SEARCH

SEARCH BY CITATION

Keywords:

  • ARM-SGP;
  • MC3E;
  • WRF-SBM;
  • disdrometer;
  • mesoscale modeling;
  • rainfall analysis

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Description of Model Simulation and Overall Condition on 25 April 2011 at SGP Site
  5. 3. Results
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[1] A unique microphysical structure of rainfall is observed by the surface laser optical Particle Size and Velocity (Parsivel) disdrometers on 25 April 2011 during Midlatitude Continental Convective Clouds Experiment (MC3E). According to the systematic differences in rainfall rate and bulk effective droplet radius, the sampling data can be divided into two groups; the rainfall mostly from the deep convective clouds has relatively high rainfall rate and large bulk effective droplet radius, whereas the reverse is true for the rainfall from the shallow wrm clouds. The Weather Research and Forecasting model coupled with spectral bin microphysics (WRF-SBM) successfully reproduces the two distinct modes in the observed rainfall microphysical structure. The results show that the up-to-date model can demonstrate how the cloud physics and the weather condition on the day are involved in forming the unique rainfall characteristic.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Description of Model Simulation and Overall Condition on 25 April 2011 at SGP Site
  5. 3. Results
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[2] Precipitation is a key component of earth's energy and water cycle. It is an integral product of hydrometeor size distribution where the size of the hydrometeors ranges from a small drizzle drop (≥0.1 mm in diameter) to a hailstone (>10 cm in diameter). Several processes are responsible for determining the size and phase of individual hydrometeors at the surface including but not limited to aerosol loading, liquid water availability, structures of vertical and horizontal wind velocities, presence or absence of mixed-phase processes, and subcloud layer evaporation.

[3] Microphysical structure of precipitation at the ground is of interest in developing cloud microphysical models, in terms of improving forecast accuracy of surface precipitation rate as well as validating cloud microphysics by regarding precipitation as a product. Bulk cloud microphysics is commonly used in the present atmospheric models; however, there is little room for change in the precipitation microphysics because certain mathematical functions are used as built-in parameterization to represent particle size distribution (PSD). In contrast, spectral bin microphysics (SBM) allows prediction of hydrometeor PSDs and subsequently explicit representation of precipitation microphysical structure. Numerical weather prediction (NWP) simulations using SBM have been utilized mostly in the research field of cloud-aerosol interactions in previous studies [e.g.,Lynn et al., 2005]. Analysis and verification of the ground precipitation microphysics can be a new application of the simulation to show possibilities of the up-to-date models.

[4] This paper reports on the analysis of distinct rainfall microphysical structures associated with deep convective system and shallow boundary layer clouds in the rainfall event on 25 April 2011. The event was observed during the Midlatitude Continental Convective Clouds Experiment (MC3E) at the Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP) site in central Oklahoma. The disdrometer-based observations for raindrop size distribution (DSD) are analyzed to determine the characteristic of the rainfall, which is compared with a simulation result of the Weather Research and Forecasting model coupled with spectral bin microphysics (WRF-SBM).

2. Description of Model Simulation and Overall Condition on 25 April 2011 at SGP Site

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Description of Model Simulation and Overall Condition on 25 April 2011 at SGP Site
  5. 3. Results
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[5] The WRF-SBM has been developed for supporting the National Aeronautics and Space Administration (NASA) Global Precipitation Measurement (GPM) mission; the MC3E field campaign is a part of GPM Ground validation (GV) program. The SBM part is the same as the microphysical scheme of the Hebrew University Cloud Model (HUCM) [Khain et al., 2011]. The scheme has been implemented into the Advanced Research WRF (ARW) version 3.1.1 [Iguchi et al., 2012]. Hydrometeors are categorized into one-water and six-ice classes, i.e., water droplets, ice crystals (plate, column, dendrite), snow aggregates, graupel, and hail; the discrete PSDs of the hydrometeor classes are represented on a grid containing 43 doubling mass bins. In order to simulate the entire rainfall event, the simulation domain covers 454 × 388 horizontal grids domain with 3 km horizontal grid spacing and inner-telescoped 642 × 498 horizontal grids domain with 1 km grid spacing. The inner domain includes the location of the ARM SGP site. The WRF-SBM was employed in 24-hours real-time forecasting on 25 April 2011 using online 2-way grid nesting configuration for the two domains (detailed information can be found inauxiliary material).

[6] The Ka-band ARM Zenith Radar (KAZR) observations [e.g.,Clothiaux et al., 2000; Kollias et al., 2007] at the SGP central facility site show that a series of deep convective clouds with the cloud top heights up to 10 km passed over the site from 09 UTC to 11 UTC on 25 April 2011 (Figure 1a). High rainfall rate up to approximately 40 mm hr−1 was observed by the collocated laser optical Particle Size and Velocity (Parsivel) disdrometer during this period (Figure 1b). On the other hand, shallow boundary layer clouds less than the height of 2 km continuously hanged over the site, and weak rainfall approximately 1 mm hr−1 was observed intermittently from 13 UTC to 23 UTC. The atmospheric thermodynamics structure captured by the radiosonde data collected at the SGP site at 08 and 23 UTC (Figures 1c and 1d) indicates several features related to the aforementioned precipitation activity. A capping inversion of temperature was observed under freezing (0°C) level with humid boundary layer. As a result, shallow warm clouds were most likely to be formed under the conditions limiting vertical development of clouds. However, cold deep layer in conditional instability is observed above 0°C level in the free atmosphere. Deep convections thus tended to be easily formed when once strong convection is triggered through the capping inversion.

image

Figure 1. Time series of (a) the Ka-band ARM Zenith Radar (KAZR) vertical reflectivity profile and (b) the rainfall rate measured by the laser optical Parsivel disdrometer over ARM SGP central facility (C1) site on 25 April 2011. Balloon-borne sounding at C1 site at (c) 0823 UTC and (d) 2039 UTC.

Download figure to PowerPoint

3. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Description of Model Simulation and Overall Condition on 25 April 2011 at SGP Site
  5. 3. Results
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

3.1. Surface Rainfall Structure: Comparison Between the Disdrometer Measurements and Simulation

[7] Laser optical Parsivel disdrometers [Löffler-Mang and Joss, 2000] were installed at 16 locations around the ARM SGP central site (36.36°N, 97.29°W) (Figure S1). The maximum distance between these disdrometers was 12.26 km. The disdrometers had a sampling of the maximum length of the falling particle in a horizontal plane and sorted particles into a matrix of 32 size intervals ranging 0 to 26.78 mm and 32 fall velocity intervals ranging from 0.05 to 20.8 m s−1; the contents in the first and second size intervals (D< 0.26 mm) were cut-off because of large noise for small detection. The data was recorded per 10 seconds and integrated to 1 minute. The DSD from Parsivel is determined from raw 32 × 32 size versus fall velocity matrix and is expressed as a function of drop diameter,D (mm):

  • display math

where j is the index of the diameter bin, Δt is observation period in seconds, Areais the cross-sectional area, which is nominally 54 cm2 since length and width of the laser beam are 180 mm and 30 mm, respectively. But it is expressed as a function of D (length × [width − (Dj/2)]) to account partially observed drops that are at the edge of the laser beam. ΔDj is the bin width, Cj is the number of drops at a given size bin and measured fall velocity, v(Dj).

[8] Integral rainfall parameters, i.e., total number concentration, Nt (drops m−3), bulk raindrop effective radius, re (μm), rainfall rate, R (mm hr−1), radar reflectivity, Z (mm6 m−3), are calculated by integrating DSD spectra as follows:

  • display math
  • display math
  • display math
  • display math

Figures 2a and 2c show scatter diagrams of Nt(D> 0.77 mm)-R and re-R correlations derived from the 15 Parsivel measurements (the data of one unit was corrupted) for 24 hours on 25 April 2011. Note that Nt is calculated using parts of DSDs with D > 0.77 mm in consideration of possible underestimation of number concentration at the small size range in Parsivel measurements [Tokay et al., 2001]. The plots are color-coded for the period from 06 to 12 UTC in blue and from 03 to 06 and from 12 to 24 UTC in red; this grouping is based on the difference in approximate time of passages of the deep convective system and shallow boundary layer clouds. The groups of the blue- and red-color dots can be recognized as two distinct modes characterized by largere and R and small re and R, respectively (Figure 2c). The threshold value of re is approximately 400 μm, and R ranges from 0.01 to 10 mm hr−1 and from 0.1 to 100 mm hr−1 in each group. On the other hand, Figure 2a shows that Ntgroup of the blue-color marks branches off with slightly smallerNtfrom the red-color group at approximatelyR = 1 mm hr−1, while both modes has a monotonous increase in Rat log-scale with a similar increase inNt up to 1000 m−3. This behavior is related to difference in re. In the red-color group, small and almost constantre suppresses an increase in R heavier than 10 mm hr−1. In contrast, R increases up to 100 mm hr−1 with an increase in rein the blue-color group. The existence of the two distinct modes for correlation between particle size and rainfall rate can be justified by the 2-dimensional video disdrometer (2DVD) measurements for the same rainfall event (Figure S2b).

image

Figure 2. Scatter diagrams (a, b) between Nt and R, (c, d) between re and Rderived from the disdrometric observational data and from the WRF-SBM simulation result, and (e, f) betweenZ and R, derived from the disdrometric observational data and from the WRF-SBM simulation result. Plots are color-coded by the sampling period, 06–12 UTC (blue) and 03–06, 12–24 UTC (red) for the measurements and 03–05, 13–24 UTC (red) for the simulation. Lines and equations in Figures 2e and 2f show the fitted lines approximating a power function and the approximate equations for each color group, respectively.

Download figure to PowerPoint

[9] In order to obtain the corresponding precipitation DSDs by the disdrometric measurement and subsequently integral rainfall parameters from the WRF-SBM simulation, so-called Parsivel simulator was designed: Momentary 43 binned PSDs of the WRF-SBM simulation were recorded per hour for all model grid points. Then, surface 32-binned precipitation DSD (equation (1)) was calculated using the 43-binned PSDs and terminal fall velocities for mass bins in the SBM. The binned size ranges of the 32 bins were identical to those used in the Parsivel disdrometer program. The contents in the first and second bins (D< 0.26 mm) were cut-off to be aligned with the actual disdrometer sampling. Then, the corresponding integral rainfall parameters were calculated followingequations (1)(5). This approach allows a reasonable straightforward comparison of rainfall parameters between the simulation and measurement, though uncertainty due to the incompatibility of sampling timing and location remains. The integral parameters were sampled over a domain of 20 km square centered at the ARM SGP site. The application of longer distance than the maximum distance between the distrometers is to fill up the difference in number of sampling between the measurements and simulation with different sampling intervals.

[10] Figures 2b and 2d show scatter diagrams of Nt(D> 0.77 mm)-R and re-Rcorrelations derived from the WRF-SBM simulation. The plots are color-coded for the period from 06 to 12 UTC in blue and from 03 to 05 and from 13 to 24 UTC in red (the reason of the different notations between the measurements and simulation is because the momentary output of the simulation was recorded per hour). The simulation reproduced overall two distinct characteristics ofNt-R and re-Rthat correspond reasonably well with those in the observed plots: Branch off of the blue-color marks group from the red-color group is simulated atR > 1 mm hr−1 (Figure 2b). Two distinct modes characterized by large re and R (blue) and small re and R (red) are simulated (Figure 2d). However, the simulated plots pattern are more scattered than that of the measurements. In particular, difference in re between the measurements and simulation is highlighted at R < 1 mm hr−1; the simulated re is scattered to larger value in both color groups. The overprediction of re at the small R range is caused by a group of DSD with small Nt less than 0.5 m−3, seen only in the simulation plot (Figure S3). The marks with overprediction of re can be almost removed by sampling only DSDs with Nt larger than 0.5 m−3 (Figure S4). Although the distrometers may capture DSDs with the tiny Nt, they are disregarded as noise [Tokay et al., 2005]. The set of large re, small Nt and R is simulated at grids probably where rainfall begins to fall and only large raindrops reach to the surface in advance. Spontaneous breakup of large drops is not considered in the present SBM, so that huge raindrops after melting of huge ice particles like hailstones may artificially reach to the ground without breakup and be sampled in the simulation.

[11] The relationship between Z and R is generally approximated to be a form of a power law function, Z = ARb. The values of the coefficients are recognized as an important aspect of the rainfall characteristics, though the values have non-negligible dependency on the types of measurement instruments and the regression techniques [Campos and Zawadzki, 2000; Tokay et al., 2001]. Figures 2e and 2f illustrate the Z-R relationships in the form of scatter diagrams derived from the measurements and simulation, respectively. The parameters, A and b, are computed using a linear regression method separately for the plot groups of red-color and blue-color, shown in the form ofZ = ARb in the panels. The values of A and b of the simulation are generally good agreement with those of the observation, except for Afor the blue-colored dots. This difference is considered to be due to the same source as in the aforementioned difference inre at the range of R < 1 mm hr−1.

[12] The timing of appearances of convective and stratiform clouds around the SGP site in the simulation is justified by the overall agreement between the observation and simulation panels in Figure 2, though the present grouping according to the time ranges may be coarse. Shallow boundary clouds with rainfall, however, seem to be oversimulated and included in the 06–12 UTC group as compared to the measurements, because a non-neglectable part of blue-color marks are overlapped by the red-color marks in the simulation panels ofFigure 2.

3.2. Relationship Between the Surface Rainfall Structures and the Cloud/Weather Conditions

[13] These distinct rainfall characteristics originated from difference in processes between the deep convective and shallow boundary layer clouds to form the precipitation. Figure 3 illustrates the horizontal distributions of re(without cutoff at particular diameter for surface DSD) and cloud-top temperature, derived from the WRF-SBM simulation at 09 and 21 UTC; the cloud-top temperature is here defined as air temperature at the highest model level containing the total water content larger than 0.3 g m−3 in each grid column. This figure clearly demonstrates that reand the cloud-top temperature are inversely correlated. In particular, a boundary around 400μm of re in Figure 3ais in good agreement with a border of 0° Celsius of the cloud-top temperature inFigure 3b. These results show that the surface precipitation of red- and blue-color groups inFigure 2mostly originates from shallow warm and deep mixed-phase rain processes, respectively.

image

Figure 3. Snapshots of horizontal distributions of (a) surface precipitation re(without diameter cutoff) and (b) cloud-top temperature of the WRF-SBM simulation at (left) 09 and (right) 21 UTC on 25 April 2011.

Download figure to PowerPoint

[14] At 09 UTC (Figure 3, left), couples of deep convective system are simulated over the ARM SGP site. The coldest cloud-top temperature is approximately −60°C, and the corresponding surface DSDre is more than 900 μm (up to 2000 μm as shown in Figure 2d). Shallow warm clouds with the cloud-top temperature over 0°C cover the wide area of southwestern segment and move toward the ARM SGP site by a southwestern wind. There of the shallow system is approximately 300 μm. At 21 UTC (right panels), a number of patchy shallow clouds remain over area including the location of ARM SGP site. The re is approximately 300–400 μm. The simulated cloud top heights over the site at 09 and 21 UTC are in reasonably agreement with those observed by the KAZR (Figure 1a). The WRF-SBM simulation poorly reproduced the homogenous distribution of shallow stratiform clouds at 21 UTC, which is speculated by the KAZR profile around the time. An accurate prediction of cloud fraction is a difficult problem even in a high-resolutional simulation [e.g.,Khairoutdinov and Randall, 2003], and this problem needs to be investigated more in future works.

[15] In the case on 25 April, 0°C temperature level were located at a height of approximately 3.5 km, so that warm clouds below this level were shallow and raindrops reached to the ground from clouds immediately. Terminal fall velocity with droplet radius of 400 μm is assumed to be approximately 2.6 m s−1 at 1000 hPa level in the model microphysics [Khain and Sednev, 1995]. Against this fall velocity, maximum updraft velocity in the simulated shallow warm clouds is roughly 2 m s−1. Thus, particles that grow larger up to the radius of approximately 400 μm cannot be held in the clouds and fall to the ground. The small updraft velocity in the shallow clouds is a factor of small surface re. On the other hand, 10 m s−1or much stronger updraft velocities in the deep convective system were simulated, so that any size particles in the model can be held in the updraft region of the clouds. Thus, much larger particles than those from the shallow warm clouds can be generated and finally fall to the ground. The surface rainfall rate increases with an increase in surface-rainfallre.

[16] Figure S5a illustrates the simulated thermodynamic fields that cause the formation of series of deep convective clouds at 09 UTC, which corresponds to those observed by the KAZR (Figure 1a). There are two major flows of warm, moist air from the southern part and cold, dry air from the northern part on 850 hPa level. Parts of the two major flows are converged around 36°N, 98°W. This is approximately the center of the deep convective system. A couple of warm cores are formed around the coordinates on the 850 hPa pressure level. Warm air mass is transported to the level of free atmosphere in conditional instability, and induces deep convections. On a north-south cross-section of the east-side of the system along 97.5°W (Figure S5b), a strong vertical wind shear is simulated under 800 hPa level. A low-level flow from south rises to a higher level at 35°N by the opposite wind, and then air mass diverges at the top of deep convection, approximately on 200 hPa level. Deep convections are continuously induced along the outflow boundaries originating from the convection close behind. The deep convective system is conveyed northeastward by the background southwestern wind.

[17] On the other hand, shallow boundary layer clouds are simulated over the southwestern segment at 09 UTC. Nocturnal boundary layer clouds are often observed over this area, and the detailed mechanism was investigated in Zhu et al. [2001]. Turbulence driven by wind shear is an important factor to maintain well-mixed boundary layer and form clouds, in place of radiative heating in daytime.Figure S5c shows that shallow clouds are simulated along remarkable wind shear near the ground. Relative humidity reaches 100% and condensation occurs under or around capping inversion of temperature (Figure S6). Unlike the segment shown in Figure S5b, deep convective system are not developed, probably because of no convergence promoting breakup of the capping and subsequently involvement in free atmosphere with conditional instability. The boundary layer structure shown in Figure S6 is similar to those observed at ARM SGP site before and after the passage of the deep convective system (Figures 1c and 1d). These results imply that the observed soundings at the times are under a sufficient condition to form shallow boundary layer clouds.

4. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Description of Model Simulation and Overall Condition on 25 April 2011 at SGP Site
  5. 3. Results
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[18] We have analyzed the Parsivel disdrometer observations at the ARM SGP site on 25 April, 2011. The measurement results show unique distinctive structures of correlations between the integral rainfall parameters, which can be classified into two groups mostly according to the types of clouds causing the precipitation. The group with large bulk effective particle size and rainfall rate originates from series of deep convective clouds, whereas the inverse is true for the group from shallow boundary layer clouds. The WRF-SBM weather prediction simulation could successfully reproduce the rainfall microphysical structures. The simulation for high rainfall rate is well, whereas the low rate rainfall is simulated relatively with defects; note that the measurements also have a large uncertainty at low rainfall rate, especiallyR< 0.1 mm/hr. The discussion on the surface rainfall microphysics was expanded to relevance to the weather and cloud conditions on the day, by referring to the three-dimensional atmospheric field in the simulation. The simulation demonstrated that the horizontal distribution of the classified precipitation was well inversely correlated with that of cloud-top temperature; that is, the distinct rainfall structures are derived from the difference between the deep mixed-phase and shallow warm rain processes in clouds aloft. The characteristic structures of cloud and precipitation were most likely to be formed under the weather condition on the day. Series of deep convective clouds were locally and rapidly developed through meso-scale feedback process. Shallow boundary layer clouds were easily formed under the atmospheric sounding with humid boundary layer under temperature inversion.

[19] This paper has presented a possibility of the high-resolution NWP simulation coupled with one of the up-to-date microphysical schemes, even though this is a single case study for one-day precipitation event. This rainfall event is set as an important case in the MC3E field campaign during April and May 2011. The three-dimensional output including hydrometeor PSDs of the WRF-SBM is planned to be provided to the MC3E and GPM science communities to support their research, e.g., for testing/developing the GPM pre-launch precipitation retrieval algorithm.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Description of Model Simulation and Overall Condition on 25 April 2011 at SGP Site
  5. 3. Results
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[20] This study was supported by the NASA Precipitation Measuring Mission (PMM) and NASA Modeling Analysis Prediction (MAP). We acknowledge the collaboration with the U.S. Department of Energy (DOE) as part of the ARM Climate Research Facility SGP site, as a part of the joint NASA-DOE MC3E field campaign.

[21] The Editor thanks the two anonymous reviewers for their assistance in evaluating this paper.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Description of Model Simulation and Overall Condition on 25 April 2011 at SGP Site
  5. 3. Results
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Description of Model Simulation and Overall Condition on 25 April 2011 at SGP Site
  5. 3. Results
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

Auxiliary material for this article contains 15 figures, 1 text file, and 2 tables.

Auxiliary material files may require downloading to a local drive depending on platform, browser, configuration, and size. To open auxiliary materials in a browser, click on the label. To download, Right-click and select “Save Target As…” (PC) or CTRL-click and select “Download Link to Disk” (Mac).

Additional file information is provided in the readme.txt.

FilenameFormatSizeDescription
grl29811-sup-0001-readme.txtplain text document6Kreadme.txt
grl29811-sup-0002-fs01.epsPS document1007KFigure S1. The locations of the Laser optical Parsivel disdrometers installed around the ARM SGP central site.
grl29811-sup-0003-fs02.epsPS document1645KFigure S2. Scatter diagrams between Nt and R, between re and R, and between Z and R derived from the 2-dimensional video disdrometer measurements.
grl29811-sup-0004-fs03.epsPS document1748KFigure S3. The same as in Figures 2a and 2b, but using a vertical axis covering with much lower values.
grl29811-sup-0005-fs04.epsPS document704KFigure S4. The same as in Figure 2d, but sampling only DSDs with Nt larger than 0.5 m-3.
grl29811-sup-0006-fs05.epsPS document3153KFigure S5. Snapshot of the WRF-SBM simulation at 09 UTC on 25 April 2011.
grl29811-sup-0007-fs06.epsPS document1197KFigure S6. Vertical profiles of temperature and relative humidity with respect to water at 36N, 99.5W and 35N, 99.5W at 0900 UTC 25 April 2011, obtained from the WRF-SBM simulation.
grl29811-sup-0008-fs07.epsPS document830KFigure S7. Nesting domain configuration used for the WRF-SBM simulation.
grl29811-sup-0009-txts01.txtplain text document3KText S1. Detailed description of the WRF-SBM model and simulation.
grl29811-sup-0010-fs08.epsPS document604KFigure S8. Droplet radii against 43 bins in the WRF-SBM.
grl29811-sup-0011-ts01.txtplain text document0KTable S1. The coordinates of the Laser optical Parsivel disdrometers installed at 16 locations around the ARM SGP central site.
grl29811-sup-0012-ts02.txtplain text document2KTable S2. The intervals between the Laser optical Parsivel disdrometers.
grl29811-sup-0013-fs09.epsPS document954KFigure S9. Thirty-two size intervals used in the Parsivel disdrometer measurements.
grl29811-sup-0014-fs10.epsPS document263KFigure S10. Thirty-two size intervals used in the Parsivel disdrometer measurements, continuation of Figure S9.
grl29811-sup-0015-fs11.epsPS document4306KFigure S11. The same as in Figure 2, but using re and R with parts of DSDs with D > 0.77 mm in place of D > 0.26 mm.
grl29811-sup-0016-fs12.epsPS document23751KFigure S12. GOES IR satellite images at 0845, 0945 and 2115 UTC April 25 2011.
grl29811-sup-0017-fs13.epsPS document1079KFigure S13. Time series of the Ka-band vertical reflectivity profile and the Parsivel rainfall rate calculated from the WRF-SBM simulation output per hour, at 36.36N, 97.29W on 25 April 2011.
grl29811-sup-0018-fs14.epsPS document1436KFigure S14. Time series of vertical profiles of water droplets effective radius, snow aggregates effective radius, graupel effective radius, and hail effective radius calculated from 43 bins PSDs of the WRF-SBM simulation output per hour, at 36.36N, 97.29W on 25 April 2011.
grl29811-sup-0019-fs15.epsPS document1133KFigure S15. Time series of vertical profiles of liquid water contents and ice water contents calculated from the WRF-SBM simulation output per hour, at 36.36N, 97.29W on 25 April 2011.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.