• Barnes, S. L., 1973: Mesoscale objective map analysis using weighted time series observations. NOAA Tech. Memo. ERL NSSL-62, 60 pp. [NTIS COM-73-10781.]

  • Barnes, S. L., 1994: Applications of the Barnes objective analysis scheme. Part I: Effects of undersampling, wave position, and station randomness. J. Atmos. Oceanic Technol., 11, 14331448.

    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., and Coauthors, 2004a: An hourly assimilation–forecast cycle: The RUC. Mon. Wea. Rev., 132, 495518.

  • Benjamin, S. G., , Grell G. A. , , Brown J. M. , , Smirnova T. G. , , and Bleck R. , 2004b: Mesoscale weather prediction with the RUC hybrid isentropic–terrain-following coordinate model. Mon. Wea. Rev., 132, 473494.

    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., , Weygandt S. , , Devenyi D. , , Brown J. M. , , Manikin G. , , Smith T. L. , , and Smirnova T. , 2004c: Improved moisture and PBL initialization in the RUC using METAR data. Preprints, 11th Conf. on Aviation, Range, and Aerospace Meteorology, Hyannis, MA, Amer. Meteor. Soc., 17.3. [Available online at http://ams.confex.com/ams/pdfpapers/82023.pdf.]

  • Benjamin, S. G., , Jamison B. D. , , Moninger W. R. , , Sahm S. R. , , Schwartz B. E. , , and Schlatter T. W. , 2010: Relative short-range forecast impact from aircraft, profiler, radiosonde, VAD, GPS-PW, METAR, and Mesonet observations via the RUC hourly assimilation cycle. Mon. Wea. Rev., 138, 13191343.

    • Search Google Scholar
    • Export Citation
  • Bothwell, P. D., , Hart J. A. , , and Thompson R. L. , 2002: An integrated three-dimensional objective analysis scheme in use at the Storm Prediction Center. Preprints, 21st Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., JP3.1. [Available online at http://ams.confex.com/ams/SLS_WAF_NWP/techprogram/paper_47482.htm.]

  • Burk, S. D., , and Thompson W. T. , 1989: A vertically nested regional numerical weather prediction model with second-order closure physics. Mon. Wea. Rev., 117, 23052324.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , Olson J. B. , , and Tongue J. S. , 2003: Multiseason verification of the MM5. Part I: Comparison with the Eta Model over the central and eastern United States and impact of MM5 resolution. Wea. Forecasting, 18, 431457.

    • Search Google Scholar
    • Export Citation
  • Coniglio, M. C., , Hwang J. Y. , , and Stensrud D. J. , 2010: Environmental factors in the upscale growth and longevity of MCSs derived from Rapid Update Cycle analyses. Mon. Wea. Rev., 138, 35143539; Corrigendum, 139, 26862688.

    • Search Google Scholar
    • Export Citation
  • Davies, J. M., 2004: Estimations of CIN and LFC associated with tornadic and nontornadic supercells. Wea. Forecasting, 19, 714726.

  • Dean, A. R., , and Schneider R. S. , 2008: Forecast challenges at the NWS Storm Prediction Center relating to the frequency of favorable severe storm environments. Preprints, 24th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 9A.2. [Available online at http://ams.confex.com/ams/24SLS/techprogram/paper_141743.htm.]

  • Devenyi, D., , Weygandt S. , , Schlatter T. , , Benjamin S. , , and Hu M. , 2007: Hourly data assimilation with the gridpoint statistical interpolation for the rapid refresh. Preprints, 22nd Conf. on Weather Analysis and Forecasting/18th Conf. on Numerical Weather Prediction, Park City, UT, Amer. Meteor. Soc., 4A.2. [Available online at http://ams.confex.com/ams/22WAF18NWP/techprogram/paper_124535.htm.]

  • Doswell, C. A., III, , and Rasmussen E. N. , 1994: The effect of neglecting the virtual temperature correction on CAPE calculations. Wea. Forecasting, 9, 625629.

    • Search Google Scholar
    • Export Citation
  • Eager, R. E., , Raman S. , , Holt T. R. , , Westphal D. , , Reid J. , , Nachamkin J. , , Liu M. , , and Mandoos A. A. , 2007: A seasonal statistical evaluation of COAMPS over the Arabian Gulf region. Pure Appl. Geophys., 164, 17471764.

    • Search Google Scholar
    • Export Citation
  • Edwards, R., , Dean A. R. , , Thompson R. L. , , and Smith B. T. , 2010: Objective environmental analyses and convective modes for U.S. tropical cyclone tornadoes from 2003–2008. Preprints, 25th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., P3.2. [Available online at http://ams.confex.com/ams/pdfpapers/175397.pdf.]

  • Evensen, G., 1994: Sequential data assimilation with a nonlinear quasi-geostrophic model using Monte-Carlo methods to forecast error statistics. J. Geophys. Res., 99 (C5), 10 14310 162.

    • Search Google Scholar
    • Export Citation
  • Fabry, F., 2006: The spatial variability of moisture in the boundary layer and its effect on convection initiation: Project-long characterization. Mon. Wea. Rev., 134, 7991.

    • Search Google Scholar
    • Export Citation
  • Fawbush, E. J., , and Miller R. C. , 1954: The types of air masses in which North American tornadoes form. Bull. Amer. Meteor. Soc., 35, 154165.

    • Search Google Scholar
    • Export Citation
  • Fujita, T., , Stensrud D. J. , , and Dowell D. C. , 2007: Surface data assimilation using an ensemble Kalman filter approach with initial condition and model physics uncertainty. Mon. Wea. Rev., 135, 18461868.

    • Search Google Scholar
    • Export Citation
  • Gilleland, E., 2010: Confidence intervals for forecast verification. NCAR Tech. Note NCAR/TN-479+STR, 71 pp.

  • Hacker, J. P., , and Snyder C. , 2005: Ensemble Kalman filter assimilation of fixed screen-height observations in a parameterized PBL. Mon. Wea. Rev., 133, 32603275.

    • Search Google Scholar
    • Export Citation
  • Johns, R. H., , and Doswell C. A. III, 1992: Severe local storms forecasting. Wea. Forecasting, 7, 588612.

  • Kain, J. S., , Weiss S. J. , , Baldwin M. E. , , Carbin G. W. , , Bright D. R. , , Levit J. J. , , and Hart J. A. , 2005: Evaluating high-resolution configurations of the WRF model that are used to forecast severe convective weather: The 2005 SPC/NSSL Spring Program. Preprints, 21st Conf. on Weather Analysis and Forecasting/17th Conf. on Numerical Weather Prediction, Washington, DC, Amer. Meteor., Soc., 2A.5. [Available online at http://ams.confex.com/ams/pdfpapers/94843.pdf.]

  • Kirkpatrick, J. C., , McCaul E. W. , , and Cohen C. , 2007: The motion of simulated convective storms as a function of basic environmental parameters. Mon. Wea. Rev., 135, 30333051.

    • Search Google Scholar
    • Export Citation
  • Kirkpatrick, J. C., , McCaul E. W. , , and Cohen C. , 2009: Variability of updraft and downdraft characteristics in a large parameter space study of convective storms. Mon. Wea. Rev., 137, 15501561.

    • Search Google Scholar
    • Export Citation
  • Loehrer, S. M., , Williams S. F. , , and Moore J. A. , 1998: Results from UCAR/JOSS quality control of atmospheric soundings from field projects. Preprints, 10th Symp. on Meteorological Observations and Instrumentation, Phoenix, AZ, Amer. Meteor. Soc., 1–6.

  • Markowski, P., , Hannon C. , , Frame J. , , Lancaster E. , , Pietrycha A. , , Edwards R. , , and Thompson R. L. , 2003: Characteristics of vertical wind profiles near supercells obtained from the Rapid Update Cycle. Wea. Forecasting, 18, 12621272.

    • Search Google Scholar
    • Export Citation
  • McCaul, E. W., Jr., , and Cohen C. , 2002: The impact on simulated storm structure and intensity of variations in the mixed layer and moist layer depths. Mon. Wea. Rev., 130, 17221748.

    • Search Google Scholar
    • Export Citation
  • Myoung, B., , and Nielsen-Gammon J. W. , 2010: Sensitivity of monthly convective precipitation to environmental conditions. J. Climate, 23, 166188.

    • Search Google Scholar
    • Export Citation
  • Nash, J., , Smout R. , , Oakley T. , , Pathack P. , , and Kurnosenko S. , 2005: WMO intercomparison of high quality radiosonde systems: Vacoas, Mauritius, 2–25 February 2005. WMO Commission on Instruments and Methods of Observation Final Rep., 118 pp. [Available online at http://www.wmo.ch/pages/prog/www/IMOP/reports/2003-2007/RSO-IC-2005_Final_Report.pdf.]

  • Newton, C. W., 1963: Dynamics of severe convective storms. Severe Local Storms, Meteor. Monogr., No. 27, Amer. Meteor. Soc., 33–58.

  • Nowotarski, C. J., , Markowski P. M. , , and Richardson Y. P. , 2011: The characteristics of numerically simulated supercell storms situated over statically stable boundary layers. Mon. Wea. Rev., 139, 31393162.

    • Search Google Scholar
    • Export Citation
  • Orlanski, I., 1975: A rational subdivision of scales for atmospheric processes. Bull. Amer. Meteor. Soc., 56, 527530.

  • Potvin, C. K., , Elmore K. L. , , and Weiss S. J. , 2010: Assessing the impacts of proximity sounding criteria on the climatology of significant tornado environments. Wea. Forecasting, 25, 921930.

    • Search Google Scholar
    • Export Citation
  • Rakesh, V., , Singh R. , , and Joshi P. C. , 2009: Intercomparison of the performance of MM5/WRF with and without satellite data assimilation in short-range forecast applications over the Indian region. Meteor. Atmos. Phys., 105, 133155.

    • Search Google Scholar
    • Export Citation
  • Rasmussen, E. N., , and Blanchard D. O. , 1998: A baseline climatology of sounding-derived supercell and tornado forecast parameters. Wea. Forecasting, 13, 11481164.

    • Search Google Scholar
    • Export Citation
  • Schneider, R. S., , and Dean A. R. , 2008: A comprehensive 5-year severe storm environment climatology for the continental United States. Preprints, 24th Conf. on Severe Local Storms, Savannah, GA, Amer. Meteor. Soc., 16A.4. [Available online at http://ams.confex.com/ams/24SLS/techprogram/paper_141748.htm.]

  • Schumacher, R. S., , and Johnson R. H. , 2009: Quasi-stationary, extreme-rain-producing convective systems associated with midlevel cyclonic circulations. Wea. Forecasting, 24, 555574.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., , Yussouf N. , , Dowell D. C. , , and Coniglio M. C. , 2009a: Assimilating surface data into a mesoscale model ensemble: Cold pool analyses from spring 2007. Atmos. Res., 93, 207220.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., and Coauthors, 2009b: Convective-scale warn-on-forecast system. Bull. Amer. Meteor. Soc., 90, 14871499.

  • Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic, 666 pp.

  • Thompson, R. L., , Hart J. A. , , Elmore K. L. , , and Markowski P. , 2003: Close proximity soundings within supercell environments obtained from the Rapid Update Cycle. Wea. Forecasting, 18, 12431261.

    • Search Google Scholar
    • Export Citation
  • Thompson, R. L., , Edwards R. , , and Mead C. M. , 2004: An update to the supercell composite and significant tornado parameters. Preprints, 22nd Conf. on Severe Local Storms, Hyannis, MA, Amer. Meteor. Soc., P8.1. [Available online at http://ams.confex.com/ams/pdfpapers/82100.pdf.]

  • Thompson, R. L., , Mead C. M. , , and Edwards R. , 2007: Effective storm-relative helicity and bulk shear in supercell thunderstorm environments. Wea. Forecasting, 22, 102115.

    • Search Google Scholar
    • Export Citation
  • Thompson, R. L., , Smith B. T. , , Grams J. S. , , Dean A. R. , , and Broyles C. , 2010: Climatology of near-storm environments with convective modes for significant thunderstorms in the contiguous United States. Preprints, 25th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 16B.6. [Available online at http://ams.confex.com/ams/25SLS/techprogram/paper_175727.htm.]

  • Vasiloff, S. V., and Coauthors, 2007: Improving QPE and very short term QPF: An initiative for a community-wide integrated approach. Bull. Amer. Meteor. Soc., 88, 18991911.

    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., , and Klemp J. B. , 1982: The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev., 110, 504520.

    • Search Google Scholar
    • Export Citation
  • Wheatley, D. M., , and Stensrud D. J. , 2010: The impact of assimilating surface pressure observations on severe weather events in a WRF mesoscale ensemble system. Mon. Wea. Rev., 138, 16731694.

    • Search Google Scholar
    • Export Citation
  • Wilson, J. W., , and Roberts R. D. , 2006: Summary of convective storm initiation and evolution during IHOP: Observational and modeling perspective. Mon. Wea. Rev., 134, 2347.

    • Search Google Scholar
    • Export Citation
  • Wu, W.-S., , Purser R. J. , , and Parrish D. F. , 2002: Three-dimensional variational analysis with spatially inhomogeneous covariances. Mon. Wea. Rev., 130, 29052916.

    • Search Google Scholar
    • Export Citation
  • Wurman, J., , Wicker L. J. , , Richardson Y. P. , , Rasmussen E. N. , , Markowski P. M. , , Dowell D. , , Burgess D. W. , , and Bluestein H. B. , 2010: An overview of the VORTEX2 field campaign. Preprints, 25th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 5.1. [Available online at http://ams.confex.com/ams/25SLS/techprogram/paper_176068.htm.]

  • Ziegler, C. L., , Mansell E. R. , , Straka J. M. , , MacGorman D. R. , , and Burgess D. W. , 2010: The impact of spatial variations of low-level stability on the life cycle of a simulated supercell storm. Mon. Wea. Rev., 138, 17381766.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Tracks of the 40 radiosondes used in this study, from the launch point (denoted by the cross) to the location at which the data terminated (denoted by the closed circle). The numbers correspond to the list of soundings in Table 1.

  • View in gallery

    Distribution of the time elapsed and horizontal distance traversed by the 40 radiosondes. Bins are every 10 min or 10 km.

  • View in gallery

    The rmsds between NWS radiosonde observations (0000 and 1200 UTC) and RUC analyses and forecasts over the entire RUC computational domain for the period 11 Sep–31 Dec 2002, from the study by Benjamin et al. (2004a), for (a) vector wind (m s−1), (b) temperature (K), and (c) relative humidity (%).

  • View in gallery

    Vertical profile of rmsd (m s−1) for the RUC analysis and 1-h forecast (a) vector wind, (b) temperature, and (c) RH. The differences between the rmsd estimates between the RUC analysis and 1-h RUC forecasts are shown by the black line. Error bars denote 95% confidence intervals on the differences in rmsd, with the thick error bars denoting statistically significant differences (i.e., the error bar does not overlap with the zero line) at the 95% level. Profiles of rmsd and their differences are only shown for levels above 875 hPa; fewer than 30 soundings had good data on higher pressure levels.

  • View in gallery

    As in Fig. 4, but for vertical profiles of mean difference (bias) on height levels above ground (m).

  • View in gallery

    Examples of comparisons of observed (black) and 1-h forecast soundings (gray) in which the PBL depth was too shallow in the RUC forecast. The horizontal lines in corresponding black and gray shading indicate the PBL depth determined by the algorithm used in this study. Computed parameters are listed for each of the observed (black text) and forecasted soundings (gray text in parentheses).

  • View in gallery

    Boxplots of PBL height errors (forecast − observation) for the RUC analysis (RUC00), 1-h RUC forecast (RUC01), and the SFCOA. The boxes encompass the interquartile range (25th–75th percentile), the dashed lines extend to the 10th and 90th percentiles, and the median is denoted by a horizontal line inside the box. The filled circle denotes the rmsd between the forecasts and observations and the hollow circle denotes the mean difference (bias) between the forecasts and observations. Error bars on the rmsd and bias estimates indicate 95% confidence intervals. Numbers along the bottom are the confidence (%) that the difference in the rmsd estimates (boldface text) or the difference in the bias estimate (lightface text) is statistically significant between the two data sources in question. For example, the statistical confidence that the RUC00 and RUC01 bias estimates are different is 85.9%, and the statistical confidence that the RUC01 and SFCOA rmsd estimates are different from each other is 93.9%.

  • View in gallery

    As in Fig. 7, but for (a) 2-m temperature and (b) 2-m dewpoint.

  • View in gallery

    As in Fig. 7 but for (a) the lowest 30-hPa average temperature and (b) dewpoint.

  • View in gallery

    As in Fig. 7, but for (a) SBCAPE, (b) MUCAPE, (c) MLCAPE, and (d) LLCAPE (see Table 1 for variable definitions).

  • View in gallery

    As in Fig. 7, but for (a) SBCIN, (b) MUCIN, and (c) MLCIN.

  • View in gallery

    As in Fig. 7, but for (a) SBLCL, (b) MULCL, and (c) MLLCL.

  • View in gallery

    As in Fig. 7, but for (a) SBLFC, (b) MULFC, and (c) MLLFC.

  • View in gallery

    As in Fig. 7, but for (a) SCP and (b) STP.

  • View in gallery

    As in Fig. 7, but for (a) effective BWD (EBWD) and (b) effective (ESRH).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 103 103 10
PDF Downloads 74 74 11

Verification of RUC 0–1-h Forecasts and SPC Mesoscale Analyses Using VORTEX2 Soundings

View More View Less
  • 1 NOAA/National Severe Storms Laboratory, Norman, Oklahoma
© Get Permissions
Full access

Abstract

This study uses radiosonde observations obtained during the second phase of the Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX2) to verify base-state variables and severe-weather-related parameters calculated from Rapid Update Cycle (RUC) analyses and 1-h forecasts, as well as those calculated from the operational surface objective analysis system used at the Storm Prediction Center (the SFCOA). The rapid growth in temperature, humidity, and wind errors from 0 to 1 h seen at all levels in a past RUC verification study by Benjamin et al. is not seen in the present study. This could be because the verification observations are also assimilated into the RUC in the Benjamin et al. study, whereas the verification observations in the present study are not. In the upper troposphere, the present study shows large errors in relative humidity, mostly related to a large moist bias. The planetary boundary layer tends to be too shallow in the RUC analyses and 1-h forecasts. Wind speeds tend to be too fast in the lowest 1 km and too slow in the 2–4-km layer. RUC and SFCOA 1-h forecast errors for many important severe weather parameters are large relative to their potential impact on convective evolution. However, the SFCOA significantly improves upon the biases seen in most of the 1-h RUC forecasts for the base-state surface variables and most of the other severe-weather-related parameters, indicating that the SFCOA has a more significant impact in reducing the biases in the 1-h RUC forecasts than on the root-mean-squared errors.

Corresponding author address: Dr. Michael C. Coniglio, National Severe Storms Laboratory, Rm. 2234, National Weather Center, 120 David L. Boren Blvd., Norman, OK 73072. E-mail: michael.coniglio@noaa.gov

Abstract

This study uses radiosonde observations obtained during the second phase of the Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX2) to verify base-state variables and severe-weather-related parameters calculated from Rapid Update Cycle (RUC) analyses and 1-h forecasts, as well as those calculated from the operational surface objective analysis system used at the Storm Prediction Center (the SFCOA). The rapid growth in temperature, humidity, and wind errors from 0 to 1 h seen at all levels in a past RUC verification study by Benjamin et al. is not seen in the present study. This could be because the verification observations are also assimilated into the RUC in the Benjamin et al. study, whereas the verification observations in the present study are not. In the upper troposphere, the present study shows large errors in relative humidity, mostly related to a large moist bias. The planetary boundary layer tends to be too shallow in the RUC analyses and 1-h forecasts. Wind speeds tend to be too fast in the lowest 1 km and too slow in the 2–4-km layer. RUC and SFCOA 1-h forecast errors for many important severe weather parameters are large relative to their potential impact on convective evolution. However, the SFCOA significantly improves upon the biases seen in most of the 1-h RUC forecasts for the base-state surface variables and most of the other severe-weather-related parameters, indicating that the SFCOA has a more significant impact in reducing the biases in the 1-h RUC forecasts than on the root-mean-squared errors.

Corresponding author address: Dr. Michael C. Coniglio, National Severe Storms Laboratory, Rm. 2234, National Weather Center, 120 David L. Boren Blvd., Norman, OK 73072. E-mail: michael.coniglio@noaa.gov

1. Introduction

A key part of assessing the potential for severe convective weather is the accurate depiction of the mesoscale environment and its vertical structure. A large number of studies have increased our knowledge of the relationships between convective storm behavior and parameters derived from vertical profiles of the preexisting atmospheric state through observational and numerical modeling studies (e.g., Newton 1963; Weisman and Klemp 1982; Rasmussen and Blanchard 1998), and this knowledge has long been transferred successfully to operational forecasting (Fawbush and Miller 1954; Johns and Doswell 1992; Thompson et al. 2004, 2007).

The average spacing of the routine National Weather Service (NWS) operational radiosonde network is coarse (~350 km) and the radiosonde launches are infrequent relative to convective and mesoscale time scales (Orlanski 1975). Therefore, convective-weather forecasters, who need to develop an understanding of the mesoscale environment for the accurate prediction of convective weather, usually rely heavily on objective analysis systems and numerical weather prediction (NWP) models to fill in the gaps where observations are not available. These objective analyses and model depictions are then used to diagnose environmental features and parameters that are relevant to severe-weather forecasting.

For operational centers like the Storm Prediction Center (SPC) that have forecasting responsibilities for the entire continental United States (CONUS) and need to develop an integrated view of the synoptic and mesoscale environment over multiple synoptic regimes and geographical locations, the hourly frequency of the network of surface observing stations, which has an average spacing of about 100 km, is indispensable. Analyses of surface observations help forecasters identify important mesoscale features that influence local or regional weather conditions (Johns and Doswell 1992). Yet, the accurate assimilation of in situ surface observations for short-range mesoscale NWP models is far from trivial (Benjamin et al. 2004c). The NWP-based parameters that are used in a forecast setting can depend significantly on the assimilation technique and the physical parameterization of surface fluxes into the planetary boundary layer (PBL), which can evolve on hourly or even subhourly time scales (Stull 1988, p. 2). Given the wide use of objective analysis systems and NWP models to diagnose the mesoscale environment in an operational forecast setting, it is important to measure the performance of these models, if their designs are to be improved and the model guidance is to be used effectively in the forecast process.

The second phase of the Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX2; Wurman et al. 2010) obtained radiosonde observations near supercell thunderstorms, some of which sampled the ambient conditions. This dataset provides the opportunity to assess the performance of mesoscale analysis and forecast systems in regions favorable for severe convective weather. This study focuses on operational forecast products based on the Rapid Update Cycle (RUC; Benjamin et al. 2004b) (both the hourly analyses and 1-h forecasts) and the RUC-based hourly surface objective analysis system (which has been called SFCOA, for surface objective analysis) currently in use at the SPC (Bothwell et al. 2002). Output from the RUC and the SFCOA are widely used in the operational forecasting of severe weather. In this study, performance measures are computed between observations and analyses–forecasts of the base-state vertical thermodynamic and kinematic structure of the atmosphere, as well as parameters derived from vertical profiles of the atmosphere. The VORTEX2 soundings provide an opportunity to evaluate the RUC-based analyses and forecasts using independent observations, since the VORTEX2 soundings were not used in the RUC data assimilation procedure or in the generation of the SFCOA.

The two goals of this study are to provide a validation of RUC analyses and 1-h forecasts, and to document the value of supplementing the RUC 1-h forecasts with analyses of surface data in an expedient manner as has been done at the SPC for the past decade. The results will be relevant to spring and early summer severe-weather regimes in the central United States. Is it important to provide up-to-date information on the performance characteristics of the RUC and SFCOA models because they are used so widely in operations, and because over the past decade, a wealth of studies have used the RUC analyses–forecasts or SFCOA analyses in climatological studies of storm environments (e.g., Markowski et al. 2003; Thompson et al. 2003; Schneider and Dean 2008; Schumacher and Johnson 2009; Coniglio et al. 2010).

Section 2 provides a description of the RUC and SFCOA and section 3 describes the sounding dataset and verification methodology. A comparison between the RUC analysis and the 1-h RUC forecasts valid at the same time is provided in section 4. Section 5 focuses on aggregate statistics of the RUC 0- and 1-h forecasts, and the SFCOA. A summary and concluding remarks are provided in section 6.

2. Mesoscale analysis and forecast systems being assessed

a. The RUC

The RUC is an operational weather prediction system that has been under development for many years within the National Oceanic and Atmospheric Administration/Earth Systems Research Laboratory (NOAA/ESRL, formerly the Forecast Systems Laboratory), consisting of an hourly analysis–assimilation procedure to initialize a numerical forecast model (Benjamin et al. 2004b). The RUC model has been run operationally at the NOAA/National Centers for Environmental Prediction (NCEP) on a 13-km grid with 50 isentropic-sigma hybrid vertical levels since June 2005. The primary purpose of the RUC is to provide guidance to aviation and severe weather forecasters in the form of frequently updated near-term weather forecasts. The hourly assimilation of observations from a variety of sources is a key contributor to the accuracy of these forecasts (Benjamin et al. 2010).

Most of the fields examined in section 4 are dependent on the assimilation of surface observations. The RUC uses the three-dimensional variational data assimilation (3DVAR) technique described in Benjamin et al. (2004a, hereafter B04), with subsequent improvements described in Benjamin et al. (2010). The method widely used in NCEP operations to assimilate surface (and other) observations into NWP models is the Gridpoint Statistical Interpolation (GSI) scheme (Wu et al. 2002). Although plans were made to apply the GSI method to surface-data assimilation into the next-generation version of the RUC—the Rapid Refresh (RR) model (Devenyi et al. 2007); both the version of the RUC evaluated in this paper, and the current operational implementation of the RR as of the time of this writing, use the 3DVAR-based method of surface data assimilation described below (C. Alexander 2011, personal communication).

As with any data assimilation scheme, covariances between the observed and model variables are needed for the adjustments to the model fields during the assimilation step. As part of the 3DVAR analysis, the RUC uses information from the PBL structure to decide how to adjust the model fields above the surface given a surface observation. As described in Benjamin et al. (2004c), the RUC diagnosis of the PBL depth (see http://ruc.noaa.gov/vartxt.html#PBL) is used to guide the depth over which the observation-minus-background values of ;temperature, moisture, and wind diagnosed at the surface are likely to be applicable. The innovations (corrections to background forecast values) to virtual potential temperature and water vapor from aviation routine weather reports (METARs) are then applied in 20-hPa increments with maximum weight at the surface and decreasing weight up to a depth of 75% of the 1-h forecast PBL depth, limited to a depth of ~200 hPa (C. Alexander 2011, personal communication). It was found that allowing surface observations to influence conditions up to 75% of the depth of the PBL with a weight that decreases with height, and not just over the lowest ~40 hPa with a constant weight as in previous versions of the RUC, resulted in an overall improvement in temperature and dewpoint forecast statistical accuracy, and improvements in CAPE forecasts (Benjamin et al. 2004c).

b. The SFCOA

The SPC (and formerly the National Severe Storms Forecast Center) has long produced analyses of surface observations, both manually and with objective analysis techniques, to diagnose mesoscale features in the environment. Since the RUC gridded fields were available routinely in SPC operations in 1998, methods used to analyze surface data at the SPC have been integrated with the upper-air analyses provided by the RUC to form the SFCOA. The SFCOA is widely used in SPC operations and NWS wide via the SPC forecast tools web page (http://www.spc.noaa.gov/exper/mesoanalysis/). Archived gridded fields from the SFCOA have also been used in numerous studies to assess climatology of storm environments (e.g., Thompson et al. 2007; Schneider and Dean 2008; Edwards et al. 2010; Thompson et al. 2010).

The SFCOA is a blend of the available surface observations with the RUC pressure-level fields above the surface valid at the same time. Although mesonet observations are used on rare occasions, the surface analysis of the SFCOA primarily uses METAR and marine observations. The surface observations are analyzed on a CONUS domain with 40-km grid spacing using a two-pass Barnes analysis (Barnes 1973). For the three-dimensional SFCOA product, these analyzed surface fields simply replace the RUC surface fields, with no alteration of the RUC fields above the surface. In operations, the hope is that the inclusion of up-to-date surface observations in hourly three-dimensional analyses improves upon the accuracy of the surface fields (and any derived fields that depend on surface observations) from those obtained in the raw RUC forecast fields.

The SFCOA is produced by SPC in real time every 15 min. The SFCOA simulations generated at 0, 15, and 30 min past the hour use the RUC 1-h forecast surface fields to quality control the surface observations prior to the Barnes analysis. For the SFCOA run generated at 45 min past the hour, the initial quality control step uses the RUC analysis valid at the same time as the observations (if it is available) instead of the RUC 1-h forecast, as well as the RUC analysis for fields above the surface. This “final” analysis performed at 45 min after the hour is archived at SPC and is the dataset that was available for this study.

The goal of this paper is to evaluate the products calculated from the SFCOA that are available routinely on the SPC forecast tools web page. Therefore, the SFCOA fields used in this study need to emulate, as best as possible, the analysis that is performed at 15 min past the hour. Since the Barnes analysis of the surface data does not use the RUC fields in any way during the analysis passes, the archived SFCOA gridded fields at the surface (2-m temperature, 2-m dewpoint temperature, surface pressure, 10-m winds) can be used simply to replace the RUC 1-h forecast surface fields valid at the same time to emulate the SFCOA product that is available on the web page. There are only two differences in the data used in this study and those that are used to generate the products on the web page. First, in this study the surface observations were quality controlled by the RUC analysis, whereas the surface fields on the web page use the RUC 1-h forecast fields valid at the same time for the initial quality control step. Second, the surface analysis used in this study (the final SFCOA analysis archived at SPC mentioned above) typically has 5%–10% more surface observations available than the analysis that is used for the web page graphics. These differences typically result in very minor differences in the gridded analysis fields that are used in this study and those that are displayed on the web page.

3. Sounding dataset and verification method

All of the soundings obtained during VORTEX2 were quality controlled by the National Center for Atmospheric Research/Earth Observing Laboratory (NCAR/EOL) [see Loehrer et al. (1998) for details on the quality control procedures] and are used as “truth” in the verification. During VORTEX2 the NOAA/National Severe Storms Laboratory (NSSL) and NCAR each operated two mobile GPS Advanced Upper-Air Sounding systems incorporating Vaisala RS92 radiosondes. Meteorological and position data (latitude and longitude) were recorded every 1 s along the balloon’s track. The final, quality controlled archive for the 2009 and 2010 seasons includes a total of 582 radiosonde soundings (raobs).

Many raobs that passed the quality control (QC) tests described below were obtained up to several hours prior to storm development and are used in this study. For those soundings that were taken with a nearby ongoing storm (or storms), the soundings that likely did not represent the inflow conditions to the storm and those deemed to be in the inflow region, but too close to convection, were not used. For the latter check, a sounding must have been no closer than 40 km from a region of composite reflectivity1 >15 dBZ at any point along the balloon’s ascent up to 200 hPa. The choice of 40 km was based on the work of Potvin et al. (2010), and the finding in the present study of a noticeable increase in vector wind and moisture errors starting above the 600-hPa level for the soundings that were located within 40 km of a storm.

A sounding was not used if it did not have valid wind or thermodynamic observations below 10 m above ground level (AGL) as determined by the EOL QC procedures. Furthermore, the wind observations below 6 m AGL were often weaker and directed inconsistently with the wind vector in the levels immediately above (likely because of the lasting effects of manually releasing the balloon) and were set to missing values in all soundings, even if they passed the EOL QC tests. The remaining wind observations were used to linearly interpolate the wind components to the 10 m AGL level for comparison to the diagnosed 10-m winds in the analyses and forecasts. For temperature and humidity, the observation closest to 2 m AGL (no higher than 10 m AGL) was used to define the surface conditions. Finally, those soundings for which RUC and SFCOA data were not available at times needed to compare to the observed soundings were removed from the dataset, along with a few soundings that appeared to have erroneous wind or humidity data in the mid- or upper levels. After all of the above QC tests were applied, only 93 of the 582 soundings remained. These 93 soundings, taken over 40 separate events, represent conditions prior to convective weather or in the inflow during the early stages of convection.

Since inferences are made later on the statistical significance of the error estimates, care must be taken to minimize the statistical dependence among the raobs. On days (e.g., 10 May 2010) during which several soundings were launched within several tens of kilometers and a few hours of each other, the soundings are likely not statistically independent. Although there are techniques that can account for dependent data when making statistical inferences (Gilleland 2010), a conservative approach is taken in this study by only using one sounding per event in the computations of the aggregate statistics. Although this leaves only 40 soundings that are used in the verification, their likely independence allows us to make general inferences on the overall performance of the RUC and SFCOA over the active days of VORTEX2. For days during which multiple soundings could be used, the sounding with the most data was selected. If more than one sounding had data up to the same pressure level, the sounding that was obtained closest to the time of the start of the convective event was selected. The 40 soundings used in the evaluations are limited to the central and southern plains region of the United States (Fig. 1), which were the general confines of the VORTEX2 operations. The times of the soundings range from 1738 UTC (1238 central daylight time, CDT) to 2336 UTC (1836 CDT) (Table 1), so the results are most representative of the RUC performance within daytime to early evening conditions.

Fig. 1.
Fig. 1.

Tracks of the 40 radiosondes used in this study, from the launch point (denoted by the cross) to the location at which the data terminated (denoted by the closed circle). The numbers correspond to the list of soundings in Table 1.

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

Table 1.

List of the 40 soundings used in this study, identified by date and time of the launch and the coordinates (lat–lon) of the launch points.

Table 1.

Because the balloons were usually launched in environments with strong flow, the balloons often drifted a considerable horizontal distance by the time they reached 200 hPa (Fig. 2). Furthermore, it was common for at least 40 min to elapse prior to the balloon reaching 200 hPa (Fig. 2). Therefore, to account for this balloon drift, the RUC gridded fields are interpolated in space and time to the time and location of the radiosonde along its path. This is accomplished by 1) determining the grid points that surround the balloon path as the balloon crosses the pressure levels in the RUC isobaric data, 2) linearly interpolating the gridded fields in time to these grid points using the RUC analyses or forecasts valid before and after the valid time of the observation, and 3) interpolating these gridded values to the balloon location using a Gaussian weighting of the surrounding grid points. To create the SFCOA soundings, the temperature, dewpoint, surface pressure, station elevation, and wind components from the 40-km SFCOA grids were simply interpolated to the time and location of the balloon at launch time, and these interpolated values replaced the surface values in the RUC 1-h forecast soundings. This last step emulates the procedure used to calculate the operational SFCOA gridded fields that are displayed on the SPC forecast tools web page.

Fig. 2.
Fig. 2.

Distribution of the time elapsed and horizontal distance traversed by the 40 radiosondes. Bins are every 10 min or 10 km.

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

The fields under consideration in this study are standard meteorological fields and derived fields (listed with acronyms defined in Table 2) that are used routinely in SPC operations and that are available on the SPC forecast tools web page at the time of this writing. The fields labeled “effective” in Table 2 are based on computations intended to reflect the storm inflow layer and follow the work of Thompson et al. (2007).2 Calculation of the PBL depth is very similar to the RUC algorithm, whereby the PBL depth is defined to be the level at which the virtual potential temperature first exceeds that averaged over the lowest 25 hPa3 plus 0.5 K (see http://ruc.noaa.gov/vartxt.html#PBL). All sounding parameters for all three data sources (RUC, SFCOA, observations) were computed using the same code, and the VORTEX2 soundings were linearly interpolated to the RUC 25-hPa isobaric levels prior to the computation of the parameters to ensure consistency in the comparisons between the model- and observed-sounding parameters. The sensitivity of the results to the interpolation method used for the VORTEX2 soundings was tested and found to be very small. The parcel calculations used the virtual temperature correction of Doswell and Rasmussen (1994).

Table 2.

List of sounding parameters calculated and illustrated in this study. The “effective layer” parameters follow Thompson et al. (2007). The parameters depending on lifted parcels use the virtual temperature correction (Doswell and Rasmussen 1994).

Table 2.

The performance of the RUC and SFCOA is measured using computations of the root-mean-squared difference (rmsd) and mean difference (bias) of the variables and parameters calculated from the observed and RUC–SFCOA soundings aggregated over the set of 40 soundings. Confidence intervals (95%) are illustrated on the rmsd and bias estimates and are computed on the differences between the rmsd and bias estimates between the two forecasts in question using the percentile bootstrap method described in Gilleland (2010). Unless noted otherwise, “significant differences” are defined to be those for which 95% confidence intervals on the resampled estimates of rmsd and bias differences do not overlap with zero, meaning there is only a 5% chance that the two estimates in question are sampled from the same underlying distribution.

4. Vertical error profiles in RUC analyses and 1-h forecasts

B04 compared error4 profiles of RUC analyses and forecasts to NWS raobs. We present some of these results herein (Fig. 3) for comparison, but the reader is reminded that many factors complicate the comparison of verification statistics among different sets of observations and forecasts. The B04 study covered primarily the autumn months over the full RUC domain for soundings only valid at 0000 and 1200 UTC that were also assimilated into the RUC analysis. The soundings used for verification in the present study were not assimilated into the RUC analysis and are valid at times other than 0000 and 1200 UTC (Table 1), but are taken over a more limited area (Fig. 1), in only May and June (Table 1), and in more limited synoptic regimes than in B04. Furthermore, numerous changes and improvements to the model design have occurred since the B04 study, so it is difficult to attribute differences in the statistics between those of B04 and in this study to any specific change. However, an interesting and robust difference between the results of B04 (Fig. 3) and those in this study is that the rapid growth in rmsd between the RUC analysis and 1-h forecasts seen for all variables in B04 is not seen in our study (Fig. 4). This discrepancy is likely found because the B04 verification observations are assimilated into the RUC, and therefore the RUC attempts to fit to the verifying observations (S. Benjamin 2011, personal communication) whereas in the present study, the verifying VORTEX2 soundings are independent of the assimilation procedure. This shows that the RUC analysis in regions of the central and southern plains that are not collocated with routine NWS raobs produces analysis errors comparable to those for the subsequent 1-h forecasts, and even worse than the 1-h forecast errors in the case of relative humidity (RH) in the upper troposphere (Fig. 4c).

Fig. 3.
Fig. 3.

The rmsds between NWS radiosonde observations (0000 and 1200 UTC) and RUC analyses and forecasts over the entire RUC computational domain for the period 11 Sep–31 Dec 2002, from the study by Benjamin et al. (2004a), for (a) vector wind (m s−1), (b) temperature (K), and (c) relative humidity (%).

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

Fig. 4.
Fig. 4.

Vertical profile of rmsd (m s−1) for the RUC analysis and 1-h forecast (a) vector wind, (b) temperature, and (c) RH. The differences between the rmsd estimates between the RUC analysis and 1-h RUC forecasts are shown by the black line. Error bars denote 95% confidence intervals on the differences in rmsd, with the thick error bars denoting statistically significant differences (i.e., the error bar does not overlap with the zero line) at the 95% level. Profiles of rmsd and their differences are only shown for levels above 875 hPa; fewer than 30 soundings had good data on higher pressure levels.

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

Biases in vector wind, temperature, and RH at levels above ground (AGL) are shown in Fig. 5. The relatively large rmsd values of RH in the middle and upper troposphere (Fig. 4) are determined mainly by a pronounced moist bias (Fig. 5c). For the 1-h forecasts, the magnitude of the RH bias is <8% up to about 5 km AGL, but the positive (moist) bias increases to 10%–15% in the 6–10-km layer.

Fig. 5.
Fig. 5.

As in Fig. 4, but for vertical profiles of mean difference (bias) on height levels above ground (m).

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

The present study uses Vaisala RS92 radiosondes, which provide much-improved humidity measurements in the troposphere compared to older radiosonde systems, and is a system that is recommended for research applications (Nash et al. 2005). However, this system may have a slight dry bias in cold temperatures and higher altitudes, so the magnitude of the moist bias aloft found in this study may be inflated somewhat by a bias in the observations. However, the moisture content in the mid- and upper troposphere is often quite variable (Nash et al. 2005), requiring the use of significant observation errors in NWP data assimilation systems, which, even with unbiased observing systems, leads to relatively large errors in model forecasts. As a result, mid- and upper-tropospheric moist biases in initial model analyses and short-term forecasts have been demonstrated for other modeling systems and regions of the earth (Colle et al. 2003; Eager et al. 2007; Rakesh et al. 2009) and are a well-known problem in NWP.

The PBL in the RUC 1-h forecasts is shallower than in the observations for 26 of the 40 soundings. Furthermore, the magnitudes of the errors in the PBL depth tend to be larger when the PBL is too shallow than when the PBL is too deep (see Fig. 6 for examples showing soundings in which the RUC 1-h forecast PBL was too shallow). This results in a low bias in the RUC-estimated thermodynamic PBL depth for both the RUC analyses and 1-h forecasts (Fig. 7). This problem of a PBL that is too shallow could be related to the use of a third-order closure version of the Burk and Thompson (1989) PBL scheme, which depends on local gradients explicitly computed on the model grid to parameterize turbulence in the PBL. Under similar synoptic regimes, regions, time of day, and times of the year, similarly based “local” PBL schemes have shown a tendency to produce PBLs that are too shallow and moisture profiles that are undermixed (Kain et al. 2005). A cool, moist bias near the surface that results from the undermixing of the PBL in local schemes (Kain et al. 2005) is also seen in this study (Figs. 5b and 5c). Note, however, that the perceived problem of undermixing does not always produce conditions near the ground that are too cool and too moist. In situations where moisture is higher aloft, the undermixing can result in conditions that are too dry near the ground (see Fig. 6d).

Fig. 6.
Fig. 6.

Examples of comparisons of observed (black) and 1-h forecast soundings (gray) in which the PBL depth was too shallow in the RUC forecast. The horizontal lines in corresponding black and gray shading indicate the PBL depth determined by the algorithm used in this study. Computed parameters are listed for each of the observed (black text) and forecasted soundings (gray text in parentheses).

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

Fig. 7.
Fig. 7.

Boxplots of PBL height errors (forecast − observation) for the RUC analysis (RUC00), 1-h RUC forecast (RUC01), and the SFCOA. The boxes encompass the interquartile range (25th–75th percentile), the dashed lines extend to the 10th and 90th percentiles, and the median is denoted by a horizontal line inside the box. The filled circle denotes the rmsd between the forecasts and observations and the hollow circle denotes the mean difference (bias) between the forecasts and observations. Error bars on the rmsd and bias estimates indicate 95% confidence intervals. Numbers along the bottom are the confidence (%) that the difference in the rmsd estimates (boldface text) or the difference in the bias estimate (lightface text) is statistically significant between the two data sources in question. For example, the statistical confidence that the RUC00 and RUC01 bias estimates are different is 85.9%, and the statistical confidence that the RUC01 and SFCOA rmsd estimates are different from each other is 93.9%.

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

Furthermore, the undermixing of the PBL does not seem to produce wind profile errors that would be expected. The RUC analyses and 1-h forecast wind speeds are too fast in the lowest 1 km AGL and too slow from 2 to 4 km AGL, with the RUC 1-h forecast wind speeds being significantly worse than the RUC analysis below 500 m AGL (Fig. 5a). Much of the high wind speed bias below 1 km is in the meridional component, whereas low biases in both the zonal and meridional winds contribute to the low wind speed bias aloft (not shown). If momentum were consistently undermixed as the daytime boundary layer develops as for the thermodynamic variables, then one would expect wind speeds that are too slow in the lowest 1 km AGL and too fast above this level for the times represented in the observations. This reveals that the low-level wind speed errors are complex and likely are driven less by the biases within the PBL scheme than the errors seen for the near-surface temperature and dewpoint.

5. Sounding parameter errors

Results are first presented for the near-surface temperature and dewpoint. All of the rmsd and bias estimates for the near-surface variables (except for the surface temperature bias) became larger for the RUC 1-h forecasts, significant at the 95% level for the temperature rmsd and for both the rmsd and bias in RH (Figs. 4 and 5). It is the goal of the SFCOA to ameliorate these errors in the 1-h RUC forecast of the surface fields, with the hope for significantly improved analyses of severe-weather-related parameters over those provided by the raw 1-h RUC forecasts.

The lowest observation below 10 m AGL in the observed sounding, which is used to verify the 2-m temperature and dewpoint fields from the RUC and SFCOA (see section 3), could be influenced by the prelaunch ground conditions and whether or not the radiosonde was sufficiently shaded and aspirated prior to launch (M. Parker 2011, personal communication). Optimal ground conditions and prelaunch preparation of each radiosonde could not be guaranteed. Therefore, results are presented for both the 2-m variables and the lowest-30-hPa average variables (Figs. 8 and 9). Any adverse effects from the ground conditions and radiosonde preparation that may affect the 2-m AGL comparisons should be reduced when examining the variables averaged over the lowest 30 hPa.

Fig. 8.
Fig. 8.

As in Fig. 7, but for (a) 2-m temperature and (b) 2-m dewpoint.

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

Fig. 9.
Fig. 9.

As in Fig. 7 but for (a) the lowest 30-hPa average temperature and (b) dewpoint.

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

Figures 8 and 9 show that the near-surface temperature and dewpoint comparisons for the 2-m and lowest-30-hPa average variables are similar. The SFCOA reduces the spread in the RUC 1-h forecast near-surface temperature errors and significantly reduces the rmsd (Figs. 8a and 9a). The temperature bias is very small (small positive for the 2-m level and small negative for the lowest-30-hPa average) and mean error magnitudes are mostly <1 K for both the 2-m and lowest-30-hPa average variables. These errors are comparable to the expected measurement accuracy in daytime boundary layers with steep lapse rates near the ground (Nash et al. 2005). The RUC 1-h forecasts produce a significantly higher near-surface moist bias with larger spread in the errors compared to the RUC analysis (Figs. 8b and 9b). The SFCOA ameliorates these errors in the 1-h RUC forecasts; the spread of the errors is reduced, and the rms and mean difference values are reduced significantly for both the 2-m and lowest-30-hPa average variables (Figs. 8b and 9b). Although a small near-surface moist bias remains in the SFCOA, the magnitudes of the errors are mostly <2 K.

The SFCOA CAPE variables show significant improvement over the 1-h RUC forecast biases (Fig. 10). The RUC analysis has a high bias for all CAPE variables examined, which becomes even higher for the RUC 1-h forecasts. For example, the CAPE is too high in about 75% of the 1-h RUC forecast soundings for all of the versions of CAPE examined. Over 50% (25%) of the 1-h RUC forecast soundings have a surface-based CAPE (SBCAPE) error > 400 (100) m2 s−2 (Figs. 10a and 10c). Although the CAPE variables from the SFCOA continue to show a high bias, the magnitude of this bias is reduced significantly in all CAPE variables. For example, the mean difference in SBCAPE (MLCAPE) decreases from about 400 m2 s−2 (150 m2 s−2) in the 1-h RUC forecasts to about 50 m2 s−2 (50 m2 s−2) in the SFCOA.

Fig. 10.
Fig. 10.

As in Fig. 7, but for (a) SBCAPE, (b) MUCAPE, (c) MLCAPE, and (d) LLCAPE (see Table 1 for variable definitions).

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

The rmsd for the full-column-integrated versions of CAPE between 400 and 600 m2 s−2 (Figs. 10a–c) are not necessarily large relative to the changes in CAPE that can have a significant impact on severe weather forecasting in the central United States, where CAPE values are routinely quite large. But changes in CAPE in the 400–600 m2 s−2 range can be important for forecasting in the southeast United States, where many significant severe weather events occur with these absolute values of CAPE (Schneider and Dean 2008). Since the soundings used herein were all obtained in the central United States, it is not known if these errors in CAPE are typical for other parts of the country or parameter spaces. If they are, then these uncertainties in CAPE on the order of the CAPE itself may be a primary reason for the poorer false alarm ratios for tornado watches in this CAPE parameter space (Dean and Schneider 2008).

The SFCOA significantly improves upon the surface-based convective inhibition (SBCIN) and mean layer CIN (MLCIN) biases in the 1-h RUC forecasts, although a high bias (indicating too little CIN) of about 15 m2 s−2 remains (Figs. 11a and 11c). The most unstable CIN (MUCIN) biases are significantly different between the SFCOA and 1-h RUC forecasts, but changes from a slight high bias to a slight low bias in MUCIN (Fig. 11b).

Fig. 11.
Fig. 11.

As in Fig. 7, but for (a) SBCIN, (b) MUCIN, and (c) MLCIN.

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

It is noteworthy that the rmsd and spread of the CIN errors are large for all three data sources and versions of CIN relative to the values of CIN that can have a significant impact on severe-weather forecasting. The rmsd is generally between 60 and 90 m2 s−2 for all CIN variables. For changes in environmental CIN in this range, storm severity (Rasmussen and Blanchard 1998; Davies 2004), storm evolution (Ziegler et al. 2010; Nowotarski et al. 2011), the general extent of precipitation over Northern Hemisphere continents in summer (Myoung and Nielson-Gammon 2010), and, in particular, the initiation of convection (Fabry 2006), are impacted greatly. These relatively large errors in CIN and the sensitivity of convective initiation and evolution to CIN speak to the difficulty in forecasting convection on time scales shorter than a few hours and spatial scales less than several tens of kilometers (Wilson and Roberts 2006). Furthermore, although the SFCOA procedure is sufficient to reduce the rather large spread in errors in the RUC 1-h SBCIN and MLCIN forecasts, improvements in physical parameterization of boundary layer processes and the adjustment of temperature and dewpoint in the boundary layer based on surface observations are likely needed to significantly lower the typical errors of CIN in mesoscale analyses and short-term forecasts.

Consistent with the high moisture bias seen for the RUC analysis and 1-h forecasts is a low lifted condensation level (LCL) bias that becomes worse for the 1-h RUC forecasts (Fig. 12). Accurate forecasts of LCL are important when assessing the potential for significant tornadoes (Thompson et al. 2003) given other parameters supportive of supercells. The SFCOA significantly reduces the bias for all LCL variables, particularly for the SBLCL, which has near-zero bias. The rmsd for the SBLCL decreases significantly for the SFCOA from near 400 m in the 1-h RUC forecasts to near 200 m for the SFCOA. The rmsd values for the SFOCA MULCL and MLLCL variables show less of an improvement compared to the SBLCL, but the rmsd still decreases significantly, to near 250 m, for the MLLCL.

Fig. 12.
Fig. 12.

As in Fig. 7, but for (a) SBLCL, (b) MULCL, and (c) MLLCL.

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

The SFCOA also alleviates some of the rather large low biases for the level of free convection (LFC) variables seen for the 1-h RUC forecasts, but the rmsd remains rather large (Fig. 13), particularly for the MLLFC (Fig. 13c). Changes to the LFC on the order of the rmsd values found in this study (500–1000 m) could impact many aspects of convective evolution, including updraft size, storm motion, and downdraft strength (McCaul and Cohen 2002; Kirkpatrick et al. 2007; Kirkpatrick et al. 2009).

Fig. 13.
Fig. 13.

As in Fig. 7, but for (a) SBLFC, (b) MULFC, and (c) MLLFC.

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

The supercell composite parameter (SCP) and significant tornado parameter (STP) were developed as a means to efficiently identify environments supporting supercells and significant tornadoes (Thompson et al. 2003). The intent was to normalize CAPE and measures of vertical wind shear to approximate threshold values for supercells and tornadoes and combine these parameters into nondimensional parameters. The calculation of SCP and STP herein follows Thompson et al. (2004, 2007) in the use of layers that are intended to reflect the potentially unstable parcels with relatively low values of convective inhibition. For the SCP, the SFCOA, once again, significantly reduces the spread of the errors and significantly reduces the mean difference and rmsd of the SCP for the 1-h RUC forecasts (Fig. 14a). The SCP mean difference is reduced by about two-thirds (~1.2 for the 1-h RUC forecasts to ~0.4 for the SFCOA). The SFCOA also improves on the 1-h RUC forecasts of STP, with slightly less, but still relatively large, confidence in the significance in the results (Fig. 14b). The STP mean difference also is reduced by about two-thirds (~0.4 for the 1-h RUC forecasts to ~0.15 for the SFCOA) and the rmsd is reduced by about a third (~0.8 for the 1-h RUC forecasts and ~0.6 for the SFCOA). The improvement in the SFCOA depiction of SCP, and particularly STP, is related to more correct analyses of zero SCP and STP in the SFCOA compared to the 1-h RUC forecasts. This indicates that the 1-h RUC forecast tends to overestimate the conditional potential for supercells and significant tornadoes much more so than the SFCOA.

Fig. 14.
Fig. 14.

As in Fig. 7, but for (a) SCP and (b) STP.

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

Like for the typical rmsd values for the CIN and LFC analyses and forecasts shown above, rmsd values of STP on the order of 0.8 are significant in practice, in that changes in the supercell environment on the order of the rmsd values found in this study can greatly impact the convective evolution and, in the case of the STP, the perceived likelihood of significant tornadoes. For instance, an increase in STP from 0.7 to 1.5 greatly changes the empirical likelihood of a significant tornado (see Fig. 3 of Thompson et al. 2004). The significantly improved depictions of the SCP and STP biases are likely driven by the depiction of the thermodynamic environment since the changes in effective bulk wind difference (BWD) and effective storm-relative helicity (SRH) from the 1-h RUC forecasts to the SFCOA (Fig. 15) are not as significant as the changes in the thermodynamic variables shown earlier.

Fig. 15.
Fig. 15.

As in Fig. 7, but for (a) effective BWD (EBWD) and (b) effective (ESRH).

Citation: Weather and Forecasting 27, 3; 10.1175/WAF-D-11-00096.1

6. Summary and concluding remarks

This study evaluates the performance of RUC analyses and 1-h forecasts, as well as the operational mesoscale analysis produced by the SPC (the SFCOA), using VORTEX2 radiosonde observations (raobs). Profiles from the RUC gridded fields are extracted along the path of the radiosonde, and verification statistics are computed for parameters derived from these profiles. The VORTEX2 soundings that were obtained in the preconvective and undisturbed near-storm environments represent conditions in regions favorable for severe convective weather and the associated model soundings represent the conditions that are used by forecasters to diagnose the environment. The goals of this study are 1) to quantify the typical errors in base-state variables in the RUC analyses and 1-h forecasts and their change with time during VORTEX2 operations and 2) to quantify the extent to which the blending of current surface observations with the 1-h RUC forecast, as used and presented operationally by the SPC, changes the typical errors and biases in severe-weather-related parameters in the raw 1-h RUC forecasts. One sounding per day over 40 VORTEX2 operations days was used in the verification, allowing general inferences to be made on statistical significance.

Results show that the rapid growth in temperature, humidity, and wind errors seen at all levels in the RUC verification study of B04 is not seen in the present study. Errors at the initial time are generally larger in the current study compared to those in B04, but the errors do not grow significantly for the 1-h forecast, like that seen in B04. This may be because the verifying raobs used in B04 were also assimilated into the RUC analysis, whereas the verifying raobs in the present study are independent of the analysis.

The PBL depths (gauged by the virtual potential temperature profile) tend to be too shallow in the RUC analyses, but improve marginally for the 1-h RUC forecasts. The analyses show a tendency for the levels just above the surface to about 1 km AGL to be too cool and too moist, but the wind speeds are too fast in the lowest 1 km AGL and too slow in the 2–4-km AGL layer.

The SFCOA significantly improves the rmsd in near-surface temperature, as well as both the rmsd and mean difference in near-surface dewpoint seen in the 1-h RUC forecasts. The rmsd for many derived variables is reduced significantly in the SFCOA, but for many variables, the rmsd values remain large relative to changes in variables that can have a large impact on the evolution of the convection. This is particularly true for the CIN and LFC variables. The largest positive impact the SFCOA has on the raw 1-h RUC forecasts is on reducing the bias in most of the thermodynamic fields examined, to a high degree of statistical confidence. The SFCOA was less impactful for the diagnosed PBL height and for the effective wind shear variables.

A motivation of the work presented herein is to form a baseline verification of the RUC and the mesoscale analysis used in SPC operations for later comparison to other mesoscale analysis systems that use very different techniques to assimilate surface observations. One of those techniques being explored for use in operations within NOAA as part of the “warn on forecast” effort (Stensrud et al. 2009b) is based on an application of the ensemble Kalman filter (EnKF) (Evensen 1994). An appealing aspect of the EnKF is that it uses an ensemble of nonlinear forecasts to estimate flow-dependent covariances for the assimilation. In theory, surface data assimilation using an EnKF approach should produce more realistic estimates of the local atmospheric state than assimilation methods, like 3DVAR, that use fixed covariances. Likewise, EnKF techniques, in theory, should mitigate analysis errors in data-void regions that can plague Gaussian-based techniques (Barnes 1994), and allow for smaller grid spacing relative to the data spacing than Gaussian-based techniques. Operational objective analyses at resolutions smaller than that currently offered by the SFCOA will be needed if the goals of the warn-on-forecast approach are to be met.

Indeed, recent studies have shown that standard surface observations assimilated via the EnKF can produce accurate depictions of the conditions within the PBL (Hacker and Snyder 2005); can improve the location and intensity of drylines and frontal boundaries, as well as PBL height and structure out to a forecast length of 6 h (Fujita et al. 2007), and can produce more realistic mesoscale temperature patterns and circulations associated with mesoscale convective systems (MCSs) (Stensrud et al. 2009a; Wheatley and Stensrud 2010). The computational cost of such methods is currently too great to implement EnKF systems into operations, but prototype systems that employ the EnKF technique are being developed within NOAA and elsewhere. The EnKF systems will be designed to produce multiscale analyses on mesoscale grid lengths that are used to drive explicit predictions of convection. Studies are currently under way to evaluate the ability of these integrated analyses and forecast systems to improve the analysis and short-term forecasts of severe-weather-related environmental parameters, so that the performance standards of the SPC SFCOA shown in this study are met before proceeding to design the portion of the system that explicitly predicts convective evolution.

Acknowledgments

The author thanks Phillip Bothwell and John Hart of the SPC, who provided the SFCOA gridded data and helpful comments. We also thank the science support ataff of the SPC, particularly Andy Dean and Jay Liang. I also want to acknowledge the VORTEX2 sounding teams whose hard work at home and in the field allowed for this study to be made (the teams were led by Dr. George Bryan and Dr. Matt Parker). Finally, the author greatly appreciates the financial support for VORTEX2 provided by NOAA and the National Science Foundation.

REFERENCES

  • Barnes, S. L., 1973: Mesoscale objective map analysis using weighted time series observations. NOAA Tech. Memo. ERL NSSL-62, 60 pp. [NTIS COM-73-10781.]

  • Barnes, S. L., 1994: Applications of the Barnes objective analysis scheme. Part I: Effects of undersampling, wave position, and station randomness. J. Atmos. Oceanic Technol., 11, 14331448.

    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., and Coauthors, 2004a: An hourly assimilation–forecast cycle: The RUC. Mon. Wea. Rev., 132, 495518.

  • Benjamin, S. G., , Grell G. A. , , Brown J. M. , , Smirnova T. G. , , and Bleck R. , 2004b: Mesoscale weather prediction with the RUC hybrid isentropic–terrain-following coordinate model. Mon. Wea. Rev., 132, 473494.

    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., , Weygandt S. , , Devenyi D. , , Brown J. M. , , Manikin G. , , Smith T. L. , , and Smirnova T. , 2004c: Improved moisture and PBL initialization in the RUC using METAR data. Preprints, 11th Conf. on Aviation, Range, and Aerospace Meteorology, Hyannis, MA, Amer. Meteor. Soc., 17.3. [Available online at http://ams.confex.com/ams/pdfpapers/82023.pdf.]

  • Benjamin, S. G., , Jamison B. D. , , Moninger W. R. , , Sahm S. R. , , Schwartz B. E. , , and Schlatter T. W. , 2010: Relative short-range forecast impact from aircraft, profiler, radiosonde, VAD, GPS-PW, METAR, and Mesonet observations via the RUC hourly assimilation cycle. Mon. Wea. Rev., 138, 13191343.

    • Search Google Scholar
    • Export Citation
  • Bothwell, P. D., , Hart J. A. , , and Thompson R. L. , 2002: An integrated three-dimensional objective analysis scheme in use at the Storm Prediction Center. Preprints, 21st Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., JP3.1. [Available online at http://ams.confex.com/ams/SLS_WAF_NWP/techprogram/paper_47482.htm.]

  • Burk, S. D., , and Thompson W. T. , 1989: A vertically nested regional numerical weather prediction model with second-order closure physics. Mon. Wea. Rev., 117, 23052324.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , Olson J. B. , , and Tongue J. S. , 2003: Multiseason verification of the MM5. Part I: Comparison with the Eta Model over the central and eastern United States and impact of MM5 resolution. Wea. Forecasting, 18, 431457.

    • Search Google Scholar
    • Export Citation
  • Coniglio, M. C., , Hwang J. Y. , , and Stensrud D. J. , 2010: Environmental factors in the upscale growth and longevity of MCSs derived from Rapid Update Cycle analyses. Mon. Wea. Rev., 138, 35143539; Corrigendum, 139, 26862688.

    • Search Google Scholar
    • Export Citation
  • Davies, J. M., 2004: Estimations of CIN and LFC associated with tornadic and nontornadic supercells. Wea. Forecasting, 19, 714726.

  • Dean, A. R., , and Schneider R. S. , 2008: Forecast challenges at the NWS Storm Prediction Center relating to the frequency of favorable severe storm environments. Preprints, 24th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 9A.2. [Available online at http://ams.confex.com/ams/24SLS/techprogram/paper_141743.htm.]

  • Devenyi, D., , Weygandt S. , , Schlatter T. , , Benjamin S. , , and Hu M. , 2007: Hourly data assimilation with the gridpoint statistical interpolation for the rapid refresh. Preprints, 22nd Conf. on Weather Analysis and Forecasting/18th Conf. on Numerical Weather Prediction, Park City, UT, Amer. Meteor. Soc., 4A.2. [Available online at http://ams.confex.com/ams/22WAF18NWP/techprogram/paper_124535.htm.]

  • Doswell, C. A., III, , and Rasmussen E. N. , 1994: The effect of neglecting the virtual temperature correction on CAPE calculations. Wea. Forecasting, 9, 625629.

    • Search Google Scholar
    • Export Citation
  • Eager, R. E., , Raman S. , , Holt T. R. , , Westphal D. , , Reid J. , , Nachamkin J. , , Liu M. , , and Mandoos A. A. , 2007: A seasonal statistical evaluation of COAMPS over the Arabian Gulf region. Pure Appl. Geophys., 164, 17471764.

    • Search Google Scholar
    • Export Citation
  • Edwards, R., , Dean A. R. , , Thompson R. L. , , and Smith B. T. , 2010: Objective environmental analyses and convective modes for U.S. tropical cyclone tornadoes from 2003–2008. Preprints, 25th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., P3.2. [Available online at http://ams.confex.com/ams/pdfpapers/175397.pdf.]

  • Evensen, G., 1994: Sequential data assimilation with a nonlinear quasi-geostrophic model using Monte-Carlo methods to forecast error statistics. J. Geophys. Res., 99 (C5), 10 14310 162.

    • Search Google Scholar
    • Export Citation
  • Fabry, F., 2006: The spatial variability of moisture in the boundary layer and its effect on convection initiation: Project-long characterization. Mon. Wea. Rev., 134, 7991.

    • Search Google Scholar
    • Export Citation
  • Fawbush, E. J., , and Miller R. C. , 1954: The types of air masses in which North American tornadoes form. Bull. Amer. Meteor. Soc., 35, 154165.

    • Search Google Scholar
    • Export Citation
  • Fujita, T., , Stensrud D. J. , , and Dowell D. C. , 2007: Surface data assimilation using an ensemble Kalman filter approach with initial condition and model physics uncertainty. Mon. Wea. Rev., 135, 18461868.

    • Search Google Scholar
    • Export Citation
  • Gilleland, E., 2010: Confidence intervals for forecast verification. NCAR Tech. Note NCAR/TN-479+STR, 71 pp.

  • Hacker, J. P., , and Snyder C. , 2005: Ensemble Kalman filter assimilation of fixed screen-height observations in a parameterized PBL. Mon. Wea. Rev., 133, 32603275.

    • Search Google Scholar
    • Export Citation
  • Johns, R. H., , and Doswell C. A. III, 1992: Severe local storms forecasting. Wea. Forecasting, 7, 588612.

  • Kain, J. S., , Weiss S. J. , , Baldwin M. E. , , Carbin G. W. , , Bright D. R. , , Levit J. J. , , and Hart J. A. , 2005: Evaluating high-resolution configurations of the WRF model that are used to forecast severe convective weather: The 2005 SPC/NSSL Spring Program. Preprints, 21st Conf. on Weather Analysis and Forecasting/17th Conf. on Numerical Weather Prediction, Washington, DC, Amer. Meteor., Soc., 2A.5. [Available online at http://ams.confex.com/ams/pdfpapers/94843.pdf.]

  • Kirkpatrick, J. C., , McCaul E. W. , , and Cohen C. , 2007: The motion of simulated convective storms as a function of basic environmental parameters. Mon. Wea. Rev., 135, 30333051.

    • Search Google Scholar
    • Export Citation
  • Kirkpatrick, J. C., , McCaul E. W. , , and Cohen C. , 2009: Variability of updraft and downdraft characteristics in a large parameter space study of convective storms. Mon. Wea. Rev., 137, 15501561.

    • Search Google Scholar
    • Export Citation
  • Loehrer, S. M., , Williams S. F. , , and Moore J. A. , 1998: Results from UCAR/JOSS quality control of atmospheric soundings from field projects. Preprints, 10th Symp. on Meteorological Observations and Instrumentation, Phoenix, AZ, Amer. Meteor. Soc., 1–6.

  • Markowski, P., , Hannon C. , , Frame J. , , Lancaster E. , , Pietrycha A. , , Edwards R. , , and Thompson R. L. , 2003: Characteristics of vertical wind profiles near supercells obtained from the Rapid Update Cycle. Wea. Forecasting, 18, 12621272.

    • Search Google Scholar
    • Export Citation
  • McCaul, E. W., Jr., , and Cohen C. , 2002: The impact on simulated storm structure and intensity of variations in the mixed layer and moist layer depths. Mon. Wea. Rev., 130, 17221748.

    • Search Google Scholar
    • Export Citation
  • Myoung, B., , and Nielsen-Gammon J. W. , 2010: Sensitivity of monthly convective precipitation to environmental conditions. J. Climate, 23, 166188.

    • Search Google Scholar
    • Export Citation
  • Nash, J., , Smout R. , , Oakley T. , , Pathack P. , , and Kurnosenko S. , 2005: WMO intercomparison of high quality radiosonde systems: Vacoas, Mauritius, 2–25 February 2005. WMO Commission on Instruments and Methods of Observation Final Rep., 118 pp. [Available online at http://www.wmo.ch/pages/prog/www/IMOP/reports/2003-2007/RSO-IC-2005_Final_Report.pdf.]

  • Newton, C. W., 1963: Dynamics of severe convective storms. Severe Local Storms, Meteor. Monogr., No. 27, Amer. Meteor. Soc., 33–58.

  • Nowotarski, C. J., , Markowski P. M. , , and Richardson Y. P. , 2011: The characteristics of numerically simulated supercell storms situated over statically stable boundary layers. Mon. Wea. Rev., 139, 31393162.

    • Search Google Scholar
    • Export Citation
  • Orlanski, I., 1975: A rational subdivision of scales for atmospheric processes. Bull. Amer. Meteor. Soc., 56, 527530.

  • Potvin, C. K., , Elmore K. L. , , and Weiss S. J. , 2010: Assessing the impacts of proximity sounding criteria on the climatology of significant tornado environments. Wea. Forecasting, 25, 921930.

    • Search Google Scholar
    • Export Citation
  • Rakesh, V., , Singh R. , , and Joshi P. C. , 2009: Intercomparison of the performance of MM5/WRF with and without satellite data assimilation in short-range forecast applications over the Indian region. Meteor. Atmos. Phys., 105, 133155.

    • Search Google Scholar
    • Export Citation
  • Rasmussen, E. N., , and Blanchard D. O. , 1998: A baseline climatology of sounding-derived supercell and tornado forecast parameters. Wea. Forecasting, 13, 11481164.

    • Search Google Scholar
    • Export Citation
  • Schneider, R. S., , and Dean A. R. , 2008: A comprehensive 5-year severe storm environment climatology for the continental United States. Preprints, 24th Conf. on Severe Local Storms, Savannah, GA, Amer. Meteor. Soc., 16A.4. [Available online at http://ams.confex.com/ams/24SLS/techprogram/paper_141748.htm.]

  • Schumacher, R. S., , and Johnson R. H. , 2009: Quasi-stationary, extreme-rain-producing convective systems associated with midlevel cyclonic circulations. Wea. Forecasting, 24, 555574.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., , Yussouf N. , , Dowell D. C. , , and Coniglio M. C. , 2009a: Assimilating surface data into a mesoscale model ensemble: Cold pool analyses from spring 2007. Atmos. Res., 93, 207220.

    • Search Google Scholar
    • Export Citation
  • Stensrud, D. J., and Coauthors, 2009b: Convective-scale warn-on-forecast system. Bull. Amer. Meteor. Soc., 90, 14871499.

  • Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic, 666 pp.

  • Thompson, R. L., , Hart J. A. , , Elmore K. L. , , and Markowski P. , 2003: Close proximity soundings within supercell environments obtained from the Rapid Update Cycle. Wea. Forecasting, 18, 12431261.

    • Search Google Scholar
    • Export Citation
  • Thompson, R. L., , Edwards R. , , and Mead C. M. , 2004: An update to the supercell composite and significant tornado parameters. Preprints, 22nd Conf. on Severe Local Storms, Hyannis, MA, Amer. Meteor. Soc., P8.1. [Available online at http://ams.confex.com/ams/pdfpapers/82100.pdf.]

  • Thompson, R. L., , Mead C. M. , , and Edwards R. , 2007: Effective storm-relative helicity and bulk shear in supercell thunderstorm environments. Wea. Forecasting, 22, 102115.

    • Search Google Scholar
    • Export Citation
  • Thompson, R. L., , Smith B. T. , , Grams J. S. , , Dean A. R. , , and Broyles C. , 2010: Climatology of near-storm environments with convective modes for significant thunderstorms in the contiguous United States. Preprints, 25th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 16B.6. [Available online at http://ams.confex.com/ams/25SLS/techprogram/paper_175727.htm.]

  • Vasiloff, S. V., and Coauthors, 2007: Improving QPE and very short term QPF: An initiative for a community-wide integrated approach. Bull. Amer. Meteor. Soc., 88, 18991911.

    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., , and Klemp J. B. , 1982: The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev., 110, 504520.

    • Search Google Scholar
    • Export Citation
  • Wheatley, D. M., , and Stensrud D. J. , 2010: The impact of assimilating surface pressure observations on severe weather events in a WRF mesoscale ensemble system. Mon. Wea. Rev., 138, 16731694.

    • Search Google Scholar
    • Export Citation
  • Wilson, J. W., , and Roberts R. D. , 2006: Summary of convective storm initiation and evolution during IHOP: Observational and modeling perspective. Mon. Wea. Rev., 134, 2347.

    • Search Google Scholar
    • Export Citation
  • Wu, W.-S., , Purser R. J. , , and Parrish D. F. , 2002: Three-dimensional variational analysis with spatially inhomogeneous covariances. Mon. Wea. Rev., 130, 29052916.

    • Search Google Scholar
    • Export Citation
  • Wurman, J., , Wicker L. J. , , Richardson Y. P. , , Rasmussen E. N. , , Markowski P. M. , , Dowell D. , , Burgess D. W. , , and Bluestein H. B. , 2010: An overview of the VORTEX2 field campaign. Preprints, 25th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 5.1. [Available online at http://ams.confex.com/ams/25SLS/techprogram/paper_176068.htm.]

  • Ziegler, C. L., , Mansell E. R. , , Straka J. M. , , MacGorman D. R. , , and Burgess D. W. , 2010: The impact of spatial variations of low-level stability on the life cycle of a simulated supercell storm. Mon. Wea. Rev., 138, 17381766.

    • Search Google Scholar
    • Export Citation
1

Composite (column maximum) reflectivity derived from the National Mosaic and Multisensor Quantitative Precipitation Estimation Project (NMQ; Vasiloff et al. 2007) was used.

2

Instead of using the wind components interpolated to 50% of the equilibrium level for the upper bound in the computation of effective shear, or BWD, a level of 6.5 km was used because sometimes there was no CAPE in the sounding or, in one case, the data that passed the QC procedures terminated before an equilibrium level could be determined.

3

A lowest-25-hPa average was used, instead of the 2-m values alone as in the RUC algorithm, as it was found that superadiabatic layers at the ground sometimes overestimated the PBL depth based on a visual inspection of the sounding on a skew T–logp diagram.

4

Hereafter, the term “error” refers to forecast minus observations and is used interchangeably with the term “difference.”

Save