• Alfaro, D. A., and M. Khairoutdinov, 2015: Thermodynamic constraints on the morphology of simulated midlatitude squall lines. J. Atmos. Sci., 72, 31163137, doi:10.1175/JAS-D-14-0295.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ashley, W. S., T. L. Mote, G. Dixon, S. L. Trotter, E. J. Powell, J. D. Durkee, and A. J. Grundstein, 2003: Distribution of mesoscale convective complex rainfall in the United States. Mon. Wea. Rev., 131, 30033017, doi:10.1175/1520-0493(2003)131<3003:DOMCCR>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., and et al. , 2016: A North American hourly assimilation and model forecast cycle: The Rapid Refresh. Mon. Wea. Rev., 144, 16691694, doi:10.1175/MWR-D-15-0242.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Berg, P., C. Moseley, and J. O. Haerter, 2013: Strong increase in convective precipitation in response to higher temperatures. Nat. Geosci., 6, 181185, doi:10.1038/ngeo1731.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bierkens, M. F. P., and et al. , 2015: Hyper-resolution global hydrological modelling: What is next? Hydrol. Processes, 29, 310320, doi:10.1002/hyp.10391.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bretherton, C. S., M. E. Peters, and L. E. Back, 2004: Relationships between water vapor path and precipitation over the tropical oceans. J. Climate, 17, 15171528, doi:10.1175/1520-0442(2004)017<1517:RBWVPA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Coniglio, M. C., 2012: Verification of RUC 0–1-h forecasts and SPC mesoscale analyses using VORTEX2 soundings. Wea. Forecasting, 27, 667683, doi:10.1175/WAF-D-11-00096.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Coniglio, M. C., J. Y. Hwang, and D. J. Stensrud, 2010: Environmental factors in the upscale growth and longevity of MCSs derived from Rapid Update Cycle analyses. Mon. Wea. Rev., 138, 35143539, doi:10.1175/2010MWR3233.1; Corrigendum, 139, 26862688, doi:10.1175/MWR-D-11-00064.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Coniglio, M. C., J. Correia Jr., P. T. Marsh, and F. Kong, 2013: Verification of convection-allowing WRF model forecasts of the planetary boundary layer using sounding observations. Wea. Forecasting, 28, 842862, doi:10.1175/WAF-D-12-00103.1.

    • Crossref
    • 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.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fritsch, J. M., and R. E. Carbone, 2004: Improving quantitative precipitation forecasts in the warm season: A USWRP research and development strategy. Bull. Amer. Meteor. Soc., 85, 955965, doi:10.1175/BAMS-85-7-955.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fritsch, J. M., R. J. Kane, and C. R. Chelius, 1986: The contribution of mesoscale convective weather systems to the warm-season precipitation in the United States. J. Appl. Meteor., 25, 13331345, doi:10.1175/1520-0450(1986)025<1333:TCOMCW>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Geerts, B., and et al. , 2017: The 2015 Plains Elevated Convection At Night field project. Bull. Amer. Meteor. Soc., 98, 767786, doi:10.1175/BAMS-D-15-00257.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Groisman, P. Y., R. W. Knight, and T. R. Karl, 2012: Changes in intense precipitation over the central United States. J. Hydrometeor., 13, 4766, doi:10.1175/JHM-D-11-039.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 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.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Houze, R. A., Jr., 2004: Mesoscale convective systems. Rev. Geophys., 42, RG4003, doi:10.1029/2004RG000150.

  • James, R. P., and P. M. Markowski, 2010: A numerical investigation of the effects of dry air aloft on deep convection. Mon. Wea. Rev., 138, 140161, doi:10.1175/2009MWR3018.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • James, R. P., P. M. Markowski, and J. M. Fritsch, 2006: Bow echo sensitivity to ambient moisture and cold pool strength. Mon. Wea. Rev., 134, 950964, doi:10.1175/MWR3109.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Keene, K. M., and R. S. Schumacher, 2013: The bow and arrow mesoscale convective structure. Mon. Wea. Rev., 141, 16481672, doi:10.1175/MWR-D-12-00172.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lackmann, G. M., 2013: The south-central U.S. flood of May 2010: Present and future. J. Climate, 26, 46884709, doi:10.1175/JCLI-D-12-00392.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mapes, B. E., 1993: Gregarious tropical convection. J. Atmos. Sci., 50, 20262037, doi:10.1175/1520-0469(1993)050<2026:GTC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Market, P., S. Allen, R. Scofield, R. Kuligowski, and A. Gruber, 2003: Precipitation efficiency of warm-season midwestern mesoscale convective systems. Wea. Forecasting, 18, 12731285, doi:10.1175/1520-0434(2003)018<1273:PEOWMM>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul, E. W., Jr., C. Cohen, and C. Kirkpatrick, 2005: The sensitivity of simulated storm structure, intensity, and precipitation efficiency to environmental temperature. Mon. Wea. Rev., 133, 30153037, doi:10.1175/MWR3015.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Melhauser, C., and F. Zhang, 2012: Practical and intrinsic predictability of severe and convective weather at the mesoscales. J. Atmos. Sci., 69, 33503371, doi:10.1175/JAS-D-11-0315.1.

    • Crossref
    • 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.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Naylor, J., and M. S. Gilmore, 2012: Convective initiation in an idealized cloud model using an updraft nudging technique. Mon. Wea. Rev., 140, 36993705, doi:10.1175/MWR-D-12-00163.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Parker, M. D., and R. H. Johnson, 2004: Simulated convective lines with leading precipitation. Part II: Evolution and maintenance. J. Atmos. Sci., 61, 16561673, doi:10.1175/1520-0469(2004)061<1656:SCLWLP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peters, J. M., and P. J. Roebber, 2014: Synoptic control of heavy-rain-producing convective training episodes. Mon. Wea. Rev., 142, 24642482, doi:10.1175/MWR-D-13-00263.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peters, J. M., and R. S. Schumacher, 2014: Objective categorization of heavy-rain-producing MCS synoptic types by rotated principal component analysis. Mon. Wea. Rev., 142, 17161737, doi:10.1175/MWR-D-13-00295.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peters, J. M., and R. S. Schumacher, 2015: The simulated structure and evolution of a quasi-idealized warm season convective system with a training convective line. J. Atmos. Sci., 72, 19872010, doi:10.1175/JAS-D-14-0215.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peters, J. M., and R. S. Schumacher, 2016: Dynamics governing a simulated mesoscale convective system with a training convective line. J. Atmos. Sci., 73, 26432664, doi:10.1175/JAS-D-15-0199.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peters, J. M., E. R. Nielsen, M. D. Parker, S. M. Hitchcock, and R. S. Schumacher, 2017: The impact of low-level moisture errors on model forecasts of an MCS observed during PECAN. Mon. Wea. Rev., doi:10.1175/MWR-D-16-0296.1, in press.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rotunno, R., J. B. Klemp, and M. L. Weisman, 1988: A theory for strong, long-lived squall lines. J. Atmos. Sci., 45, 463485, doi:10.1175/1520-0469(1988)045<0463:ATFSLL>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schumacher, R. S., 2015: Sensitivity of precipitation accumulation in elevated convective systems to small changes in low-level moisture. J. Atmos. Sci., 72, 25072524, doi:10.1175/JAS-D-14-0389.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schumacher, R. S., T. J. Galarneau Jr., and L. F. Bosart, 2011: Distant effects of a recurving tropical cyclone on rainfall in a midlatitude convective system: A high-impact predecessor rain event. Mon. Wea. Rev., 139, 650667, doi:10.1175/2010MWR3453.1.

    • Crossref
    • 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.

    • Crossref
    • Export Citation
  • Stevenson, S. N., and R. S. Schumacher, 2014: A 10-year survey of extreme rainfall events in the central and eastern United States using gridded multisensor precipitation analyses. Mon. Wea. Rev., 142, 31473162, doi:10.1175/MWR-D-13-00345.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sukovich, E. M., F. M. Ralph, F. E. Barthold, D. W. Reynolds, and D. R. Novak, 2014: Extreme quantitative precipitation forecast performance at the Weather Prediction Center from 2001 to 2011. Wea. Forecasting, 29, 894911, doi:10.1175/WAF-D-13-00061.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Takemi, T., 2006: Impacts of moisture profile on the evolution and organization of midlatitude squall lines under various shear conditions. Atmos. Res., 82, 3754, doi:10.1016/j.atmosres.2005.01.007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Takemi, T., 2010: Dependence of the precipitation intensity in mesoscale convective systems to temperature lapse rate. Atmos. Res., 96, 273285, doi:10.1016/j.atmosres.2009.09.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thompson, G., P. R. Field, R. M. Rasmussen, and W. D. 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.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wandishin, M. S., D. J. Stensrud, S. L. Mullen, and L. J. Wicker, 2008: On the predictability of mesoscale convective systems: Two-dimensional simulations. Wea. Forecasting, 23, 773785, doi:10.1175/2008WAF2007057.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wandishin, M. S., D. J. Stensrud, S. L. Mullen, and L. J. Wicker, 2010: On the predictability of mesoscale convective systems: Three-dimensional simulations. Mon. Wea. Rev., 138, 863885, doi:10.1175/2009MWR2961.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., and J. B. Klemp, 1982: The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev., 110, 504520, doi:10.1175/1520-0493(1982)110<0504:TDONSC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    Large-scale environments in the (a),(c) CTRL and (b),(d) 1KM_LARGE simulations. The 900-hPa geopotential height is contoured (every 15 m), the 900-hPa wind barbs are shown for magnitudes ≥6 m s−1 (half barb = 5, full barb = 10 kt; where 1 kt = 0.5144 m s−1), the IWV is in color shading (mm), and the 300-hPa wind speed is in gray shading (m s−1). (a),(b) Model initial conditions on domain 1 and (c),(d) conditions for t = 12 h. Geographic borders are shown only to provide a sense of horizontal scale; the model land surface is homogeneous. The location of the vertical profiles in Figs. 4 and 5 is shown by an asterisk in (a).

  • View in gallery

    As in Fig. 1, but for 900-hPa temperature (every 2 K), 900-hPa temperature advection (color shading; 10−5 K s−1), and 900-hPa wind barbs for CTRL at t = 12 h.

  • View in gallery

    Configuration of outer and inner model domains. Geographic borders are shown only to provide a sense of horizontal scale; the model land surface is homogeneous.

  • View in gallery

    Vertical profiles at the point indicated by the asterisk in Fig. 1a of water vapor mixing ratio (g kg−1, solid lines corresponding to the bottom axis) and RH (%, dashed lines corresponding to the top axis) at t = (a) 0 and (b) 12 h.

  • View in gallery

    Skew T–logp diagrams at t = 12 h at the point shown in Fig. 1. The dashed red line shows the virtual temperature. Dewpoint profiles are shown using the same color scheme as in Fig. 4. Parcel curves for the most-unstable parcel in CTRL and 1KM_LARGE are shown with the black and blue dashed lines, respectively. The thermodynamic calculations for CTRL are shown; calculations for the experiments are summarized in Table 1. A magnified view of the low levels is shown at right.

  • View in gallery

    Simulated radar reflectivity factor at 1 km AGL (color shading, dBZ), the 23°C surface potential temperature contour (blue), and 500-m AGL wind barbs (only magnitudes ≥9 m s−1 shown) at t = 15 h for (a) CTRL, (b) 600M_SMALL, (c) 600M_LARGE, (d) 1KM_SMALL, and (e) 1KM_LARGE. Red lines indicate the locations of vertical sections shown in Figs. 12 and 13.

  • View in gallery

    As in Fig. 6, but at t = 23 h.

  • View in gallery

    As in Fig. 6, but at t = 27 h.

  • View in gallery

    Total 24-h accumulated precipitation (mm) on the inner domain for (a) CTRL, (b) 600M_SMALL, (c) 600M_LARGE, (d) 1KM_SMALL, and (e) 1KM_LARGE. The red rectangle in (e) indicates the region used for area averaging in Table 2 and subsequent figures. The brown dashed rectangles in (a) and (e) indicate the averaging region for the Hovmöller diagrams in Fig. 19. In (d) and (e), the location of the maximum precipitation in CTRL is indicated with the gray plus sign (+). The area-averaged precipitation, also shown in Table 2, is given in the lower-left corner of each panel.

  • View in gallery

    Time series of area-averaged rainfall accumulation (mm) over the region shown in Fig. 9e.

  • View in gallery

    (a) Number of inner-domain grid points with hourly rainfall exceeding 50 mm h−1. (b) As in Fig. 10, but for area-averaged hourly rainfall (mm).

  • View in gallery

    Vertical sections of CAPE for parcels lifted from each vertical level (color shading in J kg−1), potential temperature (black contours every 2 K), and vertical velocity (blue contours at 0.5 m s−1 and every 1 m s−1 above that). All values have been averaged over 15 grid points (20 km) on either side of line A–B shown in Fig. 6a. Results are shown for the (left) CTRL and (right) 1KM_LARGE simulations. Times shown are t = (a),(b) 14; (c),(d) 17; (e),(f) 20; and (g),(h) 23 h.

  • View in gallery

    As in Fig. 12, but the color-shaded field is the potential temperature perturbation [defined as the difference between the full simulation and a corresponding simulation with the microphysics parameterization turned off (NOMP)].

  • View in gallery

    (a),(d) Locations of parcel trajectories originating at t = 19–24 h in (a) CTRL and (d) 1KM_LARGE. The 24°C surface temperature contour at t = 19 h is shown in blue to indicate the location of the outflow boundary in each simulation. The size of the arrowhead indicates the height of the parcel at t = 24 h, according to the key at the bottom right. Trajectories were initialized at every sixth model grid point within the red rectangle at 100, 200, and 300 m AGL for a total of 900 trajectories. In (a),(d), a random selection of 150 trajectories is shown for clarity. (b),(c),(e),(f) Vertical section along line A–B in Fig. 6a showing potential temperature perturbation (color shading, K) and the vertical location of air parcel trajectories through t = (b),(e) 22 and (c),(f) 24 h. All 900 trajectories are shown in (b),(c),(e),(f).

  • View in gallery

    As in Fig. 14, but for trajectories originating at 1.5, 1.6, and 1.7 km AGL within the red rectangles shown. (a),(c) Locations of parcel trajectories originating at t = 19–24 h in (a) CTRL and (c) 1KM_LARGE. (b),(d) Vertical section along line C–D in (a) showing the potential temperature perturbation (color shading, K) and the vertical location of air parcel trajectories through t = 22 h.

  • View in gallery

    Histogram of the maximum height attained by the air parcels between t = 19 and 24 h. For example, the leftmost red bar indicates that 25% of parcels in CTRL had a maximum height of 1 km AGL or below.

  • View in gallery

    Maximum column (shading), surface K (green contour), and maximum column w > 3 m s−1 (dark gray contours) for the (left) CTRL and (right) 1KM_LARGE simulations, valid at t = (a),(b) 20; (c),(d) 22; and (e),(f) 24 h.

  • View in gallery

    Maximum column (shading), cold pool intensity (c, red contours at intervals of 5 m s−1, defined as in PS16), and maximum column w > 3 m s−1 (dark gray contours) for the (left) CTRL and (right) 1KM_LARGE simulations, valid at tsim = (a),(b) 20; (c),(d) 22; and (e),(f) 24 h.

  • View in gallery

    (a) Hovmöller diagram of 1-h CTRL precipitation accumulation (shading, mm), (red contours), and (black contours), averaged north-to-south over the dashed box in Fig. 9a. (b) As in (a), but for the 1KM_LARGE simulation, and averaged over the dashed box in Fig. 9e. The blue arrows indicate periods when cold-pool-driven back-building was occurring.

  • View in gallery

    Vertical profiles of accumulated latent heating (K), averaged over the region shown in Fig. 9e, through t = 24 h in the CTRL (black) and 1KM_LARGE (blue) simulations.

  • View in gallery

    As in Figs. 12e,f, but for (a) 600M_SMALL, (b) 600M_LARGE, and (c) 1KM_SMALL at t = 20 h.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 67 67 7
PDF Downloads 48 48 9

Near-Surface Thermodynamic Sensitivities in Simulated Extreme-Rain-Producing Mesoscale Convective Systems

View More View Less
  • 1 Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado
© Get Permissions
Full access

Abstract

This study investigates the influences of low-level atmospheric water vapor on the precipitation produced by simulated warm-season midlatitude mesoscale convective systems (MCSs). In a series of semi-idealized numerical model experiments using initial conditions gleaned from composite environments from observed cases, small increases in moisture were applied to the model initial conditions over a layer either 600 m or 1 km deep. The precipitation produced by the MCS increased with larger moisture perturbations as expected, but the rainfall changes were disproportionate to the magnitude of the moisture perturbations. The experiment with the largest perturbation had a water vapor mixing ratio increase of approximately 2 g kg−1 over the lowest 1 km, corresponding to a 3.4% increase in vertically integrated water vapor, and the area-integrated MCS precipitation in this experiment increased by nearly 60% over the control. The locations of the heaviest rainfall also changed in response to differences in the strength and depth of the convectively generated cold pool. The MCSs in environments with larger initial moisture perturbations developed stronger cold pools, and the convection remained close to the outflow boundary, whereas the convective line was displaced farther behind the outflow boundary in the control and the simulations with smaller moisture perturbations. The high sensitivity of both the amount and location of MCS rainfall to small changes in low-level moisture demonstrates how small moisture errors in numerical weather prediction models may lead to large errors in their forecasts of MCS placement and behavior.

Current affiliation: Naval Postgraduate School, Monterey, California.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author e-mail: Prof. Russ Schumacher, russ.schumacher@colostate.edu

Abstract

This study investigates the influences of low-level atmospheric water vapor on the precipitation produced by simulated warm-season midlatitude mesoscale convective systems (MCSs). In a series of semi-idealized numerical model experiments using initial conditions gleaned from composite environments from observed cases, small increases in moisture were applied to the model initial conditions over a layer either 600 m or 1 km deep. The precipitation produced by the MCS increased with larger moisture perturbations as expected, but the rainfall changes were disproportionate to the magnitude of the moisture perturbations. The experiment with the largest perturbation had a water vapor mixing ratio increase of approximately 2 g kg−1 over the lowest 1 km, corresponding to a 3.4% increase in vertically integrated water vapor, and the area-integrated MCS precipitation in this experiment increased by nearly 60% over the control. The locations of the heaviest rainfall also changed in response to differences in the strength and depth of the convectively generated cold pool. The MCSs in environments with larger initial moisture perturbations developed stronger cold pools, and the convection remained close to the outflow boundary, whereas the convective line was displaced farther behind the outflow boundary in the control and the simulations with smaller moisture perturbations. The high sensitivity of both the amount and location of MCS rainfall to small changes in low-level moisture demonstrates how small moisture errors in numerical weather prediction models may lead to large errors in their forecasts of MCS placement and behavior.

Current affiliation: Naval Postgraduate School, Monterey, California.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author e-mail: Prof. Russ Schumacher, russ.schumacher@colostate.edu

1. Introduction

Mesoscale convective systems (MCSs; Houze 2004)—organized lines or clusters of convection—are the primary producers of heavy and extreme warm-season rainfall in many midlatitude locations, including the central United States (e.g., Fritsch et al. 1986; Ashley et al. 2003; Stevenson and Schumacher 2014). Yet the timing, location, and rainfall amounts in warm-season MCSs remain difficult to predict: forecast skill for both precipitation generally and for heavy precipitation is minimized in the summer (e.g., Fritsch and Carbone 2004; Sukovich et al. 2014). The reasons for poor rainfall predictions in the warm season are many, and they are related to the small spatial and temporal scales on which warm-season precipitation processes operate: warm-season MCSs often occur in environments in which mesoscale and storm-scale processes play important roles, and those processes in turn are very sensitive to the details of their environments. Or, in other words, warm-season MCSs, and their associated precipitation, have limited predictability (e.g., Wandishin et al. 2008, 2010; Melhauser and Zhang 2012).

As one example of this problem, Schumacher (2015, hereinafter S15) showed that the location and magnitude of the rainfall production in a simulated elevated MCS was highly sensitive to very small changes in low-level moisture. All of the simulations used the same mesoscale forcing for ascent, and the moisture changes were applied only within the near-surface stable layer and did not change the CAPE of the most unstable parcels. Yet they led to changes in both precipitation accumulation and convective structure that were disproportionate to the magnitude of the thermodynamic changes. This suggests that, in a forecast situation, errors in precipitation accumulations can be substantial even with an accurate representation of the mesoscale forcing for ascent. However, these simulations neglect potentially important aspects of real MCSs, such as horizontal gradients in moisture. It is well established that increases in atmospheric moisture are related to increases in precipitation, particularly in the tropics. For example, Bretherton et al. (2004) showed a strong, nonlinear relationship between integrated column water vapor (IWV) and precipitation over tropical oceans, and many other studies have confirmed this relationship. In midlatitude convective systems, however, the relationship between moisture and precipitation may be comparatively complicated, in part because of the tendency for MCSs to be organized by vertical wind shear and horizontal temperature gradients in ways that prolong the heavy rainfall and produce local rainfall accumulations many times larger than the local IWV. Many investigators have examined the influence of thermodynamic conditions and vertical wind shear on the characteristics of midlatitude convection, including rainfall production (e.g., Weisman and Klemp 1982; Market et al. 2003; McCaul et al. 2005; James et al. 2006; Takemi 2006, 2010; Schumacher et al. 2011; Alfaro and Khairoutdinov 2015), but there remain many questions regarding the factors responsible for limiting the predictive skill and predictability of warm-season precipitation. In particular, numerical model analyses and forecasts often have biases in low-level moisture [owing in part to a lack of mesoscale observations above the surface, e.g., Coniglio (2012); Coniglio et al. (2013)], which may in turn lead to substantial errors in the distribution of precipitation. Observations from the 2015 Plains Elevated Convection At Night (PECAN; Geerts et al. 2017) field campaign showed that the operational Rapid Refresh (Benjamin et al. 2016) analysis had low-level moisture errors of 2–4 g kg−1 near a heavily raining MCS (Peters et al. 2017). The influences of such low-level moisture errors on precipitation forecasts have not been sufficiently studied.

In this study, we build upon the findings of S15 by conducting a series of numerical experiments in which changes to the low-level thermodynamic conditions are applied to the “quasi idealized” MCS simulation described by Peters and Schumacher (2015, 2016). These experiments allow for an evaluation of the results found by S15 in an environment that is more representative of observed cases but still simplified to ease the interpretation and generalization of the results. Section 2 describes this numerical modeling framework and the design of the experiments. The results of the experiments are reported in section 3, section 4 discusses the context and implications of the results, and section 5 concludes the manuscript.

2. Design and configuration of numerical model experiments

The configuration of the control simulation (hereinafter CTRL) is nearly identical to that described in Peters and Schumacher (2015, hereinafter PS15), with the only exceptions being that version 3.7.1 of the ARW model (Skamarock et al. 2008) was used for the simulations here, slightly higher vertical resolution was used, the inner model domain was initialized slightly earlier (see below), and the simulations were run on NCAR’s Yellowstone computing facility.1 A detailed description and evaluation of the configuration of the CTRL simulation can be found in PS15, so only a brief description of this configuration is provided here. The initial and lateral boundary conditions for CTRL come from a temporal progression of composite atmospheric fields from observed warm-season heavy-rain-producing MCSs that were first developed by Peters and Schumacher (2014). The synoptic-scale conditions are characterized by an anticyclonically curved upper-level jet streak located poleward of where the MCS would develop and a low-level jet that intersects a near-surface baroclinic zone. These conditions produced strong warm-air advection (WAA) and transported low-level moisture into the vicinity of the MCS (Figs. 1 and 2). The composite fields from 15 h prior to the observed rainfall maximum were used to initialize the outer grid of the model domain (4-km horizontal grid spacing; Fig. 3), and then the lateral boundary conditions were updated every 3 h with the corresponding composite fields at 12 h, 9 h, etc., prior to the maximum rainfall time, with the model integrated for a total of 30 h. A higher-resolution nest (Fig. 3) at 1.33-km horizontal grid spacing was integrated for 24 h starting 6 h after the outer domain.2 A stretched vertical grid with 40 levels—including 8 levels in the lowest km—was used on both domains. The Yonsei University (YSU; Hong et al. 2006) planetary boundary layer (PBL) parameterization, the Thompson (Thompson et al. 2008) microphysics parameterization, the Dudhia shortwave radiation scheme (Dudhia 1989), and the Rapid Radiative Transfer Model longwave radiation scheme (Mlawer et al. 1997) were used. The lower boundary was set to a flat, constant “grassland” surface, with surface–atmosphere fluxes turned off and no land surface model used. (The advantages and disadvantages of this configuration were explored in PS15.) As in PS15, analogous simulations were run with the microphysics parameterization turned off to isolate the effects of convective processes; these simulations are abbreviated NOMP.

Fig. 1.
Fig. 1.

Large-scale environments in the (a),(c) CTRL and (b),(d) 1KM_LARGE simulations. The 900-hPa geopotential height is contoured (every 15 m), the 900-hPa wind barbs are shown for magnitudes ≥6 m s−1 (half barb = 5, full barb = 10 kt; where 1 kt = 0.5144 m s−1), the IWV is in color shading (mm), and the 300-hPa wind speed is in gray shading (m s−1). (a),(b) Model initial conditions on domain 1 and (c),(d) conditions for t = 12 h. Geographic borders are shown only to provide a sense of horizontal scale; the model land surface is homogeneous. The location of the vertical profiles in Figs. 4 and 5 is shown by an asterisk in (a).

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Fig. 2.
Fig. 2.

As in Fig. 1, but for 900-hPa temperature (every 2 K), 900-hPa temperature advection (color shading; 10−5 K s−1), and 900-hPa wind barbs for CTRL at t = 12 h.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Fig. 3.
Fig. 3.

Configuration of outer and inner model domains. Geographic borders are shown only to provide a sense of horizontal scale; the model land surface is homogeneous.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

As described by PS15, the initiation of convection was unrealistically delayed when the raw composite fields were used as ICs and LBCs. This resulted from the relative humidity being unrealistically low in the composites after averaging over many cases in a coarse reanalysis, which in turn stems partially from calculating arithmetic means of fields that are nonlinearly related (e.g., temperature and moisture, and thus CAPE and CIN) via the Clausius–Clapeyron equation. Many investigators have identified that convection is excessively suppressed in model simulations with any convective inhibition (e.g., Parker and Johnson 2004; Naylor and Gilmore 2012). To alleviate this, the initial relative humidity was increased according to the following formula:
e1
e2
where M is the added , is the from the gridded composites, is the modified relative humidity used in the ICs for CTRL, A is 10%, 900 hPa, 50 hPa, and 400 hPa. With this change applied, the initiation of convection in the quasi-idealized simulation took place at approximately t = 12 h, which is more closely aligned with the timing of convection in the observed cases that constituted the composite.
Whereas the aforementioned moisture increase was designed primarily to speed up the initiation of convection, the primary purpose of this study is to examine the influences of changes to near-surface moisture on the convection and precipitation in MCSs. To address this issue, a series of experiments was designed in which moisture was added to the initial conditions at low levels. (This approach contrasts from S15, in which moisture was reduced from an initially nearly saturated profile; here, the initial profiles are comparatively dry and moisture is added.) Four experiments were run: two with increased moisture in the lowest 1 km, and two with increased moisture in the lowest 0.6 km. Specifically, the RH below was modified on both the inner and outer grid such that
e3
where is the RH amount that was added to the composites (now expressed as a function of height), is the value of Eq. (1) at , and and are weighting coefficients. Two experiments were run with 1 km: one with and (referred to as 1KM_LARGE) and one with and (referred to as 1KM_SMALL). Two similar experiments were run with 600 m (referred to as 600M_LARGE and 600M_SMALL). In the “LARGE” perturbations, the amount of RH added to the composite grids ended up being held constant at and below . In the “SMALL” perturbations, the amount of RH added to the composites was an average between the value of Eq. (1) at and the values given by Eq. (1) below . The initial low-level moisture differences, along with how those differences evolve with time, are illustrated in Fig. 4. In the initial conditions (Fig. 4a), the low-level mixing ratio was increased by a maximum of approximately 2 g kg−1 and the RH by approximately 10% in 1KM_LARGE compared to CTRL, with progressively smaller moisture changes in the other experiments. By t = 12 h (Fig. 4b), just as convection was initiating to the north of this location, the low-level mixing ratio had increased in all simulations owing to the synoptic and mesoscale moisture advection reflected in Fig. 1), and the magnitude of the initial mixing ratio perturbations was maintained. Convection initiated within a region of mesoscale warm advection and associated low-level ascent (Fig. 2); the temperature and thermal advection fields were nearly identical prior to convection initiation in all simulations (not shown). A layer near 1500 m AGL had been lifted to saturation in all of the experiments (dashed lines in Fig. 4b), with a slightly deeper saturated layer in the 1KM experiments.
Fig. 4.
Fig. 4.

Vertical profiles at the point indicated by the asterisk in Fig. 1a of water vapor mixing ratio (g kg−1, solid lines corresponding to the bottom axis) and RH (%, dashed lines corresponding to the top axis) at t = (a) 0 and (b) 12 h.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Representative thermodynamic and kinematic profiles, evaluated just prior to convection initiation, are shown in Fig. 5, and corresponding thermodynamic calculations are given in Table 1. All of the simulations exhibited a statically stable layer near the surface, with a layer above that has been lifted to saturation as also seen in Fig. 4b. The most-unstable layer was above the surface in all simulations. The most unstable convective available potential energy (MUCAPE) and most unstable convective inhibition (MUCIN) are nearly equal in the CTRL and 600M simulations, whereas there is greater MUCAPE, and the level of the most-unstable parcel is lower, in the 1KM experiments. Surface-based parcels have substantial CAPE (as well as substantial CIN) in all of the simulations, with the added-moisture experiments having more SBCAPE and less SBCIN. To summarize the vertical profile of CAPE, the integrated CAPE (ICAPE; Mapes 1993) was also calculated, and is larger in each of the experiments going from the control to 1KM_LARGE, in accordance with the CAPE calculations mentioned above (Table 1). In the experiments, the percent increase in ICAPE above CTRL ranges from 3% in 600M_SMALL to 29% in 1KM_LARGE, and the percent increase in IWV above CTRL ranges from 0.4% in 600M_SMALL to 3.4% in 1KM_LARGE. For comparison, the magnitude of the ICAPE perturbations in S15 ranged from 3% to 15% and the IWV perturbations in S15 ranged from 0.3% to 1%. Since the moisture perturbations were made only in the low levels, the IWV integrated over only the layer between the surface and 800 hPa was also calculated (IWV800 in Table 1). The percent increases in this quantity range from 0.6% to 5.8%.

Fig. 5.
Fig. 5.

Skew T–logp diagrams at t = 12 h at the point shown in Fig. 1. The dashed red line shows the virtual temperature. Dewpoint profiles are shown using the same color scheme as in Fig. 4. Parcel curves for the most-unstable parcel in CTRL and 1KM_LARGE are shown with the black and blue dashed lines, respectively. The thermodynamic calculations for CTRL are shown; calculations for the experiments are summarized in Table 1. A magnified view of the low levels is shown at right.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Table 1.

Thermodynamic calculations for the soundings shown in Fig. 5. Differences from the control sounding are given in parentheses for selected fields; percentage differences are given for ICAPE and IWV. The prefix “MU” refers to parcels lifted from the level with the highest ; “SB” refers to parcels lifted from the lowest model level. In addition, “LPLMU” refers to the lifted parcel level, the level from which the MU parcel originated, and IWV800 is the IWV in the layer between the surface and 800 hPa. The surface dewpoint is in units of °C, CAPE and CIN fields are in J kg−1, ICAPE is in ×10−6 J m−2, the LPL and LFCs are in hPa, and IWV is in mm.

Table 1.

As further context, if the entire atmospheric column were saturated with the temperature profile unchanged, the IWV would be 66.75 mm (a 21% increase over CTRL), and if the layer from the surface to 800 hPa were saturated (with no changes to the moisture profile above this level), the IWV would be 59.97 mm (an 8.7% increase over the control) and IWV800 would be 36.94 mm (a 14.0% increase over the control). These comparisons illustrate that the low-level moisture perturbations applied in this study are substantial, but still relatively small.

3. Results

a. Overview of convective evolution and precipitation

In all of the simulations, convection initiated just before t = 12 h within the region of warm advection and organized into a forward-propagating squall line (Fig. 6), as in the simulation of PS15. The location of this initial squall line is similar in all of the simulations although the areal extent of the deep convection is larger in the added-moisture experiments (Figs. 6b–e) than in CTRL (Fig. 6a). As this squall line moves east into a region with less moisture and instability (Fig. 1), it weakens after approximately t = 18 h (not shown; see PS15 for detailed analyses). Then, between t = 18 and 22 h, a new line of convection initiates to the west of the decaying squall line (Fig. 7). However, the location and organization of this convection differs between the experiments. In CTRL, an organized west–east-oriented convective line develops well to the north of the surface outflow boundary (OFB; approximated by the 23°C potential temperature contour in Figs. 68), a process referred to as “rearward off-boundary development” (ROD; e.g., Keene and Schumacher 2013; Peters and Schumacher 2014) (Fig. 7a). In 1KM_SMALL, two bands of convection are located to the west of the decaying squall line (Fig. 7d), and in 1KM_LARGE, the new convection is less organized than in CTRL, with a collection of deep convective cores embedded within a region of lighter rainfall (Fig. 7e). Furthermore, this precipitation is located much closer to the surface OFB in the 1KM experiments. The convective structures in the 600M experiments fall somewhere in between those in the CTRL and the 1KM runs (Figs. 7b,c; also other times not shown). The reasons underlying these differences will be explored in greater detail in the next subsection. In the CTRL and 600M experiments, this convective line remains quasi-stationary for approximately 3 h, whereas the weakly organized convection moved slowly to the south in the 1KM experiments (not shown). Finally, after approximately t = 27 h, the MCS weakened and moved southward in all of the simulations, with the precipitation and associated OFB located farther south in the 1KM runs (Figs. 8d,e) than in the CTRL and 600M simulations (Figs. 8a–c).

Fig. 6.
Fig. 6.

Simulated radar reflectivity factor at 1 km AGL (color shading, dBZ), the 23°C surface potential temperature contour (blue), and 500-m AGL wind barbs (only magnitudes ≥9 m s−1 shown) at t = 15 h for (a) CTRL, (b) 600M_SMALL, (c) 600M_LARGE, (d) 1KM_SMALL, and (e) 1KM_LARGE. Red lines indicate the locations of vertical sections shown in Figs. 12 and 13.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Fig. 7.
Fig. 7.

As in Fig. 6, but at t = 23 h.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Fig. 8.
Fig. 8.

As in Fig. 6, but at t = 27 h.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

In all five simulations, extreme rainfall accumulations (>300 mm) occur as a result of this MCS (Fig. 9). Unsurprisingly, all of the added-moisture experiments yield more area-integrated accumulated precipitation than CTRL, and the increases are disproportionate to the increases in IWV (Table 2). The increases in the point-maximum precipitation are comparatively modest, with increases of 6%–19% for the low-level moisture increases applied here, and the reasons for these differences will be addressed in the next subsection. On the other hand, the total amount of rainfall produced by the MCS increases markedly with increased low-level moisture (Fig. 10). In the most extreme case, the total rainfall accumulation was approximately 60% greater in 1KM_LARGE than in CTRL, for a representative IWV increase of around 1.8 mm or 3% (Tables 1 and 2).

Fig. 9.
Fig. 9.

Total 24-h accumulated precipitation (mm) on the inner domain for (a) CTRL, (b) 600M_SMALL, (c) 600M_LARGE, (d) 1KM_SMALL, and (e) 1KM_LARGE. The red rectangle in (e) indicates the region used for area averaging in Table 2 and subsequent figures. The brown dashed rectangles in (a) and (e) indicate the averaging region for the Hovmöller diagrams in Fig. 19. In (d) and (e), the location of the maximum precipitation in CTRL is indicated with the gray plus sign (+). The area-averaged precipitation, also shown in Table 2, is given in the lower-left corner of each panel.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Table 2.

Maximum accumulated precipitation at a point and area-averaged precipitation (mm) in the numerical experiments. The area-averaged precipitation was calculated over the region shown in Fig. 9e. The differences are with respect to CTRL. Percent differences in area-averaged precipitation are equivalent to percent differences in area-integrated precipitation.

Table 2.
Fig. 10.
Fig. 10.

Time series of area-averaged rainfall accumulation (mm) over the region shown in Fig. 9e.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

In addition to the changes in rainfall amounts, there are also important changes to the rainfall locations. The location of the heavy rainfall in the 600M experiments is very similar to CTRL (cf. Figs. 9a–c), but in the 1KM experiments the heavy-rainfall axis shifts toward the southwest (Figs. 9d,e). In 1KM_SMALL, the location of the maximum rainfall occurred about 125 km to the west-southwest of that in CTRL, and in 1KM_LARGE the shift was approximately 225 km. There are also differences in the timing of the heavy rainfall across the simulations. The 1KM simulations have a much greater coverage of heavy hourly rainfall within the initial progressive squall line (from t = 14 to 16 h; see blue bars in Fig. 11a) than do the CTRL or 600M simulations. In contrast, at later times the CTRL and 600M experiments have numerous locations with heavy hourly rainfall (from t = 21 to 25 h; see black and green bars in Fig. 11a) whereas the 1KM simulations do not. Nonetheless, even with less coverage of locally heavy rainfall in the 1KM simulations later in the MCS’s lifetime, these simulations continue to produce more total (i.e., area integrated) rainfall than the CTRL and 600M simulations until after t = 26 h (Fig. 11b). These results can be partially explained by the evolution of the MCS structures highlighted above: a well-organized back-building convective line develops in CTRL, which leads to large rainfall rates within that line (Fig. 7a). With larger perturbations to the initial low-level moisture, the back-building line becomes less organized (Figs. 7b–e), but it also covers a large area with less-intense rainfall, which still results in more overall precipitation production than the focused convective line in CTRL.

Fig. 11.
Fig. 11.

(a) Number of inner-domain grid points with hourly rainfall exceeding 50 mm h−1. (b) As in Fig. 10, but for area-averaged hourly rainfall (mm).

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

To summarize: as we have seen from the convective evolution and the precipitation accumulation results presented thus far that there are substantial and complex sensitivities to low-level moisture changes in both the evolution and rainfall production in MCSs, and the following subsection will further explore the reasons for these sensitivities.

b. Differences in convective structure and evolution

What, then, led to the trends described above: the development of a strong, well-organized back-building convective line in CTRL that was displaced well north of the outflow boundary, and the evolution to a weaker, less-organized (but still heavily raining) line located nearer the gust front in the added-moisture simulations? These differences stem in large part from differences in the model atmosphere’s response to the initial squall line and its cold pool. Using the CTRL and 1KM_LARGE simulations as the extremes of the parameter space to illustrate these differences, Figs. 12 and 13 show the evolution of instability and potential temperature through the simulated MCS’s lifetime. Shortly after the initiation of deep convection, the CTRL and 1KM_LARGE simulations have similar vertical structures of CAPE and potential temperature, aside from the increased CAPE in 1KM_LARGE that results from the imposed moisture perturbation (Figs. 12a,b). However, the two simulations become rather different after the development of the forward-propagating squall line. In 1KM_LARGE, the CAPE has been eliminated between y = 120 and 200 km (Fig. 12d), a cold pool with maximum θ deficit of over 4 K has developed (Fig. 13d), and deep convection is occurring at the OFB. In contrast, CTRL has a comparatively weak and shallow cold pool (Fig. 13c), little deep convection in the plane of the vertical section, and much of the initial CAPE still remains or has increased with time (Fig. 12c). In 1KM_LARGE, the cold pool continues to strengthen and move southward with time (Figs. 13f,h), with unstable air almost exclusively confined to the south of the OFB (Figs. 12f,h). By t = 23 h, convection is comparatively weak but covers a large area to the north of the surface OFB (Fig. 12h; see also Fig. 7e). In CTRL, CAPE quickly returns to the region above the weak cold pool, although it is reduced somewhat after air is lifted at the OFB (Figs. 12e and 13e); this change in stability is analyzed in detail by Peters and Schumacher (2016, hereinafter PS16). ROD takes place, with an organized convective line initiating approximately 90 km north of the surface OFB and remaining nearly stationary for 3–4 h (Figs. 12e,g; see also Figs. 13e,g and 7a).

Fig. 12.
Fig. 12.

Vertical sections of CAPE for parcels lifted from each vertical level (color shading in J kg−1), potential temperature (black contours every 2 K), and vertical velocity (blue contours at 0.5 m s−1 and every 1 m s−1 above that). All values have been averaged over 15 grid points (20 km) on either side of line A–B shown in Fig. 6a. Results are shown for the (left) CTRL and (right) 1KM_LARGE simulations. Times shown are t = (a),(b) 14; (c),(d) 17; (e),(f) 20; and (g),(h) 23 h.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Fig. 13.
Fig. 13.

As in Fig. 12, but the color-shaded field is the potential temperature perturbation [defined as the difference between the full simulation and a corresponding simulation with the microphysics parameterization turned off (NOMP)].

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

The fates of air parcels passing through the CTRL and 1KM_LARGE MCSs further illustrate the differences in the convective structure, evolution, and location. Trajectories were calculated from 10-min model output for a total of 900 air parcels originating just south of the surface OFB at t = 19 h, at initial heights of 100, 200, and 300 m AGL; and for another 900 parcels originating in regions of mesoscale isentropic ascent to the southwest of the MCS, at initial heights of 1.5, 1.6, and 1.7 km AGL. Because the cold pool moved farther south in 1KM_LARGE than in CTRL, the initial “grid” of surface-based parcels was shifted accordingly (cf. Figs. 14a,d). The regions of mesoscale ascent (Figs. 15a,c) were identified by local maxima in the ingredients-based propagation index discussed below. Three hours after the release of the near-surface trajectories, the air parcels in CTRL rise slightly over the shallow cold pool but remain in the lowest 2 km (Fig. 14b). In contrast, many of the parcels in 1KM_LARGE are lifted along the edge of the deeper, stronger cold pool and, considering their larger CAPE and smaller CIN, ascend in deep convection (Fig. 14e). By t = 24 h, the parcel trajectories suggest that the convection in 1KM_LARGE has tilted rearward over the cold pool somewhat (Fig. 14f), consistent with the differences in the convective structures seen in Figs. 12h and 13h. In CTRL, the near-surface parcels mainly follow two paths after having been slightly lifted at the surface OFB: most remain near the surface, but some ascend in the focused line of ROD convection discussed previously (Fig. 14c). Figure 16 quantifies the fates of the near-surface air parcels in the respective simulations. In CTRL, the majority of the near-surface air parcels never rise above 2 km AGL, relatively few reach a maximum height at midlevels, but approximately 10% ascend to above 10 km AGL. In contrast, the parcels in 1KM_LARGE achieve a much more even distribution of heights, with only approximately 7% of the parcels remaining below 1 km and 20% remaining below 2 km. The evolution of the elevated parcels is somewhat different: in CTRL, many parcels originating within the mesoscale ascent to the southwest of the convection entered deep convective updrafts (Figs. 15a,b), whereas only a few of the elevated parcels from this region ascended in 1KM_LARGE (Figs. 15c,d). (By the end of the simulation, 13% of these elevated parcels had ascended above 10 km in CTRL, but only 5% of the elevated parcels in 1KM_LARGE had done so.) This result, in conjunction with the analysis of the near-surface trajectories, indicates that the CTRL MCS was “elevated” to a greater extent than the MCS in 1KM_LARGE, reinforcing S15’s findings that, in MCSs forming in an environment with a stable boundary layer, the proportion of surface-based parcels that rise in deep convection is very sensitive to the details of the low-level moisture profile.

Fig. 14.
Fig. 14.

(a),(d) Locations of parcel trajectories originating at t = 19–24 h in (a) CTRL and (d) 1KM_LARGE. The 24°C surface temperature contour at t = 19 h is shown in blue to indicate the location of the outflow boundary in each simulation. The size of the arrowhead indicates the height of the parcel at t = 24 h, according to the key at the bottom right. Trajectories were initialized at every sixth model grid point within the red rectangle at 100, 200, and 300 m AGL for a total of 900 trajectories. In (a),(d), a random selection of 150 trajectories is shown for clarity. (b),(c),(e),(f) Vertical section along line A–B in Fig. 6a showing potential temperature perturbation (color shading, K) and the vertical location of air parcel trajectories through t = (b),(e) 22 and (c),(f) 24 h. All 900 trajectories are shown in (b),(c),(e),(f).

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Fig. 15.
Fig. 15.

As in Fig. 14, but for trajectories originating at 1.5, 1.6, and 1.7 km AGL within the red rectangles shown. (a),(c) Locations of parcel trajectories originating at t = 19–24 h in (a) CTRL and (c) 1KM_LARGE. (b),(d) Vertical section along line C–D in (a) showing the potential temperature perturbation (color shading, K) and the vertical location of air parcel trajectories through t = 22 h.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Fig. 16.
Fig. 16.

Histogram of the maximum height attained by the air parcels between t = 19 and 24 h. For example, the leftmost red bar indicates that 25% of parcels in CTRL had a maximum height of 1 km AGL or below.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

To further demonstrate that the aforementioned differences in mesoscale ascent, and cold pool intensity and depth, were primarily responsible for the location differences in the MCS precipitation between CTRL and 1KM_LARGE, we employ the ingredients-based propagation index (IPI) developed by PS16. The purpose of this parameter was to identify regions of the atmosphere where air parcels (i) had nonzero CAPE, (ii) were lifting adiabatically, (iii) were close to saturation, and (iv) had little convective inhibition. These parcels were deemed to have the highest probability of convection initiation and, therefore, indicate locations toward which the MCS is likely to propagate. As was noted by PS16, IPI is not an effective method for identifying the locations of outflow-driven dynamic lifting (as was described by, e.g., Rotunno et al. 1988) and the associated MCS propagation that may occur as a result of this process. It was further discussed in PS2016, however, that the orientation of the low-level wind shear along the southern, southwestern, and western flanks of the cold pool was unfavorable for dynamically driven outflow-induced propagation, having favored comparatively gradual isentropic ascent (which IPI was designed to identify) instead. The IPI parameter was therefore defined as
e4
where and are normalization parameters and IPI is set to zero for all points where RH < 95% and CIN > 10 J kg−1. The parameter is warm-air advection (formally set to 0 for negative values), which is a good indicator of regions of adiabatic ascent within flow with nonzero horizontal velocity (see PS16 for a more in-depth justification of the inclusion of WAA in the IPI formulation). The value for of 2.5 × 10−4 K s−1 is the approximate time-averaged domain maximum WAA for all simulations, and the RH and CIN thresholds were arbitrarily defined and justified in an a posteriori manner (given the demonstrated usefulness of these parameters later in this section). The values represented the approximate time-averaged domain maximum CAPE, which was 2000 J kg−1 for the CTRL simulation and 3000 J kg−1 for the 1KM_LARGE simulation. The choice to use different values of for the different simulations was motivated by the convenience of having similar maximum values of IPI for ease of comparison of spatial patterns. The higher overall CAPE in the 1KM_LARGE simulation does not necessarily change the likelihood of convection initiation in regions where CAPE and overlap; instead, it is the times and spatial locations of maxima and minima in IPI that are important in its interpretation. In other words, because of the (somewhat arbitrary) normalization factors for CAPE, IPI should be interpreted as a relative quantity (where are the large values with respect to space and time?) rather than an absolute quantity (what is the value of IPI?). Using the same normalization factor for both runs would yield higher values of IPI everywhere in 1KM_LARGE, but the locations/times of the maxima would be the same, and thus the interpretation would also be the same.
To separate adiabatic lift from processes “external” to the MCS (such as broad-scale WAA) from lift due to processes “internal” to the MCS (such as flow ascending along the edge of a convectively generated cold pool), we divide the horizontal temperature advection () into its contributions from a simulation with full microphysics (and that contains an MCS) and its contributions from the analogous NOMP simulations:
e5
We define (where only is included), with for RH < 95%, CIN > 10 J kg−1 and . The contribution to IPI resulting from is
e6
The contribution to IPI resulting from is
e7
Then, we take the maximum column value of IPI at a given horizontal location (so that we may examine spatial maps of its distribution). It was shown in PS16 that identified regions where parcels with nonzero CAPE were being lifted within large-scale WAA, whereas identified regions where parcels with nonzero CAPE were being locally lifted along an OFB. Here, considering the differing normalization factors mentioned above, we will focus primarily on the relative magnitudes of and for each simulation individually (rather than comparing values between runs), and also the proximity of maxima in these quantities to ongoing convection.

As discussed previously (e.g., Fig. 2), the MCSs studied here formed within a region of mesoscale WAA, and this WAA persisted upstream of the MCS through t = 24 h. This time was also approximately when CAPE was maximized in this area. This yielded a local maximum in along the western flank of the MCS in CTRL through t = 24 h (Figs. 17a–e), whereas the relatively high values of in 1KM_LARGE were displaced far from the western flank of the MCS (Figs. 17b,d,f). In CTRL, the magnitudes of the maxima in were much larger than those in through t = 22 h (cf. Figs. 17a,c and 18a,c), but the opposite was true in 1KM_LARGE: the maxima in along the OFB had much greater magnitudes, and were nearer to the ongoing convection, than the maxima in (cf. Figs. 17b,d and 18b,d). These results demonstrate that in CTRL, large-scale environmental lifting played a crucial role in continuously regenerating new convective cells along the western flank of the system, whereas in 1KM_LARGE, lifting along the convectively generated outflow boundary was more important and large-scale lifting played a comparatively smaller role. In particular, whereas the southwestern OFB in the CTRL run was insufficiently deep and strong to lift air parcels to their LFCs and initiate convection (as demonstrated by PS16) until late in the simulation, the lift along the cold pool on the MCS’s southwestern flank in the 1KM_LARGE simulation was sufficient to directly initiate convection at a much earlier time than in the CTRL simulation. By t = 24 h, the story in CTRL changed somewhat, with increasing to the south of the convective line (but north of the original OFB; Fig. 18e); this preceded the southward surge of the MCS near the end of the CTRL simulation (e.g., Fig. 8a).

Fig. 17.
Fig. 17.

Maximum column (shading), surface K (green contour), and maximum column w > 3 m s−1 (dark gray contours) for the (left) CTRL and (right) 1KM_LARGE simulations, valid at t = (a),(b) 20; (c),(d) 22; and (e),(f) 24 h.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Fig. 18.
Fig. 18.

Maximum column (shading), cold pool intensity (c, red contours at intervals of 5 m s−1, defined as in PS16), and maximum column w > 3 m s−1 (dark gray contours) for the (left) CTRL and (right) 1KM_LARGE simulations, valid at tsim = (a),(b) 20; (c),(d) 22; and (e),(f) 24 h.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

The aggregate impact of the differences in cold pool intensities between the 1KM_LARGE and CTRL simulations is concisely demonstrated through the time–latitude evolution of 1-h precipitation, , and (Figs. 19a,b). The initial convection in both simulations developed in a region of , which shows that large-scale warm-air advection contributed to the initiation of the MCS. In the CTRL simulation, became prevalent along the western flank of the system after t = 20 h (Fig. 19a), in conjunction with back-building on the western flank of the system having anchored at approximately 89.5°W (Fig. 19a). In contrast, in the 1KM_LARGE simulation became prevalent on the western flank of the system by t = 14 h, and persisted through the remainder of the MCS’s lifetime, suggesting that cold-pool-driven lift played a role in upstream back-building much earlier in this simulation. Maximum was also farther west in the 1KM_LARGE simulation than in CTRL (Fig. 19b; as discussed in the previous paragraph, this is a result of having been maximized along the southwestern OFB in the 1KM_LARGE simulation, rather than northeast of the southwestern OFB in CTRL). This resulted in upstream back-building having occurred over a degree of longitude farther west in the 1KM_LARGE simulation than the CTRL simulation (Fig. 19b).

Fig. 19.
Fig. 19.

(a) Hovmöller diagram of 1-h CTRL precipitation accumulation (shading, mm), (red contours), and (black contours), averaged north-to-south over the dashed box in Fig. 9a. (b) As in (a), but for the 1KM_LARGE simulation, and averaged over the dashed box in Fig. 9e. The blue arrows indicate periods when cold-pool-driven back-building was occurring.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

It is logical to wonder at this point why the MCS did not simply move westward given the demonstrated maxima in IPI along the western flank of the system, and to further wonder why there were successive eastward-moving convective episodes along the system’s east flank. It was shown in PS16 that the mean tropospheric wind was oriented toward the east-northeast, and that dynamically driven propagation along the outflow edge occurred along the east and southeastern cold pool peripheries. Both of these factors (neither of which are explicitly identified by IPI) contributed an eastward component to the system’s motion (and the aforementioned eastward-moving convective episodes). On the western flank of the system, the westward component of propagation from isentropic ascent nearly canceled the eastward propagation from other factors, and led to nearly net-zero motion of this flank of the system.

In summary, the IPI diagnostics support the following hypothesis: (i) the cold pool on the southwestern flank of the CTRL MCS was insufficiently deep to directly initiate convection in the CTRL simulation until after t = 20 h, whereas the cold pool on the southwestern flank of the 1KM_LARGE MCS was sufficiently deep to initiate convection by t = 14 h; (ii) since the elevated inflow into the MCS was southwesterly, air needed to travel farther to the northeast in the CTRL simulation before convection initiated than in the 1KM_LARGE simulation; and (iii) these differences in the locations of convection initiation were ultimately responsible for the differing positions of the maximum accumulated rainfall between the two MCSs.

One possibly counterintuitive result shown above is that in the simulations with higher RH at low levels (e.g., 1KM_LARGE), which, all else being equal, should reduce evaporation, the convectively generated cold pools are stronger and deeper. This seeming discrepancy is easily resolved, however, when considering that the increased moisture and instability in the added-moisture runs result in a much greater amount of total condensate (and frozen hydrometeors). Averaged over the entire MCS, the near-surface latent cooling is slightly greater in 1KM_LARGE than in CTRL, as a result of the much greater latent heating (owing to condensation, freezing, and deposition of hydrometeors) aloft (Fig. 20). Furthermore, this effect is amplified in the convectively active portions of the MCS, leading to the locally much stronger and deeper cold pools in the added-moisture experiments. These results are consistent with the findings of Coniglio et al. (2010), who showed that more CAPE, and greater potential instability, supported stronger cold pools and more rapid upscale growth of MCSs. (Low-level potential instability is not explicitly shown here, but is also greater in the added-moisture experiments than in CTRL.) The results are also reminiscent of the findings of James and Markowski (2010) in their examination of dry layers aloft: despite the dry layer being favorable for more evaporation, the squall lines that formed in these environments produced less hydrometeor mass and thus less total evaporation.

Fig. 20.
Fig. 20.

Vertical profiles of accumulated latent heating (K), averaged over the region shown in Fig. 9e, through t = 24 h in the CTRL (black) and 1KM_LARGE (blue) simulations.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

Up to this point, attention has mainly been given to the CTRL and 1KM_LARGE experiments, because they represent the far ends of the parameter space considered in this study. An examination of the “intermediate” experiments reveals that there is a straightforward progression of convective behaviors between these two extremes. For example, the cold pools produced by the initial convective lines are slightly stronger in the 600M experiments than in CTRL (cf. Figs. 12e and 21a,b), and stronger still in 1KM_SMALL (Fig. 21c), but not as strong as in 1KM_LARGE (Fig. 12f). Similarly, the northward extent of high-CAPE air above the cold pool transitions smoothly across these experiments (Figs. 21a–c): the larger the initial moisture perturbation, the greater the removal of CAPE behind the OFB (or, conversely, there is less recovery of CAPE after the initial squall-line passage to support subsequent convection initiation behind the OFB). This ultimately results in the heaviest rain falling in nearly the same location in CTRL and the 600M simulations, but with increased accumulations as the initial moisture is increased (Figs. 9a–c). In 1KM_SMALL, the rainfall axis was shifted slightly to the southwest compared to CTRL, but not as far southwest as 1KM_LARGE (Figs. 9d,e). Thus, although the increases in precipitation accumulation in response to the increased moisture are very nonlinear (Table 2), the fact that the changes in the MCS structure represent a well-constrained continuum as the moisture changes are increased gives added confidence in the robustness of the results presented above.

Fig. 21.
Fig. 21.

As in Figs. 12e,f, but for (a) 600M_SMALL, (b) 600M_LARGE, and (c) 1KM_SMALL at t = 20 h.

Citation: Monthly Weather Review 145, 6; 10.1175/MWR-D-16-0255.1

4. Discussion

The results of the numerical experiments presented above add to the body of evidence demonstrating that precipitation accumulation has a highly nonlinear response to changes in atmospheric water vapor. Related studies that have examined the sensitivity of precipitation to vertical moisture profiles include Takemi (2006), S15, and Alfaro and Khairoutdinov (2015). Yet these studies did not include the atmospheric forcing representative of midlatitude warm-season MCSs, which is included here. In particular, this forcing includes synoptic-to-mesoscale ascent and rapid horizontal transport and convergence of moisture. The outcomes of the experiments—especially the 1KM experiments—show that the increase in the precipitation production of a midlatitude MCS can far exceed what might be implied by a quantity like IWV, with the 3.4% increase in IWV in the 1KM_LARGE experiment resulting in a nearly 60% increase in total MCS precipitation. These precipitation increases are also disproportionate to those found previously in the literature. For example, in the high-shear simulations of Alfaro and Khairoutdinov (2015), variations in IWV from 43.5 to 46.6 mm (approximately 7%), which corresponded to variations in ICAPE from 3.3 to 6.2 × 10−6 J m−2, resulted in variations in hourly rain rate from 20 to 27 mm, or 35%. A similar comparison of hourly rain rates from the simulations shown in the present study indicates increases of up to 66% for IWV increases of 3.4% and ICAPE increases of 4.15–5.35 × 10−6 J m−2. Of course, because the moisture increases here were applied over an area much larger than the MCS itself, and because of the synoptic and mesoscale forcing in which the MCS develops, these relatively small increases in moisture locally can be processed by the convective system over its full life cycle to result in large rainfall accumulations. These factors were not fully included in the aforementioned studies, and appear to have important influences both on rainfall production itself, and on the sensitivity of rainfall production to the environmental conditions. Or from the perspective of NWP models, the results suggest that small biases or errors in initial moisture conditions can lead to very large differences in accumulated rainfall. This may have important implications for two different lines of inquiry related to MCS precipitation.

The first is for the predictive skill and predictability of heavy rainfall. As shown here, and as well as by S15, small changes to the low-level moisture profile lead to large changes in the total MCS rainfall, and this study additionally shows that the location of the heavy precipitation can also be changed substantially. As discussed in the introduction, forecast skill for warm-season rainfall continues to be low, and the results of this study reveal some potential reasons why: even when the large-scale forcing for ascent is well represented [itself not a given in real-world scenarios; e.g., Peters and Roebber (2014)], errors in the low-level moisture profile can lead to very different MCS structures and, as a result, precipitation distributions. One could think of the MCS environments represented in this study as a simple ensemble of forecasts and, although all of the “members” of this ensemble would indicate the possibility of an extreme rainstorm, the shift in the heaviest rainfall location by over 200 km as a response to a relatively small moisture perturbation would not give a forecaster high confidence in where to expect a threat for heavy rain and flash flooding. As efforts to couple high-resolution weather forecasts with hydrologic model predictions of flooding continue to advance (e.g., Bierkens et al. 2015) these sorts of uncertainties in the location of the precipitation axis are concerning. The results do provide yet further motivation for representing heavy precipitation forecasts probabilistically, however.

The second relevant research area is the question of how extreme precipitation has changed, and may continue to change, within the context of a changing climate. All else being equal, atmospheric IWV should respond to warming at a rate roughly commensurate with the Clausius–Clapeyron relation [approximately 7% (1°C)−1 of warming; e.g., Berg et al. 2013]. However, this study shows that relatively small increases in moisture (if they occur near the surface) can lead to very large increases in precipitation accumulation. The moisture perturbations applied here are likely too idealized to reflect actual atmospheric responses to changes in radiative forcing (e.g., the temperature itself was not altered here, and perturbations were limited solely to low levels), but the high sensitivity that experiments reveal is nonetheless concerning. In a “pseudo-global-warming” experiment that reflects more representative atmospheric changes, Lackmann (2013) similarly showed large increases in rainfall and convective intensity in simulations of an observed extreme rainfall event, and Groisman et al. (2012) and Berg et al. (2013), among others, have shown that heavy convective precipitation has become more frequent in observations.

Additional areas warranting further study are the connections between (and uncertainties associated with) the land surface and low-level atmospheric moisture. This study did not directly include the effects of surface–atmosphere fluxes, nor did it address the possible influences of land surface inhomogeneity. One could envision that these processes can serve as sources of low-level moisture errors and uncertainties in models, and the importance of these processes on MCSs and their precipitation remains poorly understood.

Finally, the results presented in this study relate closely to observations that were collected during the PECAN field campaign (Geerts et al. 2017) in 2015. In particular, an MCS that shared many characteristics of the idealized MCS simulated in this study occurred on 24–25 June 2015 in eastern Iowa. Radiosonde observations taken at high temporal frequency showed rapid moistening and destabilization of the environment near the MCS, which was not well represented in operational analyses that did not assimilate these observations. Numerical simulations of this case (Peters et al. 2017) suggest that the location of this MCS was sensitive to the magnitude of these low-level moisture errors in much the same way as the simulated MCSs in this study were. Further exploration of the unprecedented observations of MCS environments during PECAN may reveal further insights regarding the reasons for errors in forecasts of MCS precipitation.

5. Conclusions

A series of numerical model experiments addressed how the distribution of precipitation produced by a semi-idealized warm-season MCS was influenced by changes to low-level moisture. In two experiments, the moisture was increased slightly over the lowest 600 m, and in two other experiments it was increased over the lowest 1 km. As expected, in all of the simulations, the total rainfall accumulation from the MCS increased in response to the added low-level moisture. The increases in the point rainfall maxima were comparatively modest, with all of the experiments producing over 300 mm locally. But the increase in total rainfall was highly disproportionate to the moisture perturbation: in the experiment with the largest moisture addition, a 3.4% increase in IWV resulted in a nearly 60% increase in total rainfall. Furthermore, in the experiments in which the moisture perturbation was applied over a 1-km layer, the location of the heaviest rainfall also shifted. In these runs, a stronger and faster-moving cold pool developed and the deep convection remained closer to the edge of this cold pool, whereas the convection was displaced farther behind the outflow boundary in the control simulation and those with the shallower moisture perturbations. Although the MCSs in the 1KM experiments produced much more rainfall in total, the control and 600M MCSs were able to produce large short-term rainfall much later in their lifetimes owing to cold-pool-induced differences in convective structures.

Acknowledgments

This research was supported by National Science Foundation Grants AGS-1359727 and AGS-PRF 1524435. High-performance computing resources from Yellowstone (ark:/85065/d7wd3xhc) were provided by NCAR’s Computational and Information Systems Laboratory, which is sponsored by the National Science Foundation. We thank three anonymous reviewers for their very thoughtful and constructive suggestions that resulted in important improvements to the manuscript.

REFERENCES

  • Alfaro, D. A., and M. Khairoutdinov, 2015: Thermodynamic constraints on the morphology of simulated midlatitude squall lines. J. Atmos. Sci., 72, 31163137, doi:10.1175/JAS-D-14-0295.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ashley, W. S., T. L. Mote, G. Dixon, S. L. Trotter, E. J. Powell, J. D. Durkee, and A. J. Grundstein, 2003: Distribution of mesoscale convective complex rainfall in the United States. Mon. Wea. Rev., 131, 30033017, doi:10.1175/1520-0493(2003)131<3003:DOMCCR>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., and et al. , 2016: A North American hourly assimilation and model forecast cycle: The Rapid Refresh. Mon. Wea. Rev., 144, 16691694, doi:10.1175/MWR-D-15-0242.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Berg, P., C. Moseley, and J. O. Haerter, 2013: Strong increase in convective precipitation in response to higher temperatures. Nat. Geosci., 6, 181185, doi:10.1038/ngeo1731.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bierkens, M. F. P., and et al. , 2015: Hyper-resolution global hydrological modelling: What is next? Hydrol. Processes, 29, 310320, doi:10.1002/hyp.10391.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bretherton, C. S., M. E. Peters, and L. E. Back, 2004: Relationships between water vapor path and precipitation over the tropical oceans. J. Climate, 17, 15171528, doi:10.1175/1520-0442(2004)017<1517:RBWVPA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Coniglio, M. C., 2012: Verification of RUC 0–1-h forecasts and SPC mesoscale analyses using VORTEX2 soundings. Wea. Forecasting, 27, 667683, doi:10.1175/WAF-D-11-00096.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Coniglio, M. C., J. Y. Hwang, and D. J. Stensrud, 2010: Environmental factors in the upscale growth and longevity of MCSs derived from Rapid Update Cycle analyses. Mon. Wea. Rev., 138, 35143539, doi:10.1175/2010MWR3233.1; Corrigendum, 139, 26862688, doi:10.1175/MWR-D-11-00064.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Coniglio, M. C., J. Correia Jr., P. T. Marsh, and F. Kong, 2013: Verification of convection-allowing WRF model forecasts of the planetary boundary layer using sounding observations. Wea. Forecasting, 28, 842862, doi:10.1175/WAF-D-12-00103.1.

    • Crossref
    • 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.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fritsch, J. M., and R. E. Carbone, 2004: Improving quantitative precipitation forecasts in the warm season: A USWRP research and development strategy. Bull. Amer. Meteor. Soc., 85, 955965, doi:10.1175/BAMS-85-7-955.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fritsch, J. M., R. J. Kane, and C. R. Chelius, 1986: The contribution of mesoscale convective weather systems to the warm-season precipitation in the United States. J. Appl. Meteor., 25, 13331345, doi:10.1175/1520-0450(1986)025<1333:TCOMCW>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Geerts, B., and et al. , 2017: The 2015 Plains Elevated Convection At Night field project. Bull. Amer. Meteor. Soc., 98, 767786, doi:10.1175/BAMS-D-15-00257.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Groisman, P. Y., R. W. Knight, and T. R. Karl, 2012: Changes in intense precipitation over the central United States. J. Hydrometeor., 13, 4766, doi:10.1175/JHM-D-11-039.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 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.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Houze, R. A., Jr., 2004: Mesoscale convective systems. Rev. Geophys., 42, RG4003, doi:10.1029/2004RG000150.

  • James, R. P., and P. M. Markowski, 2010: A numerical investigation of the effects of dry air aloft on deep convection. Mon. Wea. Rev., 138, 140161, doi:10.1175/2009MWR3018.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • James, R. P., P. M. Markowski, and J. M. Fritsch, 2006: Bow echo sensitivity to ambient moisture and cold pool strength. Mon. Wea. Rev., 134, 950964, doi:10.1175/MWR3109.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Keene, K. M., and R. S. Schumacher, 2013: The bow and arrow mesoscale convective structure. Mon. Wea. Rev., 141, 16481672, doi:10.1175/MWR-D-12-00172.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lackmann, G. M., 2013: The south-central U.S. flood of May 2010: Present and future. J. Climate, 26, 46884709, doi:10.1175/JCLI-D-12-00392.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mapes, B. E., 1993: Gregarious tropical convection. J. Atmos. Sci., 50, 20262037, doi:10.1175/1520-0469(1993)050<2026:GTC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Market, P., S. Allen, R. Scofield, R. Kuligowski, and A. Gruber, 2003: Precipitation efficiency of warm-season midwestern mesoscale convective systems. Wea. Forecasting, 18, 12731285, doi:10.1175/1520-0434(2003)018<1273:PEOWMM>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul, E. W., Jr., C. Cohen, and C. Kirkpatrick, 2005: The sensitivity of simulated storm structure, intensity, and precipitation efficiency to environmental temperature. Mon. Wea. Rev., 133, 30153037, doi:10.1175/MWR3015.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Melhauser, C., and F. Zhang, 2012: Practical and intrinsic predictability of severe and convective weather at the mesoscales. J. Atmos. Sci., 69, 33503371, doi:10.1175/JAS-D-11-0315.1.

    • Crossref
    • 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.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Naylor, J., and M. S. Gilmore, 2012: Convective initiation in an idealized cloud model using an updraft nudging technique. Mon. Wea. Rev., 140, 36993705, doi:10.1175/MWR-D-12-00163.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Parker, M. D., and R. H. Johnson, 2004: Simulated convective lines with leading precipitation. Part II: Evolution and maintenance. J. Atmos. Sci., 61, 16561673, doi:10.1175/1520-0469(2004)061<1656:SCLWLP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peters, J. M., and P. J. Roebber, 2014: Synoptic control of heavy-rain-producing convective training episodes. Mon. Wea. Rev., 142, 24642482, doi:10.1175/MWR-D-13-00263.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peters, J. M., and R. S. Schumacher, 2014: Objective categorization of heavy-rain-producing MCS synoptic types by rotated principal component analysis. Mon. Wea. Rev., 142, 17161737, doi:10.1175/MWR-D-13-00295.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peters, J. M., and R. S. Schumacher, 2015: The simulated structure and evolution of a quasi-idealized warm season convective system with a training convective line. J. Atmos. Sci., 72, 19872010, doi:10.1175/JAS-D-14-0215.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peters, J. M., and R. S. Schumacher, 2016: Dynamics governing a simulated mesoscale convective system with a training convective line. J. Atmos. Sci., 73, 26432664, doi:10.1175/JAS-D-15-0199.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Peters, J. M., E. R. Nielsen, M. D. Parker, S. M. Hitchcock, and R. S. Schumacher, 2017: The impact of low-level moisture errors on model forecasts of an MCS observed during PECAN. Mon. Wea. Rev., doi:10.1175/MWR-D-16-0296.1, in press.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rotunno, R., J. B. Klemp, and M. L. Weisman, 1988: A theory for strong, long-lived squall lines. J. Atmos. Sci., 45, 463485, doi:10.1175/1520-0469(1988)045<0463:ATFSLL>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schumacher, R. S., 2015: Sensitivity of precipitation accumulation in elevated convective systems to small changes in low-level moisture. J. Atmos. Sci., 72, 25072524, doi:10.1175/JAS-D-14-0389.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schumacher, R. S., T. J. Galarneau Jr., and L. F. Bosart, 2011: Distant effects of a recurving tropical cyclone on rainfall in a midlatitude convective system: A high-impact predecessor rain event. Mon. Wea. Rev., 139, 650667, doi:10.1175/2010MWR3453.1.

    • Crossref
    • 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.

    • Crossref
    • Export Citation
  • Stevenson, S. N., and R. S. Schumacher, 2014: A 10-year survey of extreme rainfall events in the central and eastern United States using gridded multisensor precipitation analyses. Mon. Wea. Rev., 142, 31473162, doi:10.1175/MWR-D-13-00345.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sukovich, E. M., F. M. Ralph, F. E. Barthold, D. W. Reynolds, and D. R. Novak, 2014: Extreme quantitative precipitation forecast performance at the Weather Prediction Center from 2001 to 2011. Wea. Forecasting, 29, 894911, doi:10.1175/WAF-D-13-00061.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Takemi, T., 2006: Impacts of moisture profile on the evolution and organization of midlatitude squall lines under various shear conditions. Atmos. Res., 82, 3754, doi:10.1016/j.atmosres.2005.01.007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Takemi, T., 2010: Dependence of the precipitation intensity in mesoscale convective systems to temperature lapse rate. Atmos. Res., 96, 273285, doi:10.1016/j.atmosres.2009.09.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thompson, G., P. R. Field, R. M. Rasmussen, and W. D. 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.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wandishin, M. S., D. J. Stensrud, S. L. Mullen, and L. J. Wicker, 2008: On the predictability of mesoscale convective systems: Two-dimensional simulations. Wea. Forecasting, 23, 773785, doi:10.1175/2008WAF2007057.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wandishin, M. S., D. J. Stensrud, S. L. Mullen, and L. J. Wicker, 2010: On the predictability of mesoscale convective systems: Three-dimensional simulations. Mon. Wea. Rev., 138, 863885, doi:10.1175/2009MWR2961.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., and J. B. Klemp, 1982: The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev., 110, 504520, doi:10.1175/1520-0493(1982)110<0504:TDONSC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
1

As found by many other investigators, different computer systems yield different results in WRF simulations. In our experiments, the total rainfall accumulations in CTRL did indeed change across computer systems, but the differences between the experiments and the control run were nearly identical in magnitude.

2

Based on the suggestion of an anonymous reviewer, we conducted a test with a larger inner domain, which resulted in very little change to the results.

Save