both a mean analysis and an uncertainty estimate, which is useful for creating probabilistic forecasts. Experiments generated for this research include a control that only assimilates conventional and OK Mesonet observations (McPherson et al. 2007), followed by those that assimilate conventional data with only radar or satellite data, and finally, those including all the available observations. The goal will be to determine where radar and satellite data provide the greatest impact to the model analysis and whether combining both can consistently lead to a better forecast.

Following the introduction, the 24 May 2011 event is summarized in section 2. Section 3 provides a description of the observation types used by these experiments with a section for describing the necessary CWP and radar forward operators. Section 5 provides information on the experiment design. Section 6 discusses observation diagnostics for radar reflectivity, radial velocity, and CWP. Sections 7 and 8 discuss the analysis and forecast results of each experiment, while section 9 summarizes the overall conclusions.

2. Severe weather event: 24 May 2011

On 24 May 2011, atmospheric conditions in central OK were very favorable for the development of severe thunderstorms. During the day, an upper-level trough located in the western United States transitioned eastward while taking on a negatively tilted orientation. By 1800 UTC, its influence was being felt in central OK, where a special 1800 UTC radiosonde launch from Norman, OK (KOUN), captured 25 m s−1 southwesterly winds at 500 hPa with >40 m s−1 westerlies aloft associated with the incoming jet streak (Fig. 1a). The KOUN sounding also showed a highly unstable atmosphere with a strong capping inversion in place near 825 hPa. The corresponding veering wind profile was particularly favorable for rotating supercells (Fig. 1b). Surface conditions were characterized by a warm, moist air mass ahead of a dryline located at western OK at 1800 UTC (Figs. 1c,d). The dryline located in western OK began moving eastward in response to the upper-level trough and corresponding surface cyclogenesis. Figure 1c shows 2-m surface temperatures in excess of 34°C behind the dryline, whose location approximated by the 40% relative humidity contour in Fig. 1d. Higher humidity and low-level cloud cover result in lower surface temperatures east of the dryline. CI occurred along the advancing dryline in west-central OK. The first severe thunderstorm warning was issued at 1903 UTC, with the first severe weather report following at 1930 UTC. The first tornado warning for this event was issued at 2000 UTC and was realized 20 min later with a report of the first tornado spawned from the northernmost supercell.

Fig. 1.
Fig. 1.

(a) The 1800 UTC KOUN radiosonde showing vertical profiles of temperature, dewpoint, wind speed, and direction. (b) Corresponding KOUN hodograph at 1800 UTC. (c) Ensemble mean 2-m surface temperature at 1800 UTC over west-central OK. (d) Ensemble mean 2-m relative humidity at 1800 UTC with a 40% contour indicating approximate location of the dryline at this time.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

Combined GOES-13 6.7-μm water vapor channel brightness temperatures TB and 6-km WSR-88D reflectivity data at 1902 UTC show the initial development in western OK as three separate cells aligned north to south ahead of the dryline, with the northernmost storm having the initial severe thunderstorm warning (Fig. 2a). The upper-level cloud shield originating from the storms already extends eastward of the 20-dBZ reflectivity observations. By 1930 UTC, the storms have grown in strength and size (Fig. 2b). The areal coverage of the >50 dBZ area has increased along with the downstream cirrus outflow where TB < −40°C. Intensification continued to 2000 UTC with a rapid increase in the coverage of the >50 dBZ radar reflectivity and cirrus outflow from the storm top compared to 1930 UTC (Figs. 2b,c). The TB associated with the cirrus outflow decreases to <50°C, indicating colder, higher cloud tops. By 2030 UTC (Fig. 2d), the original three cells have become well-established supercells. The rapid storm development and corresponding storm interactions with each other and the surrounding environment make this a challenging event to forecast.

Fig. 2.
Fig. 2.

GOES-13 6.7-μm water vapor channel with WSR-88D reflectivity overlaid at (a) 1900, (b) 1930, (c) 2000, and (d) 2030 UTC 24 May 2011. Satellite data have been parallax corrected. Reflectivity data are within a black outline, outside of which satellite brightness temperatures are shown.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

3. Observations

a. Conventional and Mesonet

Conventional observations, considered to be those most commonly assimilated into NWP models, include data from the Automated Surface Observing System (ASOS), Aircraft Communications Addressing and Reporting System (ACARS), and radiosonde instruments acquired from the Meteorological Assimilation Data Ingest Files System (MADIS). Surface observation types include 10-m wind speed and direction, 2-m temperature and humidity, and surface pressure observations at 5-min intervals. Radiosondes provide vertical profiles of temperature, humidity, and wind speed and direction. Finally, ACARS includes temperature and wind observations along aircraft flight tracks. Observation measurement errors for each are provided in Table 1.

Table 1.

Observation types, errors, horizontal and vertical localization radii (H. local and V. local), and the number of observations assimilated in the PATHRAD experiment between 1800 and 2000 UTC. The clear-air reflectivity sample size (marked by an asterisk) is obtained from the RAD0 experiment.

Table 1.

In addition to the observation types listed above, surface observations from the OK Mesonet are assimilated into each experiment. The OK Mesonet consists of a high-density network of sensors that measure surface temperature, humidity, pressure, and wind properties (McPherson et al. 2007). Data from each observation site are archived at 5-min intervals. The high-resolution nature of this dataset provides the opportunity to better analyze small-scale structures in the surface fields beyond that possible with ASOS observations alone. The network is maintained to a high-quality standard and observation errors are set to be the same as corresponding ASOS observations.

b. GOES CWP retrievals

The GOES imager takes multispectral images over the continental United States (CONUS) every 5–15 min, depending on the need (Menzel and Purdom 1994). Top of atmosphere radiances are sampled in five spectral bands: a 1-km visible (0.65 μm) channel, a 4-km shortwave infrared (3.9 μm) channel, a 4-km water vapor channel at 6.5 μm, a 4-km atmospheric infrared window (10.8 μm) channel, and an 8-km CO2 absorption channel at 13.3 μm (Menzel and Purdom 1994; Schmit et al. 2001).

Cloud properties are retrieved from 4-km GOES imager data for pixels classified as cloudy (Minnis et al. 2008b) using the multispectral retrieval algorithm developed by Minnis et al. (2011) and adapted to geostationary data (Minnis et al. 2008a). Daytime cloud properties are retrieved using a variant of the visible infrared shortwave-infrared split-window technique (VISST; Minnis et al. 2011). The modified CO2-absorption technique of Chang et al. (2010) uses 10.8- and 13.3-μm channels to retrieve cloud top pressure (CTP), cloud top temperature (CTT), cloud emissivity, and cloud phase for high clouds having a cloud optical thickness (COT) less than ~4. Its retrievals are sometimes used to replace the VISST retrievals of the same quantities. Cloud-base pressure is the pressure corresponding to the altitude equal to the difference between cloud top height and cloud thickness. Cloud phase classifies a cloudy pixel as either “liquid” or “ice” based on the cloud temperature and cloud effective particle size information. Optically thick clouds containing both liquid and ice phase hydrometeors are generally classified as ice clouds since the current iteration of the retrieval algorithm is unable to separately classify mixed-phase clouds. Hereafter, liquid water path (LWP) refers to cloud water path associated with liquid phase clouds only while ice water path (IWP) refers to the cloud water path for ice and mixed-phase clouds. CWP is used when discussing both LWP and IWP simultaneously. A full description of the VISST algorithm is found in Minnis et al. (2011).

The observation error is defined as a function of both LWP and IWP values, with the lowest errors defined for clear-sky retrievals and the highest defined for high CWP retrievals instead of the constant value used by Jones et al. (2013b) For CWP = 0, the observation error is 0.025 kg m−2 since high confidence exists in the retrievals for clear-sky regions. For 0.025 < CWP < 0.2 kg m−2, the observation error increases to 0.05 kg m−2; for 0.2 < CWP < 0.5 kg m−2, the observation error increases to 0.075 kg m−2; and for 0.5 < CWP < 1.0 kg m−2, the observation error increases to 0.10 kg m−2. Finally, for CWP > 2.5 kg m−2, the observation error becomes 0.15 kg m−2 because of the greater expected uncertainties in the presence of optically thick clouds (e.g., Minnis et al. 2007). The observation errors are based on the uncertainty characteristics of the retrieval algorithm and further are tuned slightly to prevent high-uncertainty, high CWP values from having too great of an impact to the model analysis.

Jones et al. (2013b) assimilated raw CWP observations at their original 4-km resolution. Later research discovered that the variations in CWP associated with deep convection are often not representative of the convective characteristics below the top of the storm. To reduce the negative impacts of these variations, the CWP data are now smoothed to a 6-km grid prior to assimilation. This smoothing removes the unwanted variations and also produces a dataset better suited to a 3-km model using the “2Δx” guideline described by Lu and Xu (2009). For tall clouds, geolocation errors exceeding 10 km can occur; thus, a parallax correction is also applied where CWP > 0.0 kg m−2. The geolocation of the raw satellite data and retrievals measure the physical condition of the cloud at its top and not its base, and since the satellite is not directly overhead, the relative locations at the surface and aloft are not the same. To correct for parallax, the retrieved cloud height is used to remap cloudy pixels to their zenith location above the surface (Wang and Huang 2014).

As convection matures, cloud optical thickness increases to the point where the satellite retrievals become saturated, artificially limiting CWP in high-precipitation regions. This also results in a high bias in the cloud-base pressure (CBP) retrieval. When only CWP retrievals are assimilated, CBP values are adjusted downward when CWP > 1.0 kg m−2 and the original CBP < 500 hPa. However, the underestimate of CWP remains, potentially weakening already mature convection in the model analysis if assimilated. Combing radar and satellite data provides a potential solution in the form of a radar reflectivity mask. Using the assumption that CWP retrievals are likely saturated in high reflectivity regions, nearby CWP retrievals are not assimilated where the column-integrated reflectivity exceeds 40 dBZ. This mask is only applied when both radar and satellite are assimilated in the same experiment.

c. WSR-88D reflectivity and radial velocity

WSR-88D reflectivity and radial velocity are obtained from three radars in central and western Oklahoma. Radar sites include: Fredrick, OK (KFDR); Vance Air Force Base, OK (KVNX); and Twin Lakes, OK (Fig. 3). The Quality Control Neural Network (QCNN) algorithm (Lakshmanan et al. 2003) is applied to the level-2 data to remove potential nonmeteorological targets such as ground clutter, anomalous propagation, and sun glint from the data. Corresponding Doppler radial velocity data are dealiased using the method developed by Eilts and Smith (1990). The cleaned radar data are objectively analyzed using the Observation Processing and Wind Synthesis (OPAWS) software1 to an evenly spaced 6-km horizontal Lambert conformal grid while retaining the conical scanning surfaces (Dowell et al. 2004; Dowell and Wicker 2009; Majcen et al. 2008) using a single-pass Cressman interpolation scheme with a 3-km radius (Dowell et al. 2004). Objectively analyzed radar reflectivity and radial velocity data are generated at 5-min intervals beginning at 1845 UTC and continuing until 2000 UTC. The 5-min data files contain all observations within ±2.5 min from the central time, which are then assumed to be valid at the central time.

Fig. 3.
Fig. 3.

Map showing the background 15-km model domain (labeled “1”), inner 3-km domain (labeled “2”), and 3-km experiment domain (labeled “3”). Locations for the three WSR-88Ds used in this research (KFDR, KTLX, KVNX) are also shown.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

Objectively analyzed radar reflectivity below 5 dBZ is set to 0 dBZ, and the resulting set of 0 dBZ data are separated out as a different variable type, “clear-air reflectivity,” with remaining positive values labeled “reflectivity.” Assimilating clear-air reflectivity has proven useful as a means of suppressing spurious convection that may be part of the initial model state or that develops over the course of the model run (e.g., Tong and Xue 2005). However, assimilating clear-air reflectivity is also very resource intensive, which is compounded when using multiple radars. Using satellite retrievals to define clear-sky areas (CWP = 0 kg m−2) should be much more resource friendly since it is a 2D (rather than 3D) variable from a single sensor. In addition, clear-air reflectivity and CWP = 0 kg m−2 have somewhat different meanings. Clear-air reflectivity indicates precipitation-free areas while CWP = 0 kg m−2 indicates precipitation and cloud-free areas. To reduce the resources required for clear-air reflectivity assimilation, clear-air values are further thinned, with only every other point in the horizontal assimilated. No thinning in the vertical occurs. Observation errors for both clear-air and positive radar reflectivity are set to 5 dBZ based with the corresponding radial velocity observation error set to 3 m s−1, similar to Dowell et al. (2004).

4. Forward operators

a. Cloud water path

Cloud hydrometeor mixing ratios in the model analysis are related to LWP and IWP retrievals using the forward operator developed by Jones et al. (2013b). In summary, simulated CWP are calculated at each model time step using the column-integrated cloud hydrometeor mixing ratio. For each grid point and model level, the mixing ratios of each hydrometeor variable (qc, qr, qi, qs, and qg) are summed to form a total cloud water mixing ratio qa. For mixed-phase clouds, the total cloud water mixing ratio is then integrated over the entire atmospheric column and divided by gravity to calculate the simulated IWP value. A similar value (qliq) is generated from only the liquid phase hydrometeors for comparison with LWP retrievals.

For both liquid and mixed-phase clouds, the vertical extent of the cloud layer at each observation location when CWP > 0.0 kg m−2 is defined by CTP and CBP and is passed to the forward operator and used to constrain the CWP calculated from the model. For example, if CTP = 400 hPa and CBP = 800 hPa, then the mixing ratios of each cloud variable are only summed between these two levels. Mixing ratios above and below the cloud layer are not considered in the forward operator. Clear-sky (CWP = 0 kg m−2) retrievals are also assimilated, except that the vertical extent of the observation remains undefined. Thus, the clear-sky retrievals are compared against mixing ratios from all model levels, with the goal of reducing positive mixing ratios should they exist in what is observed to be a cloud-free environment.

b. Radar reflectivity and radial velocity

The radar forward operator in the data assimilation software for reflectivity is similar to that used for CWP. Simulated reflectivity is generated using the prognostic cloud hydrometeor variables, temperature, and humidity as inputs. The forward operator computes the simulated reflectivity from the model at every grid point from the surface to the model top. The simulated value is then compared against observed reflectivity during the assimilation process and the appropriate adjustments made to the model fields, generating a simulated reflectivity that better matches observations.

Assimilating Doppler radial velocity is a relatively straightforward process. The forward operator uses radar location and beam direction information to determine the simulated radial velocity from model horizontal and vertical velocity variables at each grid point, which are adjusted accordingly when compared against observations during the assimilation process (e.g., Tong and Xue 2005).

5. Experiment design

a. Model configuration

Experiments are initiated from ensemble backgrounds generated from the Global Ensemble Forecast System (GEFS) analysis at 0000 UTC 24 May 2011. The GEFS is a 21-member ensemble version of the Global Forecast System (GFS) run operationally by the National Centers for Environmental Prediction (NCEP). For this event, GEFS has a 1° grid spacing with 27 vertical levels from the surface to 10 hPa. Initial and boundary conditions at 15 km (mesoscale) are generated from the original GEFS members (Fig. 3). Simultaneously, a 3-km (storm scale) grid is downscaled from the mesoscale grid using a one-way nested grid configuration with information only exchanged at the lateral boundaries at the time the storm-scale grid is generated. For both grids, vertical resolution increases to 51 levels stretching from near the surface to 10 hPa. The GFS land surface data are replaced by higher-resolution land surface data from the Noah land surface model (Ek et al. 2003). To create the full suite of 36 ensemble initial conditions required for the experiments, each of the first 18 GEFS members are run with a unique combination of cumulus parameterizations, including Kain–Fritsch, Grell, and Tiedtke (Kain and Fritsch 1993; Grell and Devenyi 2002; Tiedtke 1989); radiation schemes Rapid Radiative Transfer Model (RRTM), RRTM for General Circulation Models (RRTMG), and Dudhia (Mlawer et al. 1997; Iacono et al. 2008; Dudhia 1989); and planetary boundary layer schemes Yonsei University (YSU), Mellor–Yamada–Janjić (MYJ), and Mellor–Yamada–Nakanishi–Niino (MYNN; Hong et al. 2006; Janjić 1994; Nakanishi and Niino 2006). Members 19–36 use the same set of physics diversity, but the order of GEFS members is reversed to create a set of 36 different initial and boundary conditions. The multiphysics approach acts to increase model spread and improve the covariance structures, reducing the probability of model convergence (Stensrud et al. 2000; Fujita et al. 2007). All members use Thompson cloud microphysics (Thompson et al. 2004, 2008). No cumulus parameterization is applied on the storm-scale grid.

Thompson cloud microphysics is selected as a compromise between purely single-moment schemes such as Lin (Lin et al. 1983) and full double-moment schemes such as Milbrandt and Yau (Milbrandt and Yau 2005a,b). Thompson represents a partial double-moment scheme, which predicts number concentration for only rain and ice hydrometeors species. Double-moment schemes generally perform better for radar data assimilation experiments (e.g., Mansell 2008; Jung et al. 2012; Yussouf et al. 2013), but the improvement is more evident at a 1-km resolution than for the 3-km resolution used here. Double-moment schemes predict hydrometeor number concentrations in addition to mixing ratios, allowing a high-resolution model to better capture processes related to diffusion, evaporation, and sedimentation within ongoing convection. Previous studies (e.g., Keil et al. 2003; Chaboureau and Pinty 2006; Otkin and Greenwald 2008) have also observed a strong dependence between model microphysics and simulated satellite observations impacting both the assimilation of radiances and retrievals. Here, also, full double-moment schemes generally performed better (e.g., Grasso and Greenwald 2004; Carrio et al. 2008). The impact of assimilating CWP retrievals for different microphysics scheme represents an ongoing research exercise, but early results indicate a significant relationship. For this research, both radar and CWP data assimilation have been previously tested using Thompson with good results, leading to its use in the combined experiments here.

Members are integrated hourly, assimilating convectional observations [aviation routine weather report (METAR), marine, radiosonde, ACARS, and satellite winds]. All data are assimilated using the parallel EaKF from the Data Assimilation Research Testbed (DART) software system (Anderson and Collins 2007; Anderson et al. 2009) using a configuration similar to that employed by Wheatley and Stensrud (2010), Yussouf et al. (2013), and Jones et al. (2013b). For conventional observations, a horizontal localization radius of 460 km is applied using the Gaspari and Cohn (1999) technique (Table 1). Available Mesonet observations are only assimilated in the storm-scale domain using a smaller localization radius of 100 km. All conventional and Mesonet observations use a 6-km vertical localization radius. Using the Advanced Research Weather Research and Forecasting Model (WRF-ARW), version 3.4.1 (Skamarock et al. 2008), the ensemble members are integrated forward in time. The mesoscale and storm-scale members are integrated out to 0000 UTC 25 May to provide ample initial and boundary conditions for several potential experiments using this event.

Prognostic variables updated at each assimilation cycle are the three wind components, perturbation temperature, perturbation geopotential, perturbation surface pressure of dry air, potential temperature tendency, diabatic heating, and water vapor and hydrometeor mixing ratios. Also updated are the 10-m wind fields, 2-m temperature and water vapor fields, and surface pressure variables. All are updated as part of the parallel EaKF system for each observation type being assimilated. Both positive and clear-sky CWP points are assimilated using a horizontal localization radius of 40 km. Unlike Jones et al. (2013b), cloud height information is used to define a vertical coordinate with a corresponding vertical localization for CWP. The vertical coordinate is defined as the average of CTP and CBP approximating the center of the cloud with a 6-km vertical localization radius applied surrounding the center. The vertical localization of clear-sky retrievals remains undefined as there is no cloud height available to determine a vertical coordinate. Reflectivity and radial velocity are assimilated using a horizontal localization radius of 18 km with a vertical localization radius of 6 km (Table 1). The horizontal localization radius for radar observations is smaller than CWP retrievals since the former sample a much smaller area of the atmosphere. While the satellite retrievals are available at a nominal resolution of 4 km, the actual information content comes from a larger area, especially since the satellite is not located directly overhead.

The primary tool to maintain ensemble spread is the application of adaptive covariance inflation to the prior analysis fields at each assimilation cycle (Anderson 2007, 2009). Adaptive inflation increases prior ensemble variance to counteract the natural loss of variance owing to sampling and observation errors. This flavor of adaptive inflation derives a separate inflation value for each state-vector variable in place of a single constant value. Full descriptions of the adaptive inflation schemes used here can be found in Anderson (2007, 2009). Adaptive inflation is only partially effective in maintaining spread in storm-scale ensembles containing convection. To address this issue, Dowell and Wicker (2009) used the additive noise technique to perturb model state variables where observed reflectivity is greater than a given threshold. Additive noise represents the application of random noise to a subset of model state variables for each member to increase ensemble spread (Dowell and Wicker 2009). When assimilating high-density observations such as those generated by radar or satellite sensors, ensemble spread can decrease to the point where the ensemble spread collapses, rendering the assimilation unstable. Jones et al. (2013b) did note a less-than-ideal spread in the posterior analyses using experiments that did not apply any sort of additional ensemble perturbations outside adaptive inflation. One solution is to significantly increase the number of ensemble members, but that is not practical from a resources perspective. The other is to artificially perturb model variables in each member to increase spread using the concept of additive noise. When IWP or LWP > 1.0 kg m−2, perturbations are applied to temperature, dewpoint, zonal, and meridional wind components. The magnitude of the perturbations is 0.5 standard deviations smoothed over a 9-km horizontal radius and 6-km vertical radius. Additive noise is applied at 15-min intervals. For this research, the model variables are perturbed using the satellite retrieval data and, since high CWP values should be collocated with high reflectivity values, no additional additive noise is applied in the experiments where both satellite and radar data are assimilated. For the sake of consistency, CWP retrievals are used to define the locations of additive noise when only radar data are assimilated.

b. Assimilation experiments

The 3-km storm-scale ensembles at 1800 UTC generated using the method described above are used as initial conditions for a smaller (170 × 170) 3-km domain to test satellite and radar data assimilation. The smaller domain is centered in west-central OK and covers the western two-thirds of the state while also reaching into northern Texas and southern Kansas (Fig. 3). Boundary conditions are generated at hourly intervals using the larger 3-km domain. Data are assimilated beginning at 1800 UTC and continuing at 15-min intervals to 1845 UTC, at which time convection initiates in western OK. Following 1845 UTC, the assimilation frequency increases to 5 min and continues out to 2000 UTC. Thereafter, two sets of forecasts are generated for each member beginning at 1930 and 2000 UTC and continuing for 90 min with output generated at 5-min intervals.

Using this process, five experiments are defined, each assimilating different combinations of conventional, satellite, and radar data (Table 2). The control (CNTL) experiment only assimilates conventional observations during the assimilation experiments and represents a baseline on model performance without the aid of high-resolution remote sensing data assimilation. The PATH experiment assimilates both conventional observations and satellite CWP retrievals when available. Two separate radar data assimilation experiments are performed. One assimilates radial velocity and positive (>0 dBZ) radar reflectivity (RADP) and the other assimilates radial velocity, positive reflectivity, and clear-air (0 dBZ) reflectivity (RAD0). This is done to determine the effectiveness of clear-air reflectivity at suppressing spurious convection and to assess its resource requirements. The final experiment (PATHRAD) combines conventional, satellite, and radar data to determine whether the combination of all datasets provides skill over either radar or satellite data alone. PATHRAD does not use clear-air reflectivity and instead relies on clear-sky CWP observations to suppress spurious convection.

Table 2.

Assimilation experiments used in this research along with observation types included within each experiment. An “×” below the observation type indicates that the corresponding experiment assimilates these data when available.

Table 2.

6. Observation diagnostics

To assess how radar and satellite observations are being assimilated, observation diagnostics in the form of bias, root-mean-square innovation (RMSI), and total spread (TSPRD) are calculated for positive radar reflectivity, radial velocity, LWP, and IWP at each assimilation cycle (Dowell et al. 2004; Dowell and Wicker 2009; Dowell et al. 2011). Bias and RMSI are calculated by taking the difference between prior and posterior fields [H(x)] and comparing it against observations (y) as shown in Eqs. (1) and (2), where N represents the number of observations assimilated for a particular observation type:
e1
and
e2
Ensemble spread represents the standard deviation of observation error over each member [Eq. (3)], where E is the total number of ensemble members, 36 in this case:
e3
TSPRD combines spread with the observation error used for each data type. The consistency ratio (CR) represents the ratio of prior ensemble variance (RMSI2) to the square of TSPRD. CR values near 1.0 indicate that the ensemble variance is a good approximation of prior or posterior error variance for a given observation error (Dowell et al. 2004; Aksoy et al. 2009; Dowell and Wicker 2009). Small CR values indicate an ensemble with insufficient spread, while large CR values indicate an ensemble with too much spread because of larger than optimum observation errors.

Figure 4 shows observation diagnostics for radar reflectivity >0 dBZ, radial velocity, LWP, and IWP along with the number of observations assimilated between 1800 and 2000 UTC for each assimilation cycle calculated from the PATHRAD experiment. Similar values were observed in experiments that assimilate only radar or satellite data (not shown). Prior to 1915 UTC, reflectivity RMSI ranges between 8 and 12 dBZ, indicating that the early analyses are not accurately capturing the convection already ongoing by 1900 UTC (Fig. 2a). Between 1910 and 1920 UTC, RMSI decreases to ~6 dBZ and bias improves to near zero corresponding to the time when the model spins up convection. Reflectivity CR increases from very small values prior to 1915 UTC to near 1.0 by 1930 UTC and stabilizing thereafter (Fig. 4b). The number of observations assimilated also increases from 2000 at 1845 UTC to 5500 at 1915 UTC and finally to near 10 000 by 2000 UTC as the number of positive reflectivity observations increases as the coverage and intensity of the convection increases.

Fig. 4.
Fig. 4.

Observation diagnostics including bias, RMSI, CR, and number of observations assimilated for (a),(b) radar reflectivity; (c),(d) radial velocity; (e),(f) LWP; and (g),(h) IWP generated from the PATHRAD experiment.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

Observations diagnostics for radial velocity are provided in Figs. 4c and 4d with a RMSI of ~4 m s−1 for prior fields decreasing to ~2 m s−1 for posterior fields and remaining generally constant as a function of time. However, spread increases from ~4 m s−1 at 1845 UTC to in excess of 5 m s−1 by 1930 UTC and several cycles thereafter. The resulting CRs range between 1.4 and 2.0 after 1915 UTC. The CR for radial velocity is above what would be considered optimal for the entire assimilation period (Fig. 4d). One cause is the selection of a 3 m s−1 observation error. Previous research (Dowell and Wicker 2009; Dowell et al. 2011) and experiments with this case study used an observation error of 2 m s−1, which generated more optimal CRs between 0.8 and 1.3 over the same time period. A larger observation error was selected in this research since quality control is a purely automated process potentially leaving some poorly dealised observations in the assimilated dataset. Comparisons with 2 versus 3 m s−1 observation error experiments showed that increasing the error to more realistic levels did not have a significant impact on the model-generated convection. For both radar reflectivity and radial velocity, TSPRD (and CR) are larger near the surface and decrease as a function of height (not shown).

Observation diagnostics for satellite retrievals are broken down by LWP and IWP subsets to determine the relative effects of assimilating different cloud types. Figures 4e and 4f show observation diagnostics for LWP, with the first observations being assimilated at 1815 UTC. During the period of 5-min cycling (1845–2000 UTC), several times exist where no satellite retrievals are available (Table 3). When retrievals exist, RMSI and TSPRD range between 0.03 and 0.06 kg m−2 for most cycles, with no overall trend evident except the expected RMDI reduction in the posterior analysis compared to the prior analysis. Bias is slightly negative, indicating that the experiment slightly underestimates LWP throughout. That could be due to the use of a nonadiabatic definition of LWP in the satellite retrievals, which is 12% greater than the adiabatic definition. CR varies from a low of 0.4 at 1900 UTC to a high of 1.6 at 1945 UTC, but most values range between 0.6 and 1.3 after 1915 UTC (Fig. 4f). The number of observations assimilated decreases as a function of time as LWP observations are replaced by IWP observations as convection intensity increases.

Table 3.

Data assimilation cycles and available data assimilated at each cycle: “C” refers to conventional observations, “S” refers to GOES-13 CWP retrievals, “R” refers to radar reflectivity and radial velocity, and “A” represents clear-air reflectivity.

Table 3.

The observation diagnostics for IWP are provided in Figs. 4g and 4h and show that assimilating IWP reduces RMSI between posterior and prior analysis at all cycles. However, several features stand out. First, the prior error at 1910 UTC is near 1.2 kg m−2, which is much larger than at any other time. This is a result of missing 1905 UTC satellite retrievals and the rapidly developing nature of the convection at this time. Thus, a large difference between the 1910 UTC prior analysis and 1910 UTC retrievals becomes apparent. Following this anomaly, prior RMSI stabilized to between 0.5 and 0.6 kg m−2, with posterior values ranging between 0.2 and 0.3 kg m−2. CR varies from near zero prior to 1915 to near 2 by 1930 before settling to near 1.0 during the final three assimilation cycles (Fig. 4h). Another feature of note is the very small number of observations being assimilated prior to 1930 UTC, being less than 250 per cycle. Thus, the observation diagnostics may be somewhat unreliable because of the small sample size at these earlier times.

It is important to note that generating an ideal CR for all high-resolution observation types simultaneously is difficult as spread and error characteristics differ significantly as a function of various model fields. While some individual statistics are not ideal (e.g., radial velocity CR), overall observations diagnostics from these experiments indicate that both radar and satellite data are being assimilated in a robust manner.

7. Analysis results

The positive impacts of assimilating both satellite and radar data can be seen at several points during the assimilation cycle when comparing ensemble mean analyses with corresponding satellite and radar observations. Convection initiates in all experiments except CNTL between 1910 and 1915 UTC and rapidly increases in coverage and intensity thereafter. Figure 5 shows GOES-13 CWP retrievals (Fig. 5a) along with ensemble mean simulated CWP for the five experiments conducted here (Figs. 5b–f). The CNTL experiment fails to generate CWP near the observed convection, while at the same time generating false cloud cover to the south and east (Fig. 5b). The PATH experiment performs much better, generating a reasonably good representation of CWP in the analysis, and also suppressing high CWP present farther east in CNTL (Fig. 5c). RADP correctly generated high CWP values at these points roughly corresponding to the areas of maximum observed reflectivity (Fig. 5d). However, the placement of the southern storms is incorrect and the spurious CWP to the east remains. Assimilating clear-air reflectivity in RAD0 helps reduce this false CWP while not significantly impacting other parts of the analysis compared to RADP (Fig. 5e). CWP in nonprecipitating regions remains too high compared to observations, indicating that assimilating clear-air reflectivity is not as effective at suppressing nonprecipitating clouds as CWP as would be expected. The PATHRAD experiment that assimilates both satellite and radar data combines the advantages of PATH in the nonprecipitating/cloud-free regions with the advantages of RADP where convection is ongoing (Fig. 5f). It also develops the convection faster, which will become more evident when comparing simulated radar reflectivity from each experiment.

Fig. 5.
Fig. 5.

(a) GOES-13 CWP retrievals, and ensemble mean simulated CWP from (b) CNTL, (c) PATH, (d) RADP, (e) RAD0, and (f) PATHRAD experiments analyzed at 1915 UTC. Black contours in (b)–(f) indicate regions of where GOES-13 retrievals are CWP > 1.0 kg m−2.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

The corresponding 1915 UTC observed (Fig. 6a) and simulated reflectivity at 4 km for each experiment (Figs. 6b–f) are shown in Fig. 6. The CNTL experiment fails to generate any simulated reflectivity at these locations (Fig. 6b), while having a large area of 5–10 dBZ farther east corresponding to the false CWP seen in Fig. 5b. The PATH experiment eliminates this feature while generating weak reflectivity near the locations of the northern two cells, with a single spurious maximum biased slightly eastward (Fig. 6c). The RADP experiment is stronger with the simulated reflectivity, as expected, while retaining the faults of CNTL (Fig. 6d). RAD0 (Fig. 6e) is similar to RADP except that it corrects the reflectivity bias to the east. However, recall that significant nonprecipitating clouds remain in the model as evidenced by the corresponding CWP analysis at this time (Fig. 5e). Despite assimilating reflectivity for ~30 min, the radar-only experiment still fails to correctly capture the developing storms. When no positive reflectivity exists in the model, high reflectivity observations associated with developing convection are considered outliers and may not be assimilated. This slows down the spinup of convection within the model compared to observations. The PATHRAD experiment generates >50 dBZ reflectivity cores associated with the northern two cells with a weaker, but correctly located, reflectivity core corresponding to the southernmost cell (Fig. 6f). The faster increase in convection in PATHRAD is a result of the satellite and radar data providing a signal to the model that convection is present. For this case, the satellite data help generate high-CWP clouds and their associated cloud hydrometeors, which then allows the radar data to build upon those already present hydrometeors instead of having to do it alone, also indicating that fewer reflectivity observations are being considered outliers. Thus, the combination of satellite and radar data provides a clear benefit to the model during convective initiation and the short period thereafter.

Fig. 6.
Fig. 6.

(a) WSR-88D reflectivity at 4 km, and ensemble mean simulated reflectivity from (b) CNTL, (c) PATH, (d) RADP, (e) RAD0, and (f) PATHRAD experiments analyzed at 1915 UTC. Black contours in (b)–(f) where WSR-88D reflectivity is >35 dBZ.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

Moving ahead 15 min to 1930 UTC, the coverage of CWP > 1.0 kg m−2 increases significantly in the observations (Fig. 7a). The CNTL experiment again fails to generate anything while retaining and, in some cases, increasing the spurious CWP farther east (Fig. 7b). The remaining experiments all capture the observed CWP quite well, but several important differences are apparent. RADP retains the spurious CWP, which PATH, PATHRAD, and even RAD0 largely eliminate by this time (Figs. 7c,e,f). The effect of assimilating CWP corresponding to the ongoing convection also acts to increase simulated CWP around the core compared to the radar-only experiments. Comparing RAD0 and PATHRAD, only the latter maintains CWP > 1.0 kg m−2 to the observed 1.0 kg m−2 line plotted with the difference most evident with the northern cell. All experiments except CNTL generate maximum CWP values in excess of the GOES-13 retrievals as the latter become saturated in heavy precipitation regions.

Fig. 7.
Fig. 7.

As in Fig. 5, but at 1930 UTC.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

The simulated z = 4 km radar reflectivity for each experiment tells much the same story as CWP when compared with corresponding radar observations (Fig. 8). The CNTL experiment remains poor (Fig. 8b), while the other experiments generate reflectivity nearby observations. PATH does not generate the >50 dBZ reflectivity cores and has an overall poor representation of the southern storms (Fig. 8c), since CWP retrievals are already saturated at these locations. The radar assimilation experiments, RADP and RAD0, both generate much higher simulated reflectivity values, better matching observations in magnitude and location (Figs. 8d,e). However, both also fail to accurately capture the reflectivity core associated with a portion of the southern complex of cells. The PATHRAD experiment does capture this feature in part because of its better spinup characteristics previously observed (Figs. 7f, 8f).

Fig. 8.
Fig. 8.

As in Fig. 6, but at 1930 UTC.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

The convection continues to intensify and grow in coverage during the next 30 min, resulting in several discrete supercells evident by the final assimilation cycle at 2000 UTC. The CNTL experiment finally begins to generate CWP and reflectivity in the vicinity of the observed convection, but remains far too weak and poorly located (Figs. 9b, 10b). PATH generates simulated CWP that very closely matches observations, except that the magnitudes near the storm cores are higher because of the previously mentioned saturation problem (Fig. 9c). RADP and RAD0 perform well, but fail to capture the eastward extent of the CWP = 1.0 kg m−2 boundary associated with the outflow region each storm (Figs. 9d,e). RADP also continues to generate false CWP farther east in the domain. The PATHRAD experiment combines the CWP features from PATH while maintaining the higher CWP cores evident on RADP and RAD0 to generate a very good representation of observed CWP (Fig. 9f).

Fig. 9.
Fig. 9.

As in Fig. 5, but at 2000 UTC. Lines through southern storm indicate location of cross sections shown in Fig. 11.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

Fig. 10.
Fig. 10.

As in Fig. 6, but at 2000 UTC.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

Comparison against radar reflectivity paints a somewhat different picture. PATH generates >50 dBZ reflectivity cores associated with the two southern supercells, but places them too far east (Fig. 10c). It also fails to capture the intensity of the northern storms. Both RADP and RAD0 generate a much more realistic reflectivity analysis with the latter showing no remaining spurious reflectivity (Figs. 10d,e). The PATHRAD experiment is similar to RAD0, but with ~20 dBZ reflectivity spread slightly farther east from the cores (Fig. 10f).

The impacts of satellite data assimilation can also be observed by comparing the vertical cross section of the model hydrometeor variables through an individual supercell. The west-to-east cross section was created for the southernmost of the three primary supercells and encompasses the storm core as well as upper-level outflow region that extends eastward (Fig. 9). Cross sections are generated for the four experiments that assimilate radar and/or satellite data (PATH, RADP, RAD0, PATHRAD). The PATH experiment generates a deep core of qc, qg, and qr mixing ratios >1.0 g kg−1 near 98.5°W with qi + qs > 0.8 g kg−1 extending from the storm top at ~300 hPa eastward corresponding to the location of the cirrus outflow region (Fig. 11a). One interesting feature of this experiment is that high qr mixing ratios do not extend to the surface, indicating the satellite data have difficulty resolving low-level features of the ongoing convection. The RADP experiment generates higher mixing ratios for qc, qr, and qg over a larger area, with qr > 0.8 g kg−1 now extending all the way to the surface (Fig. 11b). Near the storm top, qi + qs is greater, but the eastward extent of qi + qs > 0.4 g kg−1 only reaches 98°W. Recall that the observed CWP is still in excess of 1.5 kg m−2 here and farther eastward, indicating the RADP experiment is not capturing the storm outflow well, consistent with the results shown in Fig. 9. RAD0 closely flows RADP in the storm core region but has even lower qi + qs mixing ratios aloft east of the core (Fig. 11c). Assimilating the clear-air reflectivity is having the undesired effect of reducing frozen hydrometeor concentrations corresponding to upper-level clouds that do exist in observations. This increases the amount of solar radiation reaching the surface in the model compared to that observed, increasing error (see following section). The PATHRAD experiment retains the high mixing ratio values of qc, qg, and qr in the core region and combines that with a more realistic representation of the cirrus outflow aloft (Fig. 11d). The area of qr + qi > 1.0 g kg−1 extends much farther eastward, better matching the satellite CWP retrievals. This difference can be important in downstream storm evolution, as the additional cloud cover in the model will impact incoming solar and outgoing terrestrial radiation characteristics and change the near-storm environment compared to a model that does not properly capture this feature.

Fig. 11.
Fig. 11.

Cross section of ensemble mean hydrometeor mixing ratios for (a) PATH, (b) RADP, (c) RAD0, and (d) PATHRAD at 2000 UTC. Filled contours represent qc mixing ratio, line contours represent qi + qs mixing ratio, black hatched areas represent the region where qg > 1.0 g kg−1, and red hatched areas represent the region where qr > 1.0 g kg−1.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

8. Forecast results

a. 1930 UTC initialization

The true test of assimilating both satellite and radar data is to measure the forecast impact relative to only assimilating each of these datasets separately. The 90-min forecasts are initiated at 1930 and 2000 UTC. By 1930 UTC, the experiments have been cycling at 5-min intervals for 45 min and reach the stage where each has spun up deep convection (Figs. 7, 8). Initiating the forecast at this time allows for the study of the importance of spinup characteristics from assimilating both radar and satellite data translating into the forecast evolution convection. To assess the impact of assimilating both satellite and radar data throughout the entire 90-min period, the probability of reflectivity >45 dBZ between 1930 and 2100 UTC is calculated for PATH, RADP, RAD0, and PATHRAD and compared against the track of observed 45-dBZ reflectivity over the same time period (Fig. 12). The probability is calculated by determining the number of ensemble members that generated reflectivity >45 dBZ at a given location at a specific time. Then, the maximum probability at that location over all forecast times is taken and used to generate the plots. PATH generates >80% probabilities associated with the convection present in the southern portion of the domain while being too fast with high probabilities extending well eastward of the end of the observed 45-dBZ path (Fig. 12a). All the satellite and radar data assimilating experiments have an eastward bias in the forecast reflectivity compared to observations. An analysis of the synoptic conditions found that the 500–200-hPa wind speeds in the model analysis are ~5 m s−1 too high compared to radiosonde observations. The wind speed bias results in faster storm motion in the model leading to the location bias observed in the forecast.

Fig. 12.
Fig. 12.

(a)–(d) Probability of simulated 4-km reflectivity from each experiment >45 dBZ for the 90-min forecast period between 1930 and 2100 UTC. Hatched area indicates the region of observed WSR-88D reflectivity >45 dBZ.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

The northern storms are also poorly captured, but recall that PATH had difficulty in generating high reflectivity cores, even though it did show evidence of development. Both RADP and RAD0 generated probabilities near 100% initially, with some storm tracks being captured all the way out to 2100 UTC (Figs. 12b,c). Interestingly, RADP has difficulty retaining the northern storm compared to RAD0, while RAD0 has slightly lower probabilities associated with the southern supercell from 1930 UTC. PATHRAD is generally similar in nature to RAD0, but has several key differences. First, the higher probabilities exist for a longer period of time with the storm beginning near 36.0°N and 99.0°W (Fig. 12d). The probabilities for the southern complex are similar, but PATHRAD does not have a 70% tongue extending out too far northeast, as present in RAD0. Finally, PATHRAD generates higher probabilities with developing convection in far southern OK than either of the radar-only experiments. From an objective standpoint, the PATHRAD experiment is generally superior, though the improvement is small compared to either RADP or RAD0.

To better quantify the difference observed between experiments, skill scores including the probability of detection (POD), false alarm rate (FAR), and Heidke skill score (HSS) are computed for both simulated ensemble mean CWP and radar reflectivity (Wilks 2006). For CWP, if the experiment ensemble mean generates CWP > 1.0 kg m−2 within ±6 km of a GOES-13 CWP retrieval >1.0 kg m−2 at a particular time, then this is considered a “hit.” If the experiment generates CWP > 1.0 kg m−2 and this threshold is not exceeded in the retrievals, then it is considered a false detection. Finally, if neither the retrievals nor the experiment exceeds this threshold, then it is considered a correct null forecast. Skill scores for reflectivity are computed in a similar manner, using a threshold of 35 dBZ. The goal is to generate an analysis where hits are maximized, resulting in a high POD, but false detections are limited, thereby resulting in a low FAR. The HSS takes both into account to generate a statistic indicating the overall skill of each experiment at forecasting both CWP and reflectivity.

Figure 13 shows POD, FAR, and HSS from 4-km forecast reflectivity for each experiment (Figs. 13a–c). CNTL essentially has no skill, while the rest of the experiments are clearly better. PATH generates lower POD and higher FAR than the radar experiment since PATH is unable to analyze the high reflectivity of the storm core and spreads out the information over a larger area. RADP and RAD0 perform similarly, but the former has a higher FAR and POD during the first 15 min, resulting in a higher overall HSS. PATHRAD generates a higher POD than RADP during this time while also retaining the lower FAR produced by RAD0. The difference between the radar assimilation experiments and PATHRAD decreases as a function of time, but PATHRAD consistently has the higher POD and HSS skill scores for a 45-min forecast out to 2015 UTC. Thereafter, PATHRAD and RAP have similar skill. One interesting result from the skill score analysis is that RADP outperforms RAD0 initially and during the latter half of the forecast period despite not having the advantage of suppressing spurious convection. On the other hand, PATHRAD outperforms RAD0 at all forecast times. This may indicate that assimilating clear-air radar reflectivity may be inferior to assimilating CWP information, suggesting some combination of the observations may be useful.

Fig. 13.
Fig. 13.

Skill scores including (a) POD, (b) FAR, and (c) HSS for 4-km ensemble mean reflectivity for the 90-min forecast period between 1930 and 2100 UTC. Skill scores are computed using an observed radar reflectivity threshold of 35 dBZ. (d) POD, (e) FAR, and (f) HSS for ensemble mean CWP for the 90-min forecast period between 1930 and 2100 UTC computed using an observed CWP threshold of 1.0 kg m−2.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

The same set of skill scores is compared between GOES-13 CWP retrievals and simulated CWP for each experiment (Figs. 13d–f). For all forecast times out to 2045 UTC, the PATHRAD experiment produces a higher POD than either of the radar experiments or the PATH experiment alone (Fig. 13d). No satellite data are available between 2045 and 2115 UTC, preventing the calculation of skill scores during that period. RAD0 has the lowest POD 30 min into the forecast as clear-air reflectivity assimilation corresponds to the lower CWP values generated in the anvil of the storm at 1930 UTC, as shown in Fig. 7. With the exception of 1935 UTC, the RADP experiment consistently has the highest FAR, which is a result of the spurious cloud cover not being suppressed in the initial analysis and maintained as the forecast progresses (Fig. 13d). Overall, HSS indicates that PATHRAD is the best performer at correctly forecasting the CWP at all forecast times out to 2045 UTC (Fig. 13e). Both RADP and RAD0 are similar out to 2000 UTC with RADP HSS decreasing relative to RAD0 and RADP thereafter.

Since neither reflectivity nor CWP offer a true independent test of model skill, forecast downward shortwave flux (SWDOWN) is compared against measurements from OK Mesonet data. These data are not being assimilated and are thus completely independent of the model analysis. SWDOWN refers to the amount of solar radiation reaching the surface. Lower values correspond to the presence of cloud cover while higher values mean less cloud cover to clear-sky conditions. Lower SWDOWN errors indicate that a particular experiment is handling the location and thickness of clouds better than an experiment with higher errors. At 1930 UTC, root-mean-square error (RMSE) for PATH is ~50 W m−2 less that both RADP and RAD0 (Fig. 14a). PATHRAD performs even better being ~80 W m−2 less than the radar-only experiments. This improvement decreases over the next 20 min until all experiments except CNTL have similar RMSE. Of the two radar-only experiments, RAD0 is generally the better performer as it reduces cloud cover that RADP maintains in the analysis and ensuing forecasts.

Fig. 14.
Fig. 14.

SWDOWN RMSE calculated between ensemble mean forecasts and OK Mesonet observations from 87 sites for (a) the 1930–2100 UTC forecast period and (b) the 2000–2130 UTC forecast period.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

b. 2000 UTC initialization

Forecasts initiated at 2000 UTC represent the culmination of 115 min of 5-min cycling with the inherent advantage of having more time to spin up organized convection. Figure 15 shows 4-km reflectivity forecast probabilities between 2000 and 2130 UTC for the PATH, RADP, RAD0, and PATHRAD experiments with contours of >45 dBZ reflectivity during this period overlaid. The PATH experiment has difficulty generating high reflectivity probabilities near the initial analyzed storm locations in the northern half of the domain, but probabilities generally increase as the model generates higher reflectivity values than those being generated from the satellite data and its corresponding saturation limitation (Fig. 15a). The advantages of assimilating radar data are clear for RADP and RAD0, which both generate reflectivity probabilities near 100% nearby the 2000 UTC location of most cells (Figs. 15b,c). The primary difference between RADP and RAD0 is that the latter reduces spurious high probability in between northern cells and maintains higher probabilities associated with the storm initially located near 35.2°N and 98.6°W. The PATHRAD experiment is generally similar to RAD0 in most areas (Fig. 15d). In southern OK, PATHRAD generates higher reflectivity probabilities corresponding to the storms that develop during the forecast period, but otherwise, no significant differences are apparent. An objective analysis of these figures indicate that assimilating CWP in addition to radar data has only limited impact on predicted reflectivity for forecasts initiated at 2000 UTC.

Fig. 15.
Fig. 15.

(a)‒(d) Probability of simulated 4-km reflectivity from each experiment >45 dBZ for the 90-min forecast period between 2000 and 2130 UTC. Hatched area indicates the region of observed WSR-88D reflectivity >45 dBZ.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

This interpretation is also evident when comparing the 4-km reflectivity forecast skill scores for each experiment (Figs. 16a–c). As expected, CNTL has much lower POD and HSS values than the other experiments for the first hour of the forecast period. The PATH experiment does better, but POD remains lower and FAR higher than either RADP or RAD0. Still, PATH remains skillful compared to assimilating no remote sensing observations. RADP and RAD0 are similar, though RADP generates slightly higher POD and HSS scores than does RAD0. For the first 30 min of the forecast period, POD and HSS are highest for RADP, though it comes with a somewhat higher FAR. Assimilating clear-air reflectivity reduces forecast reflectivity potentially to a point of being undesirable. RAD0 has higher HSS values in the 30–60-min forecast period primarily due to having lower FAR values. The PATHRAD experiment is slightly less skillful than RADP for the first 30 min, but it does have higher POD and HSS scores than RAD0, while also having a FAR very similar to RADP. Afterward, PATHRAD closely follows RADP for the remainder of the forecast period. These statistics indicate that assimilating CWP retrievals in combination with radar data once convection has become well organized and widespread does not add significant skill compared to assimilating radar data alone. However, assimilating CWP does appear to remove potential issues seen when assimilating clear-air reflectivity for the 0–30-min forecast period. Given the much lower cost of assimilating CWP versus clear-air reflectivity, this is seen as an important result.

Fig. 16.
Fig. 16.

As in Fig. 13, but for 90-min forecasts initiated at 2000 UTC.

Citation: Monthly Weather Review 143, 1; 10.1175/MWR-D-14-00180.1

The same set of skill scores was also computed against GOES-13 CWP retrievals when available for the forecast period (Figs. 16d–f). PATH generates very high POD (>0.7) and low FAR (<0.2) for CWP during the first 30-min forecast period. RAD0 and RADP perform much worse, with RADP having the advantage over RAD0 since the former is not artificially reducing cloud cover because of clear-air reflectivity assimilation. The PATHRAD experiment is generally the best performer in terms of POD and HSS. Combined with the radar reflectivity skill scores, these results indicate that assimilating CWP, in combination with radar data, does not have any significant negative impact on reflectivity forecast and potentially has additional advantages over clear-air radar reflectivity while also greatly improving the skill relative to observed CWP retrievals. For both radar reflectivity and CWP, the skill relative to CNTL becomes small by the end of the forecast period, indicating that neither dataset provides any quantitative skill beyond this point.

As with the 1930 UTC set of forecasts, RMSE was computed against Mesonet SWDOWN for the 2000–2130 UTC forecast as an independent assessment of skill between each experiment (Fig. 14b). Both PATH and PATHRAD generate lower RMSE compared to RADP and especially RAD0 for the first 20 min of the forecast period. Afterward, RMSE converges, with RADP and PATHRAD having similar values for the remainder of the forecast period. RAD0 consistently generates higher SWDOWN errors out to 1 h, which is consistent with the previous results showing that assimilating clear-air reflectivity may be negatively impacting model cloud properties, which in turn impacts the radiation balance within the model.

9. Conclusions

Assimilating both high-resolution satellite and radar data into a convection-permitting model using an EaKF approach has proven viable despite the many challenges required to be overcome to allow satellite and radar data to work together and not against each other. The advantages of assimilating both types are evident in several comparisons of analyses and forecasts against corresponding observations.

High-resolution satellite retrievals of cloud properties are able to successfully initiate and maintain convection when assimilated into a convection-permitting model. Given the 2D nature of the observations, the resulting storm structure is rather ambiguous and smoothed out, but assimilating satellite data provided significantly improved forecast skill over those forecasts generated from no assimilation of any high-resolution dataset. In situations where resources are limited, assimilating satellite data may represent a viable alternative to radar data. Assimilating CWP also generally outperformed assimilating clear-air reflectivity, even when not considering resource requirements. The experiment that assimilated clear-air reflectivity, RAD0, consistently underanalyzed and underforecast atmospheric CWP content compared to observations. Clear-air reflectivity does have the desired impact of reducing spurious convection early in the assimilation cycle, but again, CWP generally does better while having a much smaller set of observations. For the forecasts initiated at 2000 UTC, assimilating clear-air reflectivity actually reduced reflectivity forecast skill out to 30 min relative to assimilating positive reflectivity observations alone. Therefore, at least for this case, it was faster and more accurate to assimilate CWP away from convective cores rather than assimilating clear-air reflectivity.

The impacts of assimilating CWP in combination with radar data appear to be greatest early in the analysis period when suppressing spurious convection and capturing convective initiation. The combined experiment spun up high-reflectivity cores faster than either the satellite or radar data could alone. By 1930 UTC, this resulted in a much better representation of convection within the model in the PATHRAD experiment. The positive impacts of assimilating CWP appear to decrease somewhat as convection matures and retrievals become saturated. Still, assimilating CWP continues to improve the analysis and forecast of anvil-like features in the model and remaining convection-free regions within the model domain. Surface shortwave flux analyses are also improved, but the improvement does not necessarily carry over to other surface conditions such as accumulated rainfall. Overall, combining satellite and radar data produces the best analysis and forecast when comparing all radar, satellite, and surface observations for this particular case study.

Further testing of the methods employed here on additional cases and convective modes is required for the conclusions to be generalized. Research is ongoing to develop a prototype ensemble-based data assimilation system that uses methods similar to those employed here. It is hoped that this prototype system will evolve into a robust system that assimilates the best of both radar and satellite data for storm-scale applications. Early testing of this prototype showed that it could be done using modest computing resources, and the initial results are consistent with those presented here. Further into the future, research will begin to take advantage of the additional products soon to be available from the GOES-R ABI and the recently available dual polarimetric upgrade of the WSR-88Ds.

Acknowledgments

This research was supported by the NOAA National Environmental Satellite, Data, and Information Service as part of the GOES-R program. Partial funding for this research was also provided by NOAA/Office of Oceanic and Atmospheric Research under NOAA–University of Oklahoma Cooperative Agreement NA11OAR4320072, under the U.S. Department of Commerce. P. Minnis and R. Palikonda are supported by the NASA Modeling, Analysis, and Prediction (MAP) Program and by the Department of Energy Atmospheric Science Research Program under Interagency Agreement DE-SC0000991/006. The near-real-time satellite analyses can be accessed for a variety of domains at http://cloudsgate2.larc.nasa.gov/. The computing for this project was performed at the OU Supercomputing Center for Education and Research (OSCER) at the University of Oklahoma (OU).

REFERENCES

  • Aksoy, A., , D. Dowell, , and C. Snyder, 2009: A multicase comparative assessment of the ensemble Kalman filter for assimilation of radar observations. Part I: Storm-scale analyses. Mon. Wea. Rev., 137, 18051824, doi:10.1175/2008MWR2691.1.

    • Search Google Scholar
    • Export Citation
  • Aksoy, A., , D. Dowell, , and C. Snyder, 2010: A multicase comparative assessment of the ensemble Kalman filter for assimilation of radar observations. Part II: Short-range ensemble forecasts. Mon. Wea. Rev., 138, 12731292, doi:10.1175/2009MWR3086.1.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., 2001: An ensemble adjustment filter for data assimilation. Mon. Wea. Rev., 129, 28842903, doi:10.1175/1520-0493(2001)129<2884:AEAKFF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., 2007: An adaptive covariance inflation error correction algorithm for ensemble filters. Tellus, 59A, 210224, doi:10.1111/j.1600-0870.2006.00216.x.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., 2009: Spatially and temporally varying adaptive covariance inflation for ensemble filters. Tellus, 61A, 7283, doi:10.1111/j.1600-0870.2008.00361.x.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., , and N. Collins, 2007: Scalable implementations of ensemble filter algorithms for data assimilation. J. Atmos. Oceanic Technol., 24, 14521463, doi:10.1175/JTECH2049.1.

    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., , T. Hoar, , K. Raeder, , H. Liu, , N. Collins, , R. Torn, , and A. Avellano, 2009: The Data Assimilation Research Testbed: A community facility. Bull. Amer. Meteor. Soc., 90, 12831296, doi:10.1175/2009BAMS2618.1.

    • Search Google Scholar
    • Export Citation
  • Carrio, G. G., , W. R. Cotton, , D. Zupanski, , and M. Zupanski, 2008: Development of an aerosol retrieval method: Description and preliminary tests. J. Appl. Meteor. Climatol., 47, 27602776, doi:10.1175/2008JAMC1729.1.

    • Search Google Scholar
    • Export Citation
  • Chaboureau, J.-P., , and J.-P. Pinty, 2006: Validation of a cirrus parameterization with Meteosat second generation observations. Geophys. Res. Lett., 33, L03815, doi:10.1029/2005GL024725.

    • Search Google Scholar
    • Export Citation
  • Chang, F.-L., , P. Minnis, , B. Lin, , M. Khaiyer, , R. Palikonda, , and D. Spangenberg, 2010: A modified method for inferring cloud top height using GOES 12 imager 10.7 and 13.3 μm data. J. Geophys. Res., 115, D06208, doi:10.1029/2009JD012304.

    • Search Google Scholar
    • Export Citation
  • Chen, Y., , F. Weng, , Y. Han, , and Q. Liu, 2008: Validation of the Community Radiative Transfer Model by using CloudSat data. J. Geophys. Res.,113, D00A03, doi:10.1029/2007JD009561.

  • Crum, T. D., , and R. L. Alberty, 1993: The WSR-88D and the WSR-88D Operational Support Facility. Bull. Amer. Meteor. Soc., 74, 16691687, doi:10.1175/1520-0477(1993)074<1669:TWATWO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Dawson, D. T., II, , L. J. Wicker, , E. R. Mansell, , and R. L. Tanamachi, 2012: Impact of the environmental low-level wind profile on ensemble forecasts of the 4 May 2007 Greensburg, Kansas, tornadic storm and associated mesocyclones. Mon. Wea. Rev., 140, 696716, doi:10.1175/MWR-D-11-00008.1.

    • Search Google Scholar
    • Export Citation
  • Derber, J. C., , and W.-S. Wu, 1998: The use of TOVS cloud-cleared radiances in the NCEP SSI analysis system. Mon. Wea. Rev.,126, 2287–2299, doi:10.1175/1520-0493(1998)126<2287:TUOTCC>2.0.CO;2.

  • Dowell, D. C., , and L. J. Wicker, 2009: Additive noise for storm-scale ensemble forecasting and data assimilation. J. Atmos. Oceanic Technol., 26, 911927, doi:10.1175/2008JTECHA1156.1.

    • Search Google Scholar
    • Export Citation
  • Dowell, D. C., , F. Zhang, , L. J. Wicker, , C. Snyder, , and N. A. Crook, 2004: Wind and temperature retrievals in the 17 May 1981 Arcadia, Oklahoma, supercell: Ensemble Kalman filter experiments. Mon. Wea. Rev., 132, 19822005, doi:10.1175/1520-0493(2004)132<1982:WATRIT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Dowell, D. C., , L. J. Wicker, , and C. Snyder, 2011: Ensemble Kalman filter assimilation of radar observations of the 8 May 2003 Oklahoma City supercell: Influences of reflectivity observations on storm-scale analyses. Mon. Wea. Rev., 139, 272294, doi:10.1175/2010MWR3438.1.

    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46, 30773107, doi:10.1175/1520-0469(1989)046<3077:NSOCOD>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Eilts, M. D., , and S. D. Smith, 1990: Efficient dealiasing of Doppler velocities using local environment constraints. J. Atmos. Oceanic Technol., 7, 118128, doi:10.1175/1520-0426(1990)007<0118:EDODVU>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Ek, M. B., , K. E. Mitchell, , Y. Lin, , E. Rogers, , P. Grunmann, , V. Koren, , G. Gayno, , and J. D. Tarpley, 2003: Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model. J. Geophys. Res., 108, 8851, doi:10.1029/2002JD003296.

    • Search Google Scholar
    • Export Citation
  • Errico, R. M., 2000: The dynamical balance of leading singular vectors in a primitive-equation model. Quart. J. Roy. Meteor. Soc.,126, 1601–1618, doi:10.1002/qj.49712656603.

  • Fujita, T., , D. J. Stensrud, , and D. C. Dowell, 2007: Surface data assimilation using an ensemble Kalman filter approach with initial condition and model physics uncertainties. Mon. Wea. Rev., 135, 18461868, doi:10.1175/MWR3391.1.

    • Search Google Scholar
    • Export Citation
  • Gao, J., , and D. J. Stensrud, 2012: Assimilation of reflectivity data in a convective-scale, cycled 3DVAR framework with hydrometeor classification. J. Atmos. Sci., 69, 10541065, doi:10.1175/JAS-D-11-0162.1.

    • Search Google Scholar
    • Export Citation
  • Gao, J., , M. Xue, , K. Brewster, , and K. K. Droegemeier, 2004: A three-dimensional variational data assimilation method with recursive filter for single-Doppler radar. J. Atmos. Oceanic Technol., 21, 457469, doi:10.1175/1520-0426(2004)021<0457:ATVDAM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Gaspari, G., , and S. E. Cohn, 1999: Construction of correlation functions in two and three dimensions. Quart. J. Roy. Meteor. Soc., 125, 723757, doi:10.1002/qj.49712555417.

    • Search Google Scholar
    • Export Citation
  • Grasso, L. D., , and T. Greenwald, 2004: Analysis of 10.7 μm brightness temperatures of a simulated thunderstorm with two-moment microphysics. Mon. Wea. Rev., 132, 815825, doi:10.1175/1520-0493(2004)132<0815:AOMBTO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Grell, G. A., , and D. Devenyi, 2002: A generalized approach to parameterizing convection combining ensemble and data assimilation techniques. Geophys. Res. Lett.,29, 1693, doi:10.1029/2002GL015311.

  • Hong, S.-Y., , Y. Noh, , and J. Dudhia, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 23182341, doi:10.1175/MWR3199.1.

    • Search Google Scholar
    • Export Citation
  • Iacono, M. J., , J. S. Delamere, , E. J. Mlawer, , M. W. Shephard, , S. A. Clough, , and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, doi:10.1029/2008JD009944.

  • Janjić, Z. I., 1994: The step-mountain eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev.,122, 927–945, doi:10.1175/1520-0493(1994)122<0927:TSMECM>2.0.CO;2.

  • Jones, T. A., , J. Otkin, , D. J. Stensrud, , and K. Knopfmeier, 2013a: Assimilation of satellite infrared radiances and Doppler radar observations during a cool season Observing System Simulation Experiment. Mon. Wea. Rev., 141, 32733299, doi:10.1175/MWR-D-12-00267.1.

    • Search Google Scholar
    • Export Citation
  • Jones, T. A., , D. J. Stensrud, , P. Minnis, , and R. Palikonda, 2013b: Evaluation of a forward operator to assimilate cloud water path into WRF-DART. Mon. Wea. Rev., 141, 22722289, doi:10.1175/MWR-D-12-00238.1.

    • Search Google Scholar
    • Export Citation
  • Jones, T. A., , J. Otkin, , D. J. Stensrud, , and K. Knopfmeier, 2014: Forecast evaluation of an Observing System Simulation Experiment assimilating both radar and satellite data. Mon. Wea. Rev., 142, 107124, doi:10.1175/MWR-D-13-00151.1.

    • Search Google Scholar
    • Export Citation
  • Jung, Y., , M. Xue, , and M. Tong, 2012: Ensemble Kalman filter analyses of the 29–30 May 2004 Oklahoma tornadic thunderstorm using one- and two-moment bulk microphysics schemes, with verification against polarimetric radar data. Mon. Wea. Rev., 140, 14571475, doi:10.1175/MWR-D-11-00032.1.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., , and J. M. Fritsch, 1993: Convective parameterization for mesoscale models: The Kain–Fritsch scheme. The Representation of Cumulus Convection in Numerical Models, Meteor. Monogr., No. 24, Amer. Meteor. Soc., 165–170.

  • Kain, J. S., , S. R. Dembek, , S. J. Weiss, , J. L. Case, , J. J. Levit, , and R. A. Sobash, 2010: Extracting unique information from high-resolution forecast models: Monitoring selected fields and phenomena every time step. Wea. Forecasting, 25, 15361542, doi:10.1175/2010WAF2222430.1.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., and et al. , 2013: A feasibility study for probabilistic convection initiation forecasts based on explicit numerical guidance. Bull. Amer. Meteor. Soc., 94, 12131225, doi:10.1175/BAMS-D-11-00264.1.

    • Search Google Scholar
    • Export Citation
  • Kalman, R. E., 1960: A new approach to linear filtering and prediction problems. J. Basic Eng., 82, 3545, doi:10.1115/1.3662552.

  • Keil, C., , A. Tafferner, , H. Mannstein, , and U. Schättler, 2003: Evaluating high-resolution model forecasts of European winter storms by use of satellite and radar observations. Wea. Forecasting, 18, 732747, doi:10.1175/1520-0434(2003)018<0732:EHMFOE>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Lakshmanan, V., , K. Hondl, , G. Stumpf, , and T. Smith, 2003: Quality control of WSR-88D data. 31st Radar Conf., Seattle, WA, Amer. Meteor. Soc., 7.1. [Available online at https://ams.confex.com/ams/pdfpapers/62597.pdf.]

  • Lin, Y.-L., , R. D. Farley, , and H. D. Orville, 1983: Bulk parameterization of the snow field in a cloud model. J. Climate Appl. Meteor.,22, 1065–1092, doi:10.1175/1520-0450(1983)022<1065:BPOTSF>2.0.CO;2.

  • Lu, H., , and Q. Xu, 2009: Trade-offs between observation accuracy and resolutions in configuring phased-array radar velocity scans for ensemble-based storm-scale data assimilation. J. Appl. Meteor. Climatol., 48, 12301244, doi:10.1175/2008JAMC2009.1.

    • Search Google Scholar
    • Export Citation
  • Majcen, M., , P. Markowski, , Y. Richardson, , D. Dowell, , and J. Wurman, 2008: Multipass objective analyses of Doppler radar data. J. Atmos. Oceanic Technol., 25, 18451858, doi:10.1175/2008JTECHA1089.1.

    • Search Google Scholar
    • Export Citation
  • Mansell, E. R., 2008: EnKF analysis and forecast predictability of a tornadic supercell storm. 24th Conf. on Severe Local Storms, Savannah, GA, Amer. Meteor. Soc., P5.2. [Available online at https://ams.confex.com/ams/24SLS/techprogram/paper_142059.htm.]

  • McNally, A. P., , P. D. Watts, , J. A. Smith, , R. Engelen, , G. A. Kelly, , J. N. Thepaut, , and M. Matricardi, 2006: The assimilation of AIRS radiance data at ECMWF. Quart. J. Roy. Meteor. Soc.,132, 935–957, doi:10.1256/qj.04.171.

  • McPherson, R. A., and et al. , 2007: Statewide monitoring of the mesoscale environment: A technical update on the Oklahoma Mesonet. J. Atmos. Oceanic Technol., 24, 301321, doi:10.1175/JTECH1976.1.

    • Search Google Scholar
    • Export Citation
  • Mecikalski, J. R., , P. Minnis, , and R. Palikonda, 2013: Use of satellite derived cloud properties to quantify growing cumulus beneath cirrus clouds. Atmos. Res., 120–121, 192201, doi:10.1016/j.atmosres.2012.08.017.

    • Search Google Scholar
    • Export Citation
  • Menzel, W. P., , and J. F. Purdom, 1994: Introducing GOES-I: The first of a new generation of geostationary operational environmental satellites. Bull. Amer. Meteor. Soc., 75, 757781, doi:10.1175/1520-0477(1994)075<0757:IGITFO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Migliorini, S., 2012: On the equivalence between radiance and retrieval assimilation. Mon. Wea. Rev.,140, 258–265, doi:10.1175/MWR-D-10-05047.1.

  • Milbrandt, J. A., , and M. K. Yau, 2005a: A multimoment bulk microphysics parameterization. Part I: Analysis of the role of the spectral shape parameter. J. Atmos. Sci., 62, 30513064, doi:10.1175/JAS3534.1.

    • Search Google Scholar
    • Export Citation
  • Milbrandt, J. A., , and M. K. Yau, 2005b: A multimoment bulk microphysics parameterization. Part II: A proposed three-moment closure and scheme description. J. Atmos. Sci., 62, 30653081, doi:10.1175/JAS3535.1.

    • Search Google Scholar
    • Export Citation
  • Minnis, P., , J. Huang, , B. Lin, , Y. Yi, , R. F. Arduini, , T.-F. Fan, , J. K. Ayers, , and G. G. Mace, 2007: Ice cloud properties in ice-over-water cloud systems using Tropical Rainfall Measuring Mission (TRMM) visible and infrared scanner and TRMM Microwave Imager data. J. Geophys. Res., 112, D06206, doi:10.1029/2006JD007626.

    • Search Google Scholar
    • Export Citation
  • Minnis, P., and et al. , 2008a: Near-real time cloud retrievals from operational and research meteorological satellites. Remote Sensing of Clouds and the Atmosphere XIII, R. H. Picard et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 7107), 710703, doi:10.1117/12.800344.

  • Minnis, P., and et al. , 2008b: Cloud detection in non-polar regions for CERES using TRMM VIRS and Terra and Aqua MODIS data. IEEE Trans. Geosci. Remote Sens., 46, 38573884, doi:10.1109/TGRS.2008.2001351.

    • Search Google Scholar
    • Export Citation
  • Minnis, P., and et al. , 2011: CERES Edition-2 cloud property retrievals using TRMM VIRS and Terra and Aqua MODIS data, Part I: Algorithms. IEEE Trans. Geosci. Remote Sens., 49, 43744400, doi:10.1109/TGRS.2011.2144601.

    • Search Google Scholar
    • Export Citation
  • Mlawer, E. J., , S. J. Taubman, , P. D. Brown, , M. J. Iacono, , and S. A. Clough, 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102, 16 66316 682, doi:10.1029/97JD00237.

    • Search Google Scholar
    • Export Citation
  • Nakanishi, M., , and H. Niino, 2006: An improved Mellor–Yamada level-3 model: Its numerical stability and application to a regional prediction of advection fog. Bound.-Layer Meteor., 119, 397407, doi:10.1007/s10546-005-9030-8.

    • Search Google Scholar
    • Export Citation
  • Otkin, J. A., 2010: Clear and cloudy sky infrared brightness temperature assimilation using an ensemble Kalman filter. J. Geophys. Res., 115, D19207, doi:10.1029/2009JD013759.

    • Search Google Scholar
    • Export Citation
  • Otkin, J. A., , and T. Greenwald, 2008: Comparison of WRF model-simulated MODIS-derived cloud data. Mon. Wea. Rev., 136, 19571970, doi:10.1175/2007MWR2293.1.

    • Search Google Scholar
    • Export Citation
  • Pincus, R., , R. J. P. Hofmann, , J. L. Anderson, , K. Raeder, , N. Collins, , and J. S. Whitaker, 2011: Can fully accounting for clouds in data assimilation improve short-term forecast by global models? Mon. Wea. Rev., 139, 946956, doi:10.1175/2010MWR3412.1.

    • Search Google Scholar
    • Export Citation
  • Polkinghorne, R., , and T. Vukicevic, 2011: Data assimilation of cloud affected radiances in a cloud resolving model. Mon. Wea. Rev., 139, 755773, doi:10.1175/2010MWR3360.1.

    • Search Google Scholar
    • Export Citation
  • Polkinghorne, R., , T. Vukicevic, , and F. Evans, 2010: Validation of cloud resolving model background data for data assimilation. Mon. Wea. Rev., 138, 781795, doi:10.1175/2009MWR3012.1.

    • Search Google Scholar
    • Export Citation
  • Schmit, T. J., , E. M. Prins, , A. J. Schreiner, , and J. J. Gurka, 2001: Introducing the GOES-M Imager. Natl. Wea. Dig., 25, 2837. [Available online at www.nwas.org/digest/papers/2001/Vol25No34/Pg28-Schmit.pdf.]

    • Search Google Scholar
    • Export Citation
  • Skamarock, W. C., and et al. , 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp., doi:10.5065/D68S4MVH.

  • Snyder, C., , and F. Zhang, 2003: Assimilation of simulated Doppler radar observations with an ensemble Kalman filter. Mon. Wea. Rev., 131, 16631677, doi:10.1175//2555.1.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., , J.-W. Bao, , and T. T. Warner, 2000: Using initial condition and model physics perturbations in short-range ensemble simulations of mesoscale convective systems. Mon. Wea. Rev.,128, 2077–2107, doi:10.1175/1520-0493(2000)128<2077:UICAMP>2.0.CO;2.

  • Stensrud, D. J., and et al. , 2009: Convective-scale warn-on-forecast system: A vision for 2020. Bull. Amer. Meteor. Soc., 90, 14871499, doi:10.1175/2009BAMS2795.1.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., and et al. , 2013: Progress and challenges with warn-on-forecast. Atmos. Res., 123, 216, doi:10.1016/j.atmosres.2012.04.004.

    • Search Google Scholar
    • Export Citation
  • Thompson, G., , R. M. Rasmussen, , and K. Manning, 2004: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part I: Description and sensitivity analysis. Mon. Wea. Rev., 132, 519542, doi:10.1175/1520-0493(2004)132<0519:EFOWPU>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Thompson, G., , P. R. Field, , R. M. Rasmussen, , and W. R. Hall, 2008: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part II: Implementation of a new snow parameterization. Mon. Wea. Rev., 136, 50955115, doi:10.1175/2008MWR2387.1.

    • Search Google Scholar
    • Export Citation
  • Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117, 17791800, doi:10.1175/1520-0493(1989)117<1779:ACMFSF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Tong, M., , and M. Xue, 2005: Ensemble Kalman filter assimilation of Doppler radar data with a compressible nonhydrostatic model: OSS experiments. Mon. Wea. Rev., 133, 17891807, doi:10.1175/MWR2898.1.

    • Search Google Scholar
    • Export Citation
  • Vukicevic, T., , T. Greenwald, , M. Zupanski, , D. Zupanski, , T. Vonder Haar, , and A. Jones, 2004: Mesoscale cloud state estimation from visible and infrared satellite radiances. Mon. Wea. Rev., 132, 30663077, doi:10.1175/MWR2837.1.

    • Search Google Scholar
    • Export Citation
  • Vukicevic, T., , M. Sengupta, , A. S. Jones, , and T. Vonder Haar, 2006: Cloud-resolving satellite data assimilation: Information content of IR window observations and uncertainties in estimation. J. Atmos. Sci., 63, 901919, doi:10.1175/JAS3639.1.

    • Search Google Scholar
    • Export Citation
  • Wang, C., , and X. Huang, 2014: Parallax correction in the analysis of multiple satellite data sets. IEEE Geosci. Remote Sens. Lett., 11, 965969, doi:10.1109/LGRS.2013.2283573.

    • Search Google Scholar
    • Export Citation
  • Wheatley, D. M., , and D. J. Stensrud, 2010: The impact of assimilating surface pressure observations on severe weather events in a WRF mesoscale ensemble system. Mon. Wea. Rev.,138, 1673–1694, doi:10.1175/2009MWR3042.1.

  • Wilks, D. S., 2006: Statistical Methods in the Atmospheric Sciences. 2nd ed. Academic Press, 627 pp.

  • Yussouf, N., , and D. J. Stensrud, 2010: Impact of phased-array radar observations over a short assimilation period: Observing system simulation experiments using an ensemble Kalman filter. Mon. Wea. Rev., 138, 517538, doi:10.1175/2009MWR2925.1.

    • Search Google Scholar
    • Export Citation
  • Yussouf, N., , E. R. Mansell, , L. J. Wicker, , D. M. Wheatley, , and D. J. Stensrud, 2013: The ensemble Kalman filter analyses and forecasts of the 8 May 2003 Oklahoma City tornadic supercell storm using single- and double-moment microphysics schemes. Mon. Wea. Rev., 141, 33883412, doi:10.1175/MWR-D-12-00237.1.

    • Search Google Scholar
    • Export Citation
  • Zhang, F., , C. Snyder, , and J. Sun, 2004: Impacts of initial estimate and observation availability on convective-scale data assimilation with an ensemble Kalman filter. Mon. Wea. Rev., 132, 12381253, doi:10.1175/1520-0493(2004)132<1238:IOIEAO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Zhang, S. Q., , M. Zupanski, , A. Y. Hou, , X. Lin, , and S. H. Cheung, 2013: Assimilation of precipitation-affected radiances in a cloud-resolving WRF ensemble data assimilation system. Mon. Wea. Rev., 141, 754772, doi:10.1175/MWR-D-12-00055.1.

    • Search Google Scholar
    • Export Citation
  • Zupanski, D., , S. Q. Zhang, , M. Zupanski, , A. Y. Hou, , and S. H. Cheung, 2011: A prototype WRF-based ensemble data assimilation system for dynamically downscaling satellite precipitation observations. J. Hydrometeor., 12, 118134, doi:10.1175/2010JHM1271.1.

    • Search Google Scholar
    • Export Citation
1

OPAWS is freely available from https://code.google.com/p/opaws/.

Save