1. Introduction
The effect of ice microphysics on sedimentation properties has been studied in bulk microphysical parameterizations with spherical ice crystals (e.g., Hong et al. 2004). However, some bulk microphysics models treat ice crystal habit, or shape, in an empirical sense, with it being a prescribed function of size and temperature (Phillips et al. 2017) while some bin microphysics models predict the evolution of crystal shape explicitly (Phillips et al. 2018). Multimoment bulk microphysical parameterizations have been developed that allow for the nonspherical growth of ice crystals (Harrington et al. 2013a,b; Chen and Tsai 2016). Through theoretical and applied studies, Sulia and Harrington (2011) and Sulia et al. (2013) determined that the habit of ice crystals is of utmost importance to track during periods of depositional ice growth. Habit evolution can have a substantial influence on cloud phase partitioning, such as ice and liquid mass, which can thereby influence hydrometeor sedimentation rates and other cloud properties. Nucleation-controlled variations in pristine ice number concentration (Ni) and subsequent depositional growth can impact ice mass and indirectly affect the degree of exaggeration of the crystal habit. Various parameterizations exist to represent ice nucleation; the choice of parameterization affects the number concentration of ice crystals in a volume due to the differing dependencies within the parameterization equations, with potential influences on sedimentation rates and cloud glaciation times, among other microphysical factors. Highly active cold-season storms with abundant moisture sources, namely lake-effect storms (LESs), offer a ripe opportunity wherein such interactive processes and their associated effects can be investigated.
Lake-effect (LE) snow impacts the Great Lakes region (GLR) during the fall and winter months when either an Arctic or continental polar air mass traverses over the relatively warm lakes (Braham 1983; Hjelmfelt 1990; Wright et al. 2013). The air mass experiences localized supersaturation, and lapse rates steepen via low-level moistening and warming. Conditional instability is locally enhanced, favoring convection and leading to the formation of cumuliform clouds (Kristovich and Laird 1998). These clouds can organize into widespread cells or an organized cloud band depending on wind magnitude and direction relative to the long lake axis, presence of a secondary circulation due to thermal differences between the lake and land, and the influence of differential surface roughness on low-level convergence (Bergmaier et al. 2017). Note that the ascent associated with the secondary circulation may also assist with the triggering of convection and increase the lifespan of convective processes (Bergmaier and Geerts 2016; Bergmaier et al. 2017). When these clouds reach the downwind side of a lake, any present topography can contribute to further forcing for ascent, intensifying the storm (Ahrens 2013). Climatologically, a greater frequency of LE clouds tends to exist during December, January, and February when the air–lake temperature difference is typically maximized (Laird et al. 2017). From October to March, LE clouds are present about 60%–80% of days each month over the GLR (Laird et al. 2017).
While the mesoscale dynamics of LE clouds are foundationally understood, the effect of microphysical ice processes in numerical models is less so. As such, the purpose of this paper is to diagnose the potential effects of ice crystal nucleation and mode of growth on in-cloud LE microphysical processes and forecast precipitation magnitude, type, and spatial pattern. Harnessing a deeper understanding of why certain microphysical processes lead to varying numerical solutions is necessary when aiming to understand the discrepancies between model output and observations, and even among varying model solutions.
2. Model description
a. Adaptive habit microphysics model
Ice crystals can grow to relatively large sizes at the expense of liquid droplets in mixed-phase clouds via the Wegener–Bergeron–Findeisen process (Sulia and Harrington 2011). Due to enhanced depositional growth for nonspherical crystals, assuming spherical growth at all temperatures underestimates the final crystal mass, affecting ice and liquid water contents, and subsequent precipitation processes. The adaptive habit microphysics model (AHM) combines vapor diffusional mass growth from the classical capacitance model (Pruppacher and Klett 1997) and aspect ratio (ϕ) evolution based on the inherent growth ratio (Γ) of the crystal, which varies with temperature and describes the distribution of mass between the major and minor crystal axes. Harrington et al. (2013a) developed the AHM with four prognostic ice variables: mass, number, and spheroidal volume-weighted a- and c-axis-length mixing ratios to predict ice crystal properties. Ice crystal shape is characterized as spheroidal by relating its a and c axes as ϕ = c/a where an oblate spheroid represents a platelike habit (i.e., a > c thus ϕ < 1), and a prolate spheroid represents a columnar habit (i.e., c > a thus ϕ > 1). By means of tracking ϕ, the AHM predicts the evolution of ice crystal habit through a historical tracking parameter, which takes account of ϕ evolution through a temporal average of Γ, capturing the nonlinear growth via vapor diffusion and subsequent effect on phase partitioning (Sulia et al. 2013). This model has been tested in a parcel model framework (Harrington et al. 2013b), two-dimensional kinematic model (Sulia et al. 2013; Jensen et al. 2017), and both ideal and real Weather Research and Forecasting (WRF) Model simulations (Sulia et al. 2014; Sulia and Kumjian 2017a,b; Jensen et al. 2018).
The accurate prediction of precipitation from high-impact snowfall events depends on the parameterization of thermodynamic and microphysical processes, including the formation and subsequent growth of frozen hydrometeors. While synoptic and mesoscale dynamics provide the environmental conditions and ascent needed for mixed-phase cloud formation, surface precipitation quantities are dependent upon the liquid and ice water contents and hence the growth mechanisms of ice crystals, including vapor deposition and subsequent collection. Before crystals can undergo growth or decay, they must first nucleate.
b. Ice nucleation parameterizations
The formation of ice crystals has been modeled using various nucleation parameterizations representing different nucleation modes with varying dependencies (e.g., temperature, ice supersaturation, aerosol concentration), as a result of in situ measurements and laboratory studies. Despite numerous field campaigns and laboratory studies focused on detailing nucleation processes (e.g., DeMott 1990; Meyers et al. 1992; Phillips et al. 2008; DeMott et al. 2010; Ardon-Dryer 2012; Murray et al. 2011; DeMott et al. 2015; Hiron and Flossman 2015; Solomon et al. 2015), albeit with varying assumptions or goals of quantifying different aerosol inputs, the prediction of ice nucleating particles (INPs) remains challenging. Uncertainty is reflected in the varying forms of ice nucleation parameterizations modeling different or even combined nucleation modes (e.g., deposition, contact freezing, immersion freezing) aimed at representing the formation of ice crystals.
Pruppacher and Klett (1997) noted that certain ice nucleation modes may produce greater INP concentrations than others over a range of subfreezing temperatures, suggesting that contact freezing nucleation is the dominant mode, followed by deposition and immersion freezing. However, more recent research has demonstrated that contact freezing is the least important mode of heterogeneous ice nucleation (Phillips et al. 2007), even when considering the direction of freezing (e.g., inside out vs outside in; Phillips et al. 2017). Immersion freezing has been suggested to be of greatest importance within mixed-phase clouds while deposition nucleation is expected to have a secondary contribution to Ni (Kanji et al. 2017). Condensation freezing has historically been treated as a combined nucleation mode (e.g., immersion–condensation freezing; Kanji et al. 2017). As such, this paper focuses on condensation and immersion freezing nucleation as parameterized by DeMott et al. (2015) and condensation and deposition freezing by Meyers et al. (1992).
1) Meyers et al. (1992)
2) DeMott et al. (2015)
c. Advanced particle microphysics model
Levin et al. (2005) suggest that precipitation amounts and rates decrease as aerosol pollution increases within a cloud due to the aerosol indirect effect on cloud microphysics and precipitation. Ardon-Dryer (2012) showed that the onset of precipitation may be delayed with an increased amount of environmental aerosols. These and many other studies support the need for high-resolution aerosol data in numerical models due to their direct and indirect effects on cloud properties. The advanced particle microphysics (APM) model was developed by Yu and Luo (2009) to explain observations of size-resolved atmospheric particles. The processes of nucleation, condensation and evaporation, coagulation, local thermodynamic equilibrium, and dry deposition are considered within the APM, which can be used as an independent box model or coupled with other chemistry-focused models. It was first implemented into a global three-dimensional atmospheric composition model, Goddard Earth Observing System (GEOS-Chem; Yu and Luo 2009), and later integrated with the chemistry-coupled WRF Model (WRF-Chem; Luo and Yu 2011). The APM is optimized to simulate secondary particle formation and subsequent growth to sizes typical of cloud condensation nuclei (CCN), with increased size resolution for critical aerosol size ranges. Here, secondary particles refer to those formed from new particle formation (or particle nucleation) in the atmosphere, while primary particles include dust, sea salt, black carbon, and primary organic carbon. The APM uses 40 bins to represent secondary particles in the dry size range of 1.2 nm to 12 μm. While freshly nucleated particles are only a few nanometers, growth to CCN sizes can occur. The APM also allows the calculation of CCN as a function of liquid supersaturation.
3. Methodology
a. Implementation into the adaptive habit microphysics model
Both parameterizations developed by Meyers et al. (1992) and DeMott et al. (2015) were implemented into the AHM to test the subsequent sensitivity of sedimentation rates and accumulation amounts during the Ontario Winter Lake-effect Systems (OWLeS) intensive observing period 4 (IOP4). Unlike in previous studies using the AHM where ice nucleation was dependent upon the predetermined ice concentration and existing ice in a grid cell, these parameterizations developed from empirical data allow for a spatially and temporally evolving Ni. A nucleation rate is calculated in each grid cell via Eq. (1) or Eq. (2), with extrapolation used, if −35° < T < −5°C and the ice supersaturation is ≥5%. Neither the Meyers et al. (1992) or DeMott et al. (2015) parameterizations consider the present ice number concentration before diagnosing the Ni or INP, respectively, at the current time step. This could lead to potential overprediction of ice number and mass within a cloud system. However, the DeMott et al. (2015) rate is applied only if there is enough cloud droplet mass to sustain immersion freezing processes. Therefore, no immersion freezing will ensue in unphysical conditions, where no droplets exist for INPs to be immersed within and subsequently induce freezing. Also, while secondary ice processes are represented within the AHM, even though the quantification of such processes is challenging due to limited understanding (Field et al. 2017), the associated effects are not investigated within this paper.
For this paper, the APM was run independently of the AHM in the WRF-Chem, version 3.7.1, using the same initialization time, simulation time period, and namelist options discussed in section 3b. The APM-produced INP and CCN aerosol data were then updated within the AHM every 3 h. While the background aerosol profile is reset every 3 h, the changes from one time step to the next are negligible, so nudging is not performed on these data. Dust particles have a negligible contribution to CCN but are important for heterogeneous ice nucleation and so are considered the dominating environmental aerosol for ice nucleation. Due to unclear physics of ice nucleation of carbonarious aerosols, only the contribution of dust to ice nucleation is considered; this dust is not specifically defined as mineral dust by the APM. Although this is not the explicit requirement for the DeMott et al. (2015) parameterization, the use of dust data in specific is adequate for the purposes of this paper relative to using all hydrophobic aerosol with no regard to composition. Dry dust with sizes larger than 0.5 μm act as initial input (na) into the DeMott et al. (2015) parameterization and are assumed to be an upper limit for Ni calculations by Meyers et al. (1992). Only 20% of the APM dust data are used in the AHM with this size threshold specified by DeMott et al. (2015). For a given supersaturation with respect to liquid, the CCN number provided by the APM is used to calculate cloud droplet number concentrations in the model. APM aerosol were treated as background in the AHM, meaning that no amount of aerosol was removed during CCN or INP activation or scavenging processes, or added during complete droplet evaporation or particle sublimation therefore eliminating consideration of any aerosol recycling processes.
To directly compare Meyers et al. (1992) to DeMott et al. (2015) at varying temperatures, the saturation vapor pressure with respect to liquid and ice is calculated for a range of temperatures. Assuming liquid water saturation, the ice supersaturation is calculated and used in the Meyers et al. (1992) deposition nucleation equation. In Fig. 1, the DeMott et al. (2015) parameterization is evaluated with aerosol concentration for particles larger than 0.5 μm for 0.01 and 0.1 cm−3. The predicted Ni for each aerosol case stay consistently different for DeMott et al. (2015) with decreasing temperature. As the temperature decreases, there is an exponential increase in INP concentration, which is a typical trend and not exclusive to Meyers et al. (1992) and DeMott et al. (2015) (Cziczo et al. 2017). An Ni difference of about six orders of magnitude exists between these parameterizations at warmer temperatures with this difference decreasing as temperatures decrease. From these results, it is expected that Meyers et al. (1992) nucleates a greater Ni than DeMott et al. (2015) at all considered temperatures with INP concentrations representative of this LES event, discussed in section 4c.
Meyers et al. (1992) predicted ice parameterization for deposition–condensation nucleation compared to the DeMott et al. (2015) parameterization for ice nucleating particle number prediction by means of immersion–condensation freezing (L−1). Two concentrations of dust >0.5 μm at standard temperature and pressure are considered for DeMott et al. (2015): 0.01 (orange) and 0.1 cm−3 (blue).
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Meyers et al. (1992) predicted ice parameterization for deposition–condensation nucleation compared to the DeMott et al. (2015) parameterization for ice nucleating particle number prediction by means of immersion–condensation freezing (L−1). Two concentrations of dust >0.5 μm at standard temperature and pressure are considered for DeMott et al. (2015): 0.01 (orange) and 0.1 cm−3 (blue).
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Meyers et al. (1992) predicted ice parameterization for deposition–condensation nucleation compared to the DeMott et al. (2015) parameterization for ice nucleating particle number prediction by means of immersion–condensation freezing (L−1). Two concentrations of dust >0.5 μm at standard temperature and pressure are considered for DeMott et al. (2015): 0.01 (orange) and 0.1 cm−3 (blue).
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
b. Experimental setup
The LES of interest occurred during IOP4 of the OWLeS field project (Kristovich et al. 2017), details of which are discussed in section 4. Numerical simulations of this LES and its associated microphysical characteristics were completed using version 3.7.1 of the WRF Model. The simulation was run with three two-way nested domains centered at 43.605°N, 76.721°W varying in horizontal resolution from 25 km [Domain 1 (D01)], 5 km [Domain 2 (D02)], and 1 km [Domain 3 (D03); Fig. 2], and 30 nonlinear vertical levels, extending to about 15 km. The time step was 150 s in D01, 30 s in D02, and 6 s in D03. D03 focused on eastern Lake Ontario and its downwind regions; previous sensitivity tests determined negligible precipitation pattern differences when shifting this domain to include the entirety of Lake Ontario. Initial and boundary conditions were provided by the 12-km North American Mesoscale Forecast System (NAM) Analysis; offline tests indicated increased forecast proficiency of NAM IOP4 LES snowfall location relative to the 0.5° Global Forecast System Analysis data.
Nested domains for all WRF simulations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Nested domains for all WRF simulations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Nested domains for all WRF simulations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Each simulation was run for the period of 11–20 December 2013, allowing 4 days of model spinup to capture a synoptic event that preceded the LES that occurred from 1800 UTC 15 December to 0800 UTC 16 December. Ample spinup time allowed for accurate representation of this synoptic event, triggering the instability leading to the LES formation, while decreasing the spinup time resulted in a poor spatial LES precipitation forecast. The Rapid Radiative Transfer Model adapted for global climate models (RRTMG; Iacono et al. 2008) was used to determine longwave radiative fluxes. The Dudhia scheme was chosen to calculate shortwave radiative processes (Dudhia 1989). Boundary layer physics were determined by the Yonsei University scheme (YSU; Hong and Noh 2006). The Kain–Fritsch cumulus scheme was used in D01 and D02 (Kain 2004) but turned off in D03 due to its high convection-allowing resolution. The aforementioned AHM was used as the microphysics option, within which ice nucleation varied between the Meyers et al. (1992) and DeMott et al. (2015) parameterizations, described in section 2b. The four simulations of interest and their associated modeling options diverging from the namelist options discussed are summarized in Table 1. The labels MEY92H, MEY92S, DEM15H, and DEM15S are used to refer to specific simulations and their model specifications. MEY92 and DEM15 refer to the parameterization used for ice nucleation within the AHM, Meyers et al. (1992) and DeMott et al. (2015), respectively. The S and H subscripts on these labels refer to ice growth occurring spherically and nonspherically, respectively.
All model simulations discussed in this paper.
c. Quantitative precipitation estimates
Radar-derived precipitation data are the best option for model validation in this area of very limited in situ ground observations. As radar is a remote sensing tool, its data quality issues are inherently problematic. These errors can be exacerbated by environmental conditions and mechanical specifications. The most obvious potential error source is the reflectivity-to-rainfall (Z–R) relationship developed by Marshall et al. (1947), Z = ARb, where A and b are constants that change substantially depending on the hydrometeor class and associated size distribution (Stout and Mueller 1968). Campbell et al. (2016) noted that during an earlier OWLeS IOP, setting A = 75, the constant more regularly used for snowfall in the western United States, and b = 2 allowed for better liquid-equivalent forecasts when compared to the measurements at the stations in Sandy Creek and North Redfield, New York. As such, these constants are used in the derivations presented herein. This Z–R relationship is less reliable for a mixture of hydrometeor types or when a transition is occurring from snow to rain, for example. Of course, other problems may arise from radar, such as location and elevation, distance of the sampling data, and coverage.
The 24-h quantitative precipitation estimate (QPE) valid at 1200 UTC 16 December from the National Weather Service Advanced Hydrologic Prediction Service (AHPS) was used to evaluate modeled quantitative precipitation forecasts (QPFs). For clarity, QPE and QPF are not interchangeable and refer to precipitation observations and forecasts, respectively. To build this dataset, precipitation estimates from the next-generation Weather Surveillance Radar-1988 Doppler (WSR-88D) were compared to reports of precipitation from rain gauges by the AHPS. Based on this comparison, a bias was computed and applied to the radar data. The radar and gauge precipitation data were then combined into the QPE field and monitored every hour. This AHPS QPE product has a spatial resolution of 16 km2 and temporal resolution of 24 h, where a hydrologic day runs from 1200 to 1200 UTC. Note that communication with staff at the AHPS has identified the lack of documentation on A and b used during a specific event. As a supplement to the AHPS observation, accumulated precipitation was derived herein from the 0.5° plan position indicator (PPI) scans (fixed-elevation azimuthal scans of radar) provided about every 5 min by the KTYX radar (Montague, New York, located east of Lake Ontario) using the aforementioned Z–R relationship where A = 75 and b = 2 and interpolated to a 16 km2 grid. These results are discussed in section 4d.
Sulia and Kumjian (2017a,b) show that an offline forward operator can produce modeled polarimetric radar quantities (Ryzhkov et al. 2016) from AHM model output. To better compare observed and modeled precipitation, QPFs are calculated using the Z–R relationship described above. The derived radar data from the forward operator are used to calculate the accumulated precipitation using the same Z–R relationship as was used for the KTYX ZH data. While a loss of accuracy is expected with this two-step derivation (i.e., model output to radar reflectivity to precipitation), the results of this test discussed in section 4d provide insight into how the accumulation amounts from the derived radar data differ from the explicit modeled precipitation quantities.
As with observations, the daily precipitation accumulation maps for the model simulations presented in section 4 were produced for 1200 UTC 15 December to 1200 UTC 16 December. The accumulated precipitation was calculated in the AHM for MEY92 and DEM15 as a sum of rain and liquid-equivalent snow, ice, and graupel that precipitated to the surface; this is referred to as the explicit modeled precipitation. Using these forecasts instead of snowfall depth removes the uncertainty regarding the highly variable snow-to-liquid ratios in this region (Baxter et al. 2005) and allows for direct comparison to AHPS measurements and the KTYX-derived QPE. In addition to the daily QPEs, hourly snow–water liquid-equivalent measurements were recorded at two stations east of Lake Ontario during the OWLeS field campaign operating between 5 December 2013 and 29 January 2014. The North Redfield station [43.624 45°N, 75.877 08°W; elevation of 385 m above mean sea level (MSL)] and the Sandy Creek station (43.6402°N, 76.097 15°W; elevation of 143 m MSL) measurements were used in analyses for both daily precipitation accumulation maps and a time series of precipitation.
4. Case study
OWLeS IOP4 will be introduced in both the synoptic and mesoscale senses through NAM analysis and a combination of in situ and remote observations. The dust conditions as modeled by the APM are introduced to ground further discussion of the effects on ice crystal nucleation and subsequent processes in the AHM. Then, the WRF simulations outlined in Table 1 will be analyzed to further elucidate the microphysical structure of the storm as well as the impacts that ice nucleation and subsequent growth mode have on the QPF.
a. Synoptic conditions
As confirmed by the sensitivity to model spinup, the OWLeS IOP4 case focused on a locally confined, shallow cumuliform LE cloud that responded to the latent conditions of the preceding synoptic circulation. The 12-km NAM analysis shows that a trough at 500-hPa propagated eastward from 1800 UTC 15 December to 0600 UTC 16 December 2013, bringing with it increased absolute vorticity over Lake Ontario (Fig. 3). The circulation associated with a low pressure system off the coast of Maine at 1800 UTC (Fig. 3) ushered in subfreezing temperatures at both the surface and 850 hPa into the eastern GLR (Fig. 4). As a result, there was a 16°–20°C temperature difference between the lake surface and 850 hPa, satisfying the 13°C minimum temperature difference needed for LES to develop (approximately the dry adiabatic lapse rate; Braham 1983), which allowed for the destabilization of the lower atmosphere and moisture and heat fluxes to propagate upward. Weak westerly surface and 850-hPa winds (Fig. 4) over the relatively warm Lake Ontario (i.e., 4.35°C; NOAA 2013) allowed the overlying air mass to travel over the longest fetch of the east–west-oriented lake, maximizing the amount of potential heat and moisture fluxes.
NAM analysis of (left) mean sea level pressure (hPa) at intervals of 2 hPa and (right) 500-hPa geopotential heights (m) and absolute vorticity (s−1) at (top) 1800 UTC 15 Dec, (middle) 0000 UTC 16 Dec, and (bottom) 0600 UTC 16 Dec 2013.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
NAM analysis of (left) mean sea level pressure (hPa) at intervals of 2 hPa and (right) 500-hPa geopotential heights (m) and absolute vorticity (s−1) at (top) 1800 UTC 15 Dec, (middle) 0000 UTC 16 Dec, and (bottom) 0600 UTC 16 Dec 2013.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
NAM analysis of (left) mean sea level pressure (hPa) at intervals of 2 hPa and (right) 500-hPa geopotential heights (m) and absolute vorticity (s−1) at (top) 1800 UTC 15 Dec, (middle) 0000 UTC 16 Dec, and (bottom) 0600 UTC 16 Dec 2013.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
NAM analysis of temperature (°C) and wind (m s−1) at the (left) surface and (right) 850-hPa at (top) 1800 UTC 15 Dec, (middle) 0000 UTC 16 Dec, and (bottom) 0600 UTC 16 Dec 2013.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
NAM analysis of temperature (°C) and wind (m s−1) at the (left) surface and (right) 850-hPa at (top) 1800 UTC 15 Dec, (middle) 0000 UTC 16 Dec, and (bottom) 0600 UTC 16 Dec 2013.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
NAM analysis of temperature (°C) and wind (m s−1) at the (left) surface and (right) 850-hPa at (top) 1800 UTC 15 Dec, (middle) 0000 UTC 16 Dec, and (bottom) 0600 UTC 16 Dec 2013.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
b. Mesoscale conditions and observations
Multiple radiosondes were launched from three locations near Lake Ontario during IOP4 by mobile sounding teams led by Millersville University, SUNY Oswego, and Hobart and William Smith (HWS) Colleges (Fig. 5). All soundings indicated that the atmosphere was nearly to completely saturated from the surface to approximately 700 hPa, where the capping inversion inhibited any additional vertical growth of the LES. The winds were primarily westerly at all sampled atmospheric levels indicating little to no directional shear. Surface winds ranged from 5 to 15 m s−1, validating the aforementioned NAM analysis winds (Fig. 4) with a slight increase in magnitude. From these soundings, cloud-base and cloud-top temperatures are estimated to be around −8° and −27°C, respectively. As seen in PPI scans of reflectivity ZH by the KTYX radar, the LES was well organized from 0000 to 0700 UTC (Figs. 6a–h), with distinctly strong banding features from 0400 to 0700 UTC (Figs. 6e–h). During its greatest intensity, the maximum ZH was > 35 dBZ. The system intensity decreased as the LES moved southward around 0600 UTC (Fig. 6g), rapidly decreased in strength at 0800 UTC (Fig. 6i), and lost all coherent structure by 0900 UTC (Fig. 6j).
Skew T–logp diagrams [temperature (red), dewpoint temperature (green), and wind barbs (full and half barbs represent 5 and 2.5 m s−1, respectively)] at (bottom right) three locations marked on the map by the following mobile observing teams: (top) Millersville University at 2055 UTC 15 Dec and 0215 UTC 16 Dec, (middle) SUNY Oswego at 2315 UTC 15 Dec and 0215 UTC 16 Dec, and (bottom left) Hobart and William Smith Colleges (HWS) at 2316 UTC 15 Dec.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Skew T–logp diagrams [temperature (red), dewpoint temperature (green), and wind barbs (full and half barbs represent 5 and 2.5 m s−1, respectively)] at (bottom right) three locations marked on the map by the following mobile observing teams: (top) Millersville University at 2055 UTC 15 Dec and 0215 UTC 16 Dec, (middle) SUNY Oswego at 2315 UTC 15 Dec and 0215 UTC 16 Dec, and (bottom left) Hobart and William Smith Colleges (HWS) at 2316 UTC 15 Dec.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Skew T–logp diagrams [temperature (red), dewpoint temperature (green), and wind barbs (full and half barbs represent 5 and 2.5 m s−1, respectively)] at (bottom right) three locations marked on the map by the following mobile observing teams: (top) Millersville University at 2055 UTC 15 Dec and 0215 UTC 16 Dec, (middle) SUNY Oswego at 2315 UTC 15 Dec and 0215 UTC 16 Dec, and (bottom left) Hobart and William Smith Colleges (HWS) at 2316 UTC 15 Dec.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
NEXRAD (KTYX) 0.5° reflectivity at times closest to (a) 0000, (b) 0100, (c) 0200, (d) 0300, (e) 0400, (f) 0500, (g) 0600, (h) 0700, (i) 0800, and (j) 0900 UTC 16 Dec 2013. Black line in (f) corresponds to the location of cross sections at 0500 UTC 16 Dec in Fig. 15.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
NEXRAD (KTYX) 0.5° reflectivity at times closest to (a) 0000, (b) 0100, (c) 0200, (d) 0300, (e) 0400, (f) 0500, (g) 0600, (h) 0700, (i) 0800, and (j) 0900 UTC 16 Dec 2013. Black line in (f) corresponds to the location of cross sections at 0500 UTC 16 Dec in Fig. 15.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
NEXRAD (KTYX) 0.5° reflectivity at times closest to (a) 0000, (b) 0100, (c) 0200, (d) 0300, (e) 0400, (f) 0500, (g) 0600, (h) 0700, (i) 0800, and (j) 0900 UTC 16 Dec 2013. Black line in (f) corresponds to the location of cross sections at 0500 UTC 16 Dec in Fig. 15.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
During IOP4, mobile surface snow observations were taken in Oswego, New Haven, Mexico, and Altmar, New York. The timing and types of precipitation observed at each location are outlined in Fig. 7. Snow was observed at all locations, with Oswego, New Haven, and Mexico experiencing more dendritic crystals during either a portion of or the entire event. Graupel was reported in New Haven, Mexico, and Altmar, indicating riming at differing times and locations. The presence of graupel suggests the existence of liquid water in the cloud available for accretion processes. The OWLeS King Air mission summary notes a considerable amount of liquid water at the cloud top near 3 km, causing icing-induced instrument malfunction. Also included in the summary is the lack of liquid water over land, around 1.8 km. However, the liquid water content did reach up to 1.3 gm−3 in the convective cells of the LES that had the ability to support high supersaturation production.
(bottom) Timelines of precipitation types observed by OWLeS spotters between 2200 and 0700 UTC at Oswego, New Haven, Mexico, and Altmar, New York. Snow is denoted by blue and graupel by green. Unless otherwise noted, the presence of colored bars represents the observation of that hydrometeor via the key. (top) A map of spotter locations is provided for spatial reference.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
(bottom) Timelines of precipitation types observed by OWLeS spotters between 2200 and 0700 UTC at Oswego, New Haven, Mexico, and Altmar, New York. Snow is denoted by blue and graupel by green. Unless otherwise noted, the presence of colored bars represents the observation of that hydrometeor via the key. (top) A map of spotter locations is provided for spatial reference.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
(bottom) Timelines of precipitation types observed by OWLeS spotters between 2200 and 0700 UTC at Oswego, New Haven, Mexico, and Altmar, New York. Snow is denoted by blue and graupel by green. Unless otherwise noted, the presence of colored bars represents the observation of that hydrometeor via the key. (top) A map of spotter locations is provided for spatial reference.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
For the majority of its lifetime, the LES was precipitating directly over and to the east of Lake Ontario (Fig. 6), so it is not surprising that the maximum 24-h accumulated liquid-equivalent precipitation of 17.92 mm reported by the AHPS was focused immediately east of Lake Ontario (Fig. 8a). The gradient of the precipitation was considerably tight, as much as 0.5 mm km−1, in both the north–south and east–west directions. The precipitation was generally bounded by 43°–44°N, 75°–77°W. Precipitation south of 43°N is not included in the detailed analysis of this event. The daily accumulations from Sandy Creek and North Redfield, New York, vary slightly compared to the AHPS data (Fig. 8a). While the North Redfield station reported 16.76 mm of liquid-equivalent precipitation, located in the 16–18 mm range measured by the AHPS, Sandy Creek reported 22.34 mm, which was 37%–46% larger than the AHPS range of 12–14 mm. Interestingly, a gauge-corrected NEXRAD ZH-derived product provided by the National Centers for Environmental Prediction (NCEP) was found by Welsh et al. (2016) to report about half of the nonspherical snow–water liquid-equivalent amounts observed at the Sandy Creek and North Redfield sites during an earlier OWLeS IOP. Based on this discrepancy and inherent data quality issues associated with radar retrievals discussed in section 3, caution is required when using these blended in situ and remotely retrieved observations as ground truth. Figure 8b displays the QPE derived for this paper from the 0.5° KTYX ZH during the same period. Although these data are not bias corrected by rain gauges, the result closely resembles that of AHPS, which is most likely a consequence of the lack of gauges in this confined area.
Quantitative precipitation estimates for the time period from 1200 UTC 15 Dec to 1200 UTC 16 Dec 2013 from (a) AHPS with daily snow–water liquid-equivalent measurements at the Sandy Creek and North Redfield stations, denoted by western and eastern triangles, respectively and (b) the KTYX 0.5° ZH derivation. The provided color scale is representative of both the AHPS and station observations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Quantitative precipitation estimates for the time period from 1200 UTC 15 Dec to 1200 UTC 16 Dec 2013 from (a) AHPS with daily snow–water liquid-equivalent measurements at the Sandy Creek and North Redfield stations, denoted by western and eastern triangles, respectively and (b) the KTYX 0.5° ZH derivation. The provided color scale is representative of both the AHPS and station observations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Quantitative precipitation estimates for the time period from 1200 UTC 15 Dec to 1200 UTC 16 Dec 2013 from (a) AHPS with daily snow–water liquid-equivalent measurements at the Sandy Creek and North Redfield stations, denoted by western and eastern triangles, respectively and (b) the KTYX 0.5° ZH derivation. The provided color scale is representative of both the AHPS and station observations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
As previously mentioned, the LES was well organized between 0000 and 0700 UTC 16 December. In Fig. 9, modeled and observed precipitation intensity is compared at both Sandy Creek and North Redfield. The maximum and minimum accumulations from MEY92S, MEY92H, DEM15S, and DEM15H serve as the bounds to the shading in Fig. 9. Indicated by the observations, the precipitation was most intense between 0000 and 0700 UTC with the maximum one-hour accumulation at 0500 UTC. There was a sharp drop in accumulations afterward as the LES propagated south of the observation sites. All simulations lag the maximum precipitation by an hour at North Redfield but accurately capture the timing at Sandy Creek. The simulations underforecast the precipitation at Sandy Creek before 0500 UTC and generally overforecast afterward as well as during all times at North Redfield. Despite the large spread in point hourly QPFs, the simulations capture the LES evolution fairly well.
Time series between 1200 UTC 15 Dec and 1200 UTC 16 Dec of hourly snow–water liquid-equivalent precipitation measurements (solid) at the (top) Sandy Creek and (bottom) North Redfield stations compared to the maximum and minimum simulated hourly accumulations (bounding the shaded area) from the MEY92S, MEY92H, DEM15S, and DEM15H simulations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Time series between 1200 UTC 15 Dec and 1200 UTC 16 Dec of hourly snow–water liquid-equivalent precipitation measurements (solid) at the (top) Sandy Creek and (bottom) North Redfield stations compared to the maximum and minimum simulated hourly accumulations (bounding the shaded area) from the MEY92S, MEY92H, DEM15S, and DEM15H simulations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Time series between 1200 UTC 15 Dec and 1200 UTC 16 Dec of hourly snow–water liquid-equivalent precipitation measurements (solid) at the (top) Sandy Creek and (bottom) North Redfield stations compared to the maximum and minimum simulated hourly accumulations (bounding the shaded area) from the MEY92S, MEY92H, DEM15S, and DEM15H simulations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
c. Dust concentrations during IOP4
While large spatial and temporal variations in CCN and na exist regionally, there is no considerable variation in the immediate downwind region of Lake Ontario during IOP4. D03-averaged background CCN and na data from the APM from 1200 UTC 15 December to 1200 UTC 16 December 2013 are provided in Fig. 10 along with the modeled temperature at 0.32, 2.1, and 3.8 km above ground level. These altitudes are approximately at cloud base, midcloud, and cloud top, respectively. CCN at 0.4% supersaturation are on the order of 100 cm−3 at the base and midcloud, but decrease to less than 0.1 cm−3 at cloud top during the LES (Fig. 10, purple lines). The value of na varies about 106 kg−1 (1 cm−3) at cloud base and decreases to about 105 kg−1 (0.1 cm−3) at midcloud and cloud top (Fig. 10, green lines). The data are representative of the background CCN values used in all simulations, and the na values are used to constrain nucleation rates in MEY92S and MEY92H and calculate nucleation rates in DEM15S and DEM15H. The temperatures are subfreezing during IOP4, ranging from −6° to −10°C at cloud base, −8° to −22°C at midcloud, and −17° to −27°C at cloud top (Fig. 10, black lines). Temperatures at all levels decrease throughout the event, with the midcloud to cloud-top cooling faster than the base. These temperature ranges cover all ice crystal growth zones, especially the dendritic growth zone, and are also amenable for riming and aggregation processes. Nucleation rates from both the Meyers et al. (1992) and DeMott et al. (2015) parameterizations will increase with these decreasing temperatures as altitude increases (Fig. 1).
Evolution of (top) D03-averaged background cloud condensation nuclei (CCN) at 0.4% supersaturation (purple; cm−3) and number concentration of dust >0.5 μm (na; green; kg−1) as simulated by the APM and (bottom) temperature (°C) at 3.8 km (approximately cloud top; dotted), 2.1 km (approximately midcloud; solid), and 0.32 km (approximately cloud base; dashed) above ground level in D03 from 1200 UTC 15 Dec to 1200 UTC 16 Dec 2013.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Evolution of (top) D03-averaged background cloud condensation nuclei (CCN) at 0.4% supersaturation (purple; cm−3) and number concentration of dust >0.5 μm (na; green; kg−1) as simulated by the APM and (bottom) temperature (°C) at 3.8 km (approximately cloud top; dotted), 2.1 km (approximately midcloud; solid), and 0.32 km (approximately cloud base; dashed) above ground level in D03 from 1200 UTC 15 Dec to 1200 UTC 16 Dec 2013.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Evolution of (top) D03-averaged background cloud condensation nuclei (CCN) at 0.4% supersaturation (purple; cm−3) and number concentration of dust >0.5 μm (na; green; kg−1) as simulated by the APM and (bottom) temperature (°C) at 3.8 km (approximately cloud top; dotted), 2.1 km (approximately midcloud; solid), and 0.32 km (approximately cloud base; dashed) above ground level in D03 from 1200 UTC 15 Dec to 1200 UTC 16 Dec 2013.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
d. Precipitation: WRF forecast versus AHPS observations
The AHPS data (Fig. 8a) are only available for accumulated precipitation over 24-h periods. For further analysis and understanding of the spatial pattern of the QPFs, the MEY92S simulation was split into 6-h time periods (Fig. 11a). As expected, there was little to no precipitation between 1200 and 1800 UTC 15 December. While the LES initiated between 1800 and 0000 UTC, the QPF increased to a maximum of 10.6 mm. A QPF of 15.0 mm occurred between 0000 and 0600 UTC with a matching contribution south of the area of main interest between 0600 and 1200 UTC. All three remaining simulations (Table 1) follow this evolution, with variation in QPF magnitude. Both the rapid intensification and decay of the precipitation associated with the LES is elucidated in this evolution culminating in a total 24-h QPF for each simulation valid at 1200 UTC 16 December in Figs. 11b–e. Each simulation provides a QPF with maxima > 25 mm, ranging from 25.7 mm in MEY92S (Fig. 11d) to 41.3 mm in DEM15S (Fig. 11b).
The 24-h model QPF (mm) valid at 1200 UTC 16 Dec for D03, which includes the geographic area shown in each of the panels above. (a) The 24-h temporal evolution of model precipitation accumulation for MEY92S in D03: (left to right) QPFs represent accumulations between 1200 and 1800 UTC 15 Dec, 1800 UTC 15 Dec and 0000 UTC 16 Dec, 0000 and 0600 UTC 16 Dec, and 0600 and 1200 UTC 16 Dec. QPFs are modeled with (b),(c) DeMott et al. (2015) and (d),(e) Meyers et al. (1992). The models are run with (b),(d) spherical ice growth and (c),(e) nonspherical ice growth. For spatial reference, the AHPS observations are overlaid in the 24-h QPF panels (white dashed contours) in the interest area where QPE = 4–18 mm; the dashed box over the color bar in (b) represents this range. Refer to Fig. 8a for specific magnitudes.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
The 24-h model QPF (mm) valid at 1200 UTC 16 Dec for D03, which includes the geographic area shown in each of the panels above. (a) The 24-h temporal evolution of model precipitation accumulation for MEY92S in D03: (left to right) QPFs represent accumulations between 1200 and 1800 UTC 15 Dec, 1800 UTC 15 Dec and 0000 UTC 16 Dec, 0000 and 0600 UTC 16 Dec, and 0600 and 1200 UTC 16 Dec. QPFs are modeled with (b),(c) DeMott et al. (2015) and (d),(e) Meyers et al. (1992). The models are run with (b),(d) spherical ice growth and (c),(e) nonspherical ice growth. For spatial reference, the AHPS observations are overlaid in the 24-h QPF panels (white dashed contours) in the interest area where QPE = 4–18 mm; the dashed box over the color bar in (b) represents this range. Refer to Fig. 8a for specific magnitudes.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
The 24-h model QPF (mm) valid at 1200 UTC 16 Dec for D03, which includes the geographic area shown in each of the panels above. (a) The 24-h temporal evolution of model precipitation accumulation for MEY92S in D03: (left to right) QPFs represent accumulations between 1200 and 1800 UTC 15 Dec, 1800 UTC 15 Dec and 0000 UTC 16 Dec, 0000 and 0600 UTC 16 Dec, and 0600 and 1200 UTC 16 Dec. QPFs are modeled with (b),(c) DeMott et al. (2015) and (d),(e) Meyers et al. (1992). The models are run with (b),(d) spherical ice growth and (c),(e) nonspherical ice growth. For spatial reference, the AHPS observations are overlaid in the 24-h QPF panels (white dashed contours) in the interest area where QPE = 4–18 mm; the dashed box over the color bar in (b) represents this range. Refer to Fig. 8a for specific magnitudes.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
The bulk of the forecast precipitation is downwind of Lake Ontario following the observations in Fig. 8a overlaid in Figs. 11b–e (white dashed contours), but the simulations include some widespread light accumulations (2–4 mm) throughout most of D03. Although the QPF magnitude is in major disagreement between each simulation and observations (about 97.5% greater than observations), the locations of maximum observed and modeled precipitation as defined as the area inside the innermost contour are relatively similar with the exception of MEY92S. DEM15H (Fig. 11c) provides the most accurate QPF in terms of the location of maximum precipitation, only 4.2 km from the observed maximum. The greatest distance of 23.4 km lies within the MEY92S QPF (Fig. 11d), whereas DEM15S and MEY92H (Figs. 11b,e) are 9.1 and 15.5 km from the observed maximum, respectively. This highlights a connection between the location of maximum QPF and habit in all simulations: nonspherical ice growth lends to a slightly better forecast spatially. Additionally, DEM15 is more adept at location placement than MEY92 due to location of the graupel accumulation maxima (not shown). The split QPF maxima produced by all simulations except MEY92H (Figs. 11b,c,d) was a result of the continued, slightly less intense precipitation to the southwest of the main maximum during the southward movement of the storm. Last, the spatial distribution of precipitation does not change drastically within the same nucleation parameterization. On average, the nonspherical ice growth method serves to increase the QPF: the average D03 QPF increases from 5.37 to 5.45 mm in MEY92 and from 5.37 to 5.50 mm in DEM15.
The 24-h period of these forecasts allows for direct quantitative comparison to the AHPS observations in Fig. 8a, keeping in mind that the precipitation may be underestimated. While the simulations seem to overforecast the LES precipitation, each captures the location of the precipitation. A more quantitative approach to analyzing the precipitation magnitude and location with respect to AHPS observations is provided in Fig. 12. Again, each simulation is forecasting more precipitation than what was reported by the AHPS. Both DEM15 simulations provide a solution that better matches the observations of the location of the average maximum precipitation (75.85°W), while the MEY92 simulations place the maximum to the east (75.7°W). While the absolute QPF maximum is greater in DEM15S (Fig. 11b) than DEM15H (Fig. 11c), nonspherical ice growth generally results in a greater averaged QPF due to the increase in north–south-oriented spatial coverage of high QPF near the absolute maximum. The relationship between DEM15 and MEY92 is less clear as the relative amount of precipitation varies longitudinally (Fig. 12) with DEM15 greater west of roughly 75.80°W and MEY92 greater to the east. The greatest mean squared error (MSE) of 12.7 is between DEM15S and MEY92S and the least, 3.3, is between DEM15H and DEM15S, indicating that the greatest source of forecast variance is between nucleation parameterizations with the secondary source being ice growth mode.
The 24-h liquid-equivalent precipitation (mm) valid at 1200 UTC 16 Dec 2013 averaged between 43° and 44°N in D03 for the explicit model QPFs from MEY92H, MEY92S, DEM15H, and DEM15S and for AHPS observations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
The 24-h liquid-equivalent precipitation (mm) valid at 1200 UTC 16 Dec 2013 averaged between 43° and 44°N in D03 for the explicit model QPFs from MEY92H, MEY92S, DEM15H, and DEM15S and for AHPS observations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
The 24-h liquid-equivalent precipitation (mm) valid at 1200 UTC 16 Dec 2013 averaged between 43° and 44°N in D03 for the explicit model QPFs from MEY92H, MEY92S, DEM15H, and DEM15S and for AHPS observations.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
As described in section 3c, radar variables were computed from each model simulation. Observed and forward-operator-generated 0.5° ZH are compared on 16 December during the LES lifetime (Fig. 13). At 0000 UTC, all simulations produce an LE cloud band of relatively high ZH that is farther south and more structurally banded compared to observations. Between 0200 and 0600 UTC, there is an increase in ZH just east of the lake in observations, and the simulation maxima move from over the lake to east of the lake. The simulations all over and underestimate ZH to a varying degree in D03 at all analysis times (presented quantitatively below) with DEM15 too aggressive with ZH and MEY92 presenting the opposite issue. Finally, the maximum ZH values in DEM15 are greater than those in MEY92. Despite the discrepancy between the observed and simulated ZH magnitudes, the simulations are able to generally capture the evolution of the LES. This increases confidence in the model to accurately forecast this LES and its associated physical processes.
PPIs of 0.5° reflectivity (dBZ) at (left to right) 0000, 0200, 0400, and 0600 UTC 16 Dec 2013 observed by KTYX and simulated by (top to bottom) DEM15S, DEM15H, MEY92S, and MEY92H.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
PPIs of 0.5° reflectivity (dBZ) at (left to right) 0000, 0200, 0400, and 0600 UTC 16 Dec 2013 observed by KTYX and simulated by (top to bottom) DEM15S, DEM15H, MEY92S, and MEY92H.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
PPIs of 0.5° reflectivity (dBZ) at (left to right) 0000, 0200, 0400, and 0600 UTC 16 Dec 2013 observed by KTYX and simulated by (top to bottom) DEM15S, DEM15H, MEY92S, and MEY92H.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
The general similarities in reflectivity magnitudes between KTYX observations and WRF simulations (Fig. 13) give pause to the larger modeled precipitation values (Figs. 11b–e) relative to the KTYX-derived precipitation (Fig. 8b) for all simulations. Hence, the simulated radar ZH (i.e., Fig. 13) derived from modeled microphysical quantities is used in the aforementioned Z–R relationship to derive a secondary precipitation product for each simulation (not shown). This derivation method allows for a closer comparison with the highly used reflectivity–rainfall relationship (e.g., AHPS).
The QPFs derived from the modeled ZH are closer to the observations derived from KTYX ZH; the explicit QPFs are around 97.5% greater than the AHPS QPE whereas the derived QPF are 1.2% less to 21.9%–56.5% greater than the KTYX-derived QPE, with MEY92S providing a lesser QPF and DEM15H producing the greatest difference in QPF. The ability of the model to capture the ZH signature relatively well and produce the derived forecasts close to KTYX-derived forecasts for MEY92 suggests that the explicit model QPF (i.e., Fig. 11) may be closer to ground truth for this specific LES than the AHPS observations or KTYX-derived precipitation (Fig. 8). This result implies that pertinent information is lost in the assumptions of the precipitation derivation, which is most likely a function of the accumulating hydrometeors (e.g., snow, ice, graupel, rain). It is difficult to express the size distributions of one hydrometeor type in two constant values, never mind a mixture. In addition, liquid water is more reflective than ice, so any liquid water present can exacerbate QPE errors (Markowski and Richardson 2010).
The major difference between DEM15 and MEY92 is the presence of rain and graupel (Fig. 14), especially over the lake (Figs. 15e–h), during the lifetime of the LES. The rain and graupel aloft, most of which is not falling to the surface but sampled by the radar over the lake, causes the large uptick in ZH values in DEM15 (Fig. 13) and therefore derived QPF (not shown). The model QPF (Fig. 11) does not display the same signature because the large amount of rain and graupel that the radar is sensing is greater than what is truly accumulating at the surface. To confirm, the DEM15 D03 QPF sum calculated for a theoretical scanning angle of 0.01° decreased by 13.3% due to the reduced concentrations of rain and graupel near the surface (not shown). It is hypothesized that if this LES only produced snow, then the observed QPE (i.e., Fig. 8) and explicit model QPF (i.e., Fig. 11) may have resulted in similar values.
Evolution of rain (purple), ice and snow (turquoise), and graupel (green) accumulated at the surface in D03 from 1800 UTC 15 Dec to 0800 UTC 16 Dec 2013 for (top to bottom) DEM15S, DEM15H, MEY92S, and MEY92H. Total accumulation magnitudes (mm) of each hydrometeor class are provided in the legends of each subplot for the presented time period.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Evolution of rain (purple), ice and snow (turquoise), and graupel (green) accumulated at the surface in D03 from 1800 UTC 15 Dec to 0800 UTC 16 Dec 2013 for (top to bottom) DEM15S, DEM15H, MEY92S, and MEY92H. Total accumulation magnitudes (mm) of each hydrometeor class are provided in the legends of each subplot for the presented time period.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Evolution of rain (purple), ice and snow (turquoise), and graupel (green) accumulated at the surface in D03 from 1800 UTC 15 Dec to 0800 UTC 16 Dec 2013 for (top to bottom) DEM15S, DEM15H, MEY92S, and MEY92H. Total accumulation magnitudes (mm) of each hydrometeor class are provided in the legends of each subplot for the presented time period.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Cross sections along 43.5°N of (a),(b) ice mixing ratio (g kg−1), (c),(d) snow mixing ratio (g kg−1), (e),(f) rain mixing ratio (g kg−1), (g),(h) graupel mixing ratio (g kg−1), (i),(j) cloud mixing ratio (g kg−1), (k),(l) Ni (L−1), and (m),(n) aspect ratio overlaid with temperature (°C) for (left) MEY92H and (right) DEM15H at 0500 UTC 16 Dec 2013. Filled black contours at the bottom of each subplot indicate terrain height.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Cross sections along 43.5°N of (a),(b) ice mixing ratio (g kg−1), (c),(d) snow mixing ratio (g kg−1), (e),(f) rain mixing ratio (g kg−1), (g),(h) graupel mixing ratio (g kg−1), (i),(j) cloud mixing ratio (g kg−1), (k),(l) Ni (L−1), and (m),(n) aspect ratio overlaid with temperature (°C) for (left) MEY92H and (right) DEM15H at 0500 UTC 16 Dec 2013. Filled black contours at the bottom of each subplot indicate terrain height.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Cross sections along 43.5°N of (a),(b) ice mixing ratio (g kg−1), (c),(d) snow mixing ratio (g kg−1), (e),(f) rain mixing ratio (g kg−1), (g),(h) graupel mixing ratio (g kg−1), (i),(j) cloud mixing ratio (g kg−1), (k),(l) Ni (L−1), and (m),(n) aspect ratio overlaid with temperature (°C) for (left) MEY92H and (right) DEM15H at 0500 UTC 16 Dec 2013. Filled black contours at the bottom of each subplot indicate terrain height.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
e. Microphysical responses
As previously discussed, the inhomogeneous mixture of hydrometeors include contributions from rain, snow and ice, and graupel, which vary slightly among ice habit and greatly among nucleation sensitivity simulations (Fig. 14). When ice evolves nonspherically, the total amount of snow and ice accumulating at the surface decreases in MEY92 and increases in DEM15. In addition, the amount of snow and ice decreases and both rain and graupel increase when changing the nucleation parameterization from Meyers et al. (1992) to DeMott et al. (2015). A surprising finding from this precipitation breakdown was the presence of accumulated rain in DEM15, which was not observed by OWLeS spotters at any of the locations from 2200 to 0700 UTC (Fig. 7). The time series of multiple hydrometeor classes that accumulated hourly provided in Fig. 14 reveals that in DEM15 the majority of rainfall occurred almost entirely before 0100 UTC 16 December, with smaller amounts during the LES. The magnitude of accumulated rain as simulated in DEM15 was much less than that of snow and ice when the LES was of greatest intensity from 0100 to 0700 UTC (Fig. 14). Again, due to the diversity of hydrometeors in this LES and the lack of plentiful ground-based in situ observations, quantifying the precipitation that accumulated on the ground proves to be difficult.
To better understand the hydrometeor type differences between MEY92 and DEM15, microphysical properties are investigated. Due to similarities between hydrometeor types for a given nucleation parameterization, only the nonspherical results between MEY92 and DEM15 are assessed. Cross sections of the LES at 0500 UTC 16 December are provided in Fig. 15. Differences exist among each due to the variance of microphysical properties. The main contributor to these differences lie within the INP number concentration. As discussed previously, the Meyers et al. (1992) parameterization nucleates a greater number of ice crystals than that of DeMott et al. (2015) (Fig. 1); this holds true in the simulations. The immersion freezing nucleation rates in the DEM15 simulation are lower than the deposition–condensation freezing nucleation rates in the MEY92 simulations (Fig. 16). As such, the ice mass and Ni are greater in MEY92 than in DEM15 (Figs. 15a,b,k,l and 17).
Autoconversion, immersion freezing nucleation, deposition nucleation, graupel deposition, riming, and snow–ice aggregation rates (g kg−1 s−1) summed through D03 for (a) DEM15 and (b) MEY92 between 1800 UTC 15 Dec and 0800 UTC 16 Dec 2013. Spherical rates are denoted with solid lines and nonspherical rates are denoted with dashed lines.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Autoconversion, immersion freezing nucleation, deposition nucleation, graupel deposition, riming, and snow–ice aggregation rates (g kg−1 s−1) summed through D03 for (a) DEM15 and (b) MEY92 between 1800 UTC 15 Dec and 0800 UTC 16 Dec 2013. Spherical rates are denoted with solid lines and nonspherical rates are denoted with dashed lines.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Autoconversion, immersion freezing nucleation, deposition nucleation, graupel deposition, riming, and snow–ice aggregation rates (g kg−1 s−1) summed through D03 for (a) DEM15 and (b) MEY92 between 1800 UTC 15 Dec and 0800 UTC 16 Dec 2013. Spherical rates are denoted with solid lines and nonspherical rates are denoted with dashed lines.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Hourly ice, snow, rain, cloud, and graupel mixing ratios (g kg−1) summed through D03 for (a) DEM15 and (b) MEY92 between 1800 UTC 15 Dec and 0800 UTC 16 Dec 2013. Mixing ratios for spherical ice growth are denoted with solid lines and those for nonspherical growth are denoted with dashed lines.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Hourly ice, snow, rain, cloud, and graupel mixing ratios (g kg−1) summed through D03 for (a) DEM15 and (b) MEY92 between 1800 UTC 15 Dec and 0800 UTC 16 Dec 2013. Mixing ratios for spherical ice growth are denoted with solid lines and those for nonspherical growth are denoted with dashed lines.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Hourly ice, snow, rain, cloud, and graupel mixing ratios (g kg−1) summed through D03 for (a) DEM15 and (b) MEY92 between 1800 UTC 15 Dec and 0800 UTC 16 Dec 2013. Mixing ratios for spherical ice growth are denoted with solid lines and those for nonspherical growth are denoted with dashed lines.
Citation: Journal of the Atmospheric Sciences 76, 11; 10.1175/JAS-D-19-0004.1
Aggressive ice formation by the Meyers et al. (1992) parameterization potentially leads to preferential growth of ice at the expense of liquid drops. This decreases the number of cloud drops, reducing autoconversion processes substantially. A lack of small cloud and large rain droplets effectively reduces the riming processes within the cloud and the rain accumulating at the surface. Thus, many ice crystals will dominate depositional growth at the expense of available moisture in the cloud and sediment as ice and aggregate snow. The snow mass is slightly greater in MEY92 (Figs. 15c,d) due to increased aggregation of ice by snow (Fig. 16). Conversely, the DeMott et al. (2015) parameterization nucleates fewer INP and the initiation of ice depends on the existence of cloud droplets within the cloud. Therefore, less ice nucleates based on the lower initial INP, compared to MEY92. Droplets can still grow in the presence of this ice, reaching sizes representative of rain, leading to autoconversion. The strong presence of liquid water in the cloud is conducive to riming processes. A deeper look into the process rates between MEY92 and DEM15 indicates that there is a greater conversion rate to graupel after riming processes due to the greater amount of available liquid in the DEM15 simulations (Fig. 16). Ultimately, this results in less snow and ice, more rain, and more graupel. The number of nucleated ice crystals in a mixed-phase cloud with supercooled liquid has a strong effect on the hydrometeor types that grow in the cloud and subsequently sediment. Therefore, due to the sometimes extreme differences in nucleated ice number, hydrometeor differences that extend from in the cloud down to the surface exist between MEY92 and DEM15. Finally, the nonspherical simulations experience greater ice–snow aggregation rates due to the increased ice mass via depositional growth (Fig. 16).
Cross sections highlight that MEY92 has a greater Ni (Figs. 15k,l), by a few orders of magnitude, and ice mass (Figs. 15a,b) compared to DEM15, especially over Lake Ontario. A D03 average of mixing ratios during the entire LES event shows that the MEY92 simulations have more ice and snow during the LES, but there are more cloud droplets, rain, and graupel in the DEM15 simulations (Figs. 14 and 17); these differences are also evident in the cross sections at 0500 UTC (Figs. 15a–j). As seen in Fig. 16, DEM15 has an autoconversion rate about two orders of magnitude greater than that of MEY92, which can be directly related to the abundance of cloud drops in both DEM15 simulations. With more rain, the collection of droplets by snow to form graupel (riming) is also larger in DEM15, with subsequent increases in graupel deposition (Fig. 16). The depositional growth rate for nonspherical ice crystals is greater than that for spherical ice crystals (not shown), leading to greater ice mass (Fig. 17) and exaggerated habits (Figs. 15m,n) in the nonspherical simulations. The ice crystals at this time are mostly plates with some pockets of columns in DEM15H (Fig. 15n) but are largely plates in MEY92 with some possible dendrites as suggested by ϕ ≪ 1 (Fig. 15m). With all of these factors combined, it is not surprising that the DEM15H simulation produced the greatest average precipitation forecast (Fig. 11c) with the relatively large presence of rain and graupel in addition to efficient depositional growth.
The LES in IOP4 responds systematically to changes to ice nucleation parameterization and crystal growth within the WRF Model. To better quantify these responses, percent errors are provided in Table 2 for precipitation, mixing ratios, and sedimentation rates with MEY92S as the control simulation due to its independence from aerosol data and its spherical mode of ice growth. As previously discussed, the parameterization of Meyers et al. (1992) nucleates a greater number of ice crystals than that of DeMott et al. (2015) in both spherical and nonspherical simulations. There is more rain and graupel mass and therefore greater associated sedimentation rates in DEM15 likely due to the aforementioned increased autoconversion. With a decrease in ice crystal number, the sedimentation rates of ice and snow are also comparatively lower in DEM15. There are also intriguing microphysical relationships valid for both MEY92 and DEM15 when ice evolves nonlinearly. Ice, snow, and graupel mass increase whereas rain mass decreases with nonspherical growth. Ice mass increases as a direct result of the increased depositional growth rate, which then increases the sedimentation rates for ice and snow, but only in DEM15. The snow sedimentation rate increases in MEY92 as well, but the ice sedimentation rate decreases suggesting subsequent processes such as aggregation may be overtaking the mass effect. In addition, the crystals have a nonspherical habit (ϕ < 1 or ϕ > 1) with different levels of exaggeration due to this growth. The rain sedimentation rate likely decreases due to the growth of ice; vapor is now preferentially growing crystals at the expense of smaller droplets.
Percentage error in D03-summed quantities compared to a simulation run with Meyers et al. (1992) nucleation and spherical ice growth, where ni is Ni, Qi is ice mass mixing ratio, Qs is snow mass mixing ratio, Qg is graupel mass mixing ratio, Qr is rain mass mixing ratio, iDep is ice deposition rate, iSed is ice sedimentation rate, sSed is snow sedimentation rate, gSed is graupel sedimentation rate, and rSed is rain sedimentation rate.
There is an unexpected response in the decrease of Ni with nonspherical growth in MEY92 by 9.9% (Table 2). The nucleation parameterization is not manipulated with this habit change as all particles are nucleated with ϕ = 1. Referring back to Eqs. (1) and (2), it is evident that these differences must stem from changes in ice supersaturation and temperature, respectively. The driving force of existing temperature differences is increased latent heating as a result of the greater depositional ice growth (Table 2) in the nonspherical simulations. This growth also serves to decrease the ice supersaturation through increased uptake of vapor by ice crystals. Between 1200 UTC 15 December and 1200 UTC 16 December where there were pristine ice crystals in D03, the MEY92H simulation is, on average, 0.046°C warmer than the MEY92S simulation, and the ice supersaturation decreased by 0.023%. The slightly warmer temperatures and lesser ice supersaturations fed into the nucleation parameterizations lead directly to smaller nucleation rates in the nonspherical simulation. Offline tests demonstrate that a supersaturation difference of this magnitude has the ability to suppress Meyers et al. (1992) Ni nucleation by 1–100 kg−1, depending on the ambient ice supersaturation. A similar pattern is observed for DEM15 simulations due to their temperature dependence. Finally, the accumulated precipitation responds similarly in the AHM: average precipitation increases by up to 0.8% when changing parameterizations from Meyers et al. (1992) to DeMott et al. (2015) (not shown) and increases by 1.6–2.3% with nonspherical ice growth (Table 2).
5. Conclusions
The Meyers et al. (1992) and DeMott et al. (2015) ice nucleation parameterizations were implemented into the AHM to investigate their influence on cloud properties and sedimentation processes. There are no traditional WRF microphysics options that allow for the nonlinear evolution of nonspherical ice growth, which has the potential to impact the forecast accuracy of high-impact cold-season events. However, note that the AHM is now available via the new Ice-Spheroids Habit Model with Aspect-Ratio Evolution (ISHMAEL; Jensen et al. 2017), released in version 4.1 of WRF. With increasing computational efficiency, it is reasonable to make use of the AHM in mesoscale modeling. The choice of ice nucleation parameterization resulted in a greater QPF difference compared to the ice crystal growth process due to the increase in hydrometeor diversity that composed the LES when employing the DeMott et al. (2015) parameterization. The presence or lack of homogeneity among the hydrometeors stemmed from the number of nucleated crystals Ni, which either depleted the number of cloud droplets to grow ice in MEY92 simulations (i.e., glaciation) or allowed for simultaneous growth of ice and cloud droplets in DEM15. The resulting Ni from the individual nucleation parameterizations significantly impacted concentration-dependent microphysical process rates such as aggregation, riming, and deposition, substantially altering the mixture of hydrometeors, sedimentation rates, radar reflectivity calculations, and precipitation quantities. Hence, the analysis in this paper should provide not only insight, but caution on the cascading influence of microphysical processes on the system as a whole. In this regard, direct observations of Ni are important to constrain and improve the model. The secondary controlling difference among the simulations of nonspherical ice growth increased the ice mass in the LES and its associated QPF. Validating the QPFs proved difficult due to the sparseness of ground-based in situ precipitation measurements. Skepticism was placed on radar-derived precipitation in IOP4 due to the mixture of hydrometeors in this LES leading to less reliable constants in the Z–R relationship. This suggests that the explicit model QPFs may be more representative of ground truth than the radar-derived estimates. Further research is necessary to increase confidence in such precipitation datasets during events where in situ observations are relatively limited, leading to more robust forecast verification and model validation. It is important to note that these conclusions should not be extrapolated and applied to other cold-season LES cases, as they are specific only to this 15–16 December 2013 case. Analysis of multiple LES is necessary to take the next step in understanding how best to model these events to elicit the best forecast in terms of timing, location, and magnitude of precipitation.
Acknowledgments
This research was supported by the National Science Foundation Partnership for International Research and Education Program between the United States and Taiwan, OISE-1545917, awarded to the University at Albany, SUNY. The authors are grateful for the comments and suggestions of three anonymous reviewers, which significantly improved this manuscript. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation (NSF). OWLeS data were provided by NCAR/EOL under the sponsorship of the NSF.
REFERENCES
Ahrens, D. C., 2013: Meteorology Today. 10th ed. Brooks/Cole, 569 pp.
Ardon-Dryer, K., 2012: Ice nuclei, their concentration and efficiency in clean and polluted air and their effects on clouds and precipitation. Ph.D. thesis, Tel Aviv University, 187 pp.
Baxter, M. A., C. E. Graves, and J. T. Moore, 2005: A climatology of snow-to-liquid ratio for the contiguous United States. Wea. Forecasting, 20, 729–744, https://doi.org/10.1175/WAF856.1.
Bergmaier, P. T., and B. Geerts, 2016: Airborne radar observations of lake-effect snowbands over the New York Finger Lakes. Mon. Wea. Rev., 144, 3895–3914, https://doi.org/10.1175/MWR-D-16-0103.1.
Bergmaier, P. T., B. Geerts, L. S. Campbell, and W. J. Steenburgh, 2017: The OWLeS IOP2b lake-effect snowstorm: Dynamics of the secondary circulation. Mon. Wea. Rev., 145, 2437–2459, https://doi.org/10.1175/MWR-D-16-0462.1.
Braham, R. R., 1983: The Midwest snow storm of 8–11 December 1977. Mon. Wea. Rev., 111, 253–272, https://doi.org/10.1175/1520-0493(1983)111<0253:TMSSOD>2.0.CO;2.
Campbell, L. S., W. J. Steenburgh, P. G. Veals, T. W. Letcher, and J. R. Minder, 2016: Lake-effect mode and precipitation enhancement over the tug hill plateau during OWLeS IOP2b. Mon. Wea. Rev., 144, 1729–1748, https://doi.org/10.1175/MWR-D-15-0412.1.
Chen, J.-P., and T.-C. Tsai, 2016: Triple-moment modal parameterization for the adaptive growth habit of pristine ice crystals. J. Atmos. Sci., 73, 2105–2122, https://doi.org/10.1175/JAS-D-15-0220.1.
Cziczo, D. J., L. Ladino, Y. Boose, Z. A. Kanji, P. Kupiszewski, S. Lance, S. Mertes, and H. Wex, 2017: Measurements of ice nucleating particles and ice residuals. Ice Formation and Evolution in Clouds and Precipitation: Measurement and Modeling Challenges, Meteor. Monogr., No. 58, Amer. Meteor. Soc., https://doi.org/10.1175/AMSMONOGRAPHS-D-16-0008.1.
de Boer, G., H. Morrison, M. D. Shupe, and R. Hildner, 2011: Evidence of liquid dependent ice nucleation in high-latitude stratiform clouds from surface remote sensors. Geophys. Res. Lett., 38, L01803, https://doi.org/10.1029/2010GL046016.
DeMott, P. J., 1990: An exploratory study of ice nucleation by soot aerosols. J. Appl. Meteor., 29, 1072–1079, https://doi.org/10.1175/1520-0450(1990)029<1072:AESOIN>2.0.CO;2.
DeMott, P. J., and Coauthors, 2010: Predicting global atmospheric ice nuclei distributions and their impacts on climate. Proc. Natl. Acad. Sci. USA, 107, 11 217–11 222, https://doi.org/10.1073/pnas.0910818107.
DeMott, P. J., and Coauthors, 2015: Integrating laboratory and field data to quantify the immersion freezing ice nucleation activity of mineral dust particles. Atmos. Chem. Phys., 15, 393–409, https://doi.org/10.5194/acp-15-393-2015.
DeMott, P. J., and Coauthors, 2017: Comparative measurements of ambient atmospheric concentrations of ice nucleating particles using multiple immersion freezing methods and a continuous flow diffusion chamber. Atmos. Chem. Phys., 17, 11 227–11 245, https://doi.org/10.5194/acp-17-11227-2017.
Dudhia, J., 1989: Numerical study of convection observed during the Winter Monsoon Experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46, 3077–3107, https://doi.org/10.1175/1520-0469(1989)046<3077:NSOCOD>2.0.CO;2.
Field, P., and Coauthors, 2017: Secondary ice production: Current state of the science and recommendations for the future. Ice Formation and Evolution in Clouds and Precipitation: Measurement and Modeling Challenges, Meteor. Monogr., No. 58, Amer. Meteor. Soc., https://doi.org/10.1175/AMSMONOGRAPHS-D-16-0014.1.
Garimella, S., D. A. Rothenberg, M. J. Wolf, R. O. David, Z. A. Kanji, C. Wang, M. Rosch, and D. J. Czizco, 2017: Uncertainty in counting ice nucleating particles with continuous flow diffusion chambers. Atmos. Chem. Phys., 17, 10 855–10 864, https://doi.org/10.5194/acp-17-10855-2017.
Garimella, S., D. A. Rothenberg, M. J. Wolf, C. Wang, and D. J. Czizco, 2018: How uncertainty in field measurements of ice nucleating particles influences modeled cloud forcing. J. Atmos. Sci., 75, 179–187, https://doi.org/10.1175/JAS-D-17-0089.1.
Harrington, J. Y., K. Sulia, and H. Morrison, 2013a: A method for adaptive habit prediction in bulk microphysical models. Part I: Theoretical development. J. Atmos. Sci., 70, 349–364, https://doi.org/10.1175/JAS-D-12-040.1.
Harrington, J. Y., K. Sulia, and H. Morrison, 2013b: A method for adaptive habit prediction in bulk microphysical models. Part II: Parcel model corroboration. J. Atmos. Sci., 70, 365–376, https://doi.org/10.1175/JAS-D-12-0152.1.
Hiron, T., and A. I. Flossman, 2015: A study of the role of the parameterization of heterogeneous ice nucleation for the modeling of microphysics and precipitation of a convective cloud. J. Atmos. Sci., 72, 3322–3339, https://doi.org/10.1175/JAS-D-15-0026.1.
Hjelmfelt, M. R., 1990: Numerical study of the influence of environmental conditions on lake-effect snowstorms over Lake Michigan. Mon. Wea. Rev., 118, 138–150, https://doi.org/10.1175/1520-0493(1990)118<0138:NSOTIO>2.0.CO;2.
Hong, S. Y., and Y. Noh, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 2318–2341, https://doi.org/10.1175/MWR3199.1.
Hong, S. Y., J. Dudhia, and S. H. Chen, 2004: A revised approach to ice microphysical processes for the bulk parameterization of clouds and precipitation. Mon. Wea. Rev., 132, 103–120, https://doi.org/10.1175/1520-0493(2004)132<0103:ARATIM>2.0.CO;2.
Iacono, M. J., J. S. Delamere, E. J. Mlawer, M. W. Shephard, S. A. Clough, and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, https://doi.org/10.1029/2008JD009944.
Jensen, A. A., J. Y. Harrington, H. Morrison, and J. A. Milbrandt, 2017: Predicting ice shape evolution in a bulk microphysics model. J. Atmos. Sci., 74, 2081–2104, https://doi.org/10.1175/JAS-D-16-0350.1.
Jensen, A. A., J. Y. Harrington, and H. Morrison, 2018: Microphysical characteristics of squall-line stratiform precipitation and transition zones simulated using an ice particle property-evolving model. Mon. Wea. Rev., 146, 723–743, https://doi.org/10.1175/MWR-D-17-0215.1.
Kain, J. S., 2004: The Kain–Fritsch convective parameterization: An update. J. Appl. Meteor., 43, 170–181, https://doi.org/10.1175/1520-0450(2004)043<0170:TKCPAU>2.0.CO;2.
Kanji, Z. A., L. A. Ladino, H. Wex, Y. Boose, M. Burkert-Kohn, D. J. Cziczo, and M. Kramer, 2017: Overview of ice nucleating particles. Ice Formation and Evolution in Clouds and Precipitation: Measurement and Modeling Challenges, Meteor. Monogr., No. 58, Amer. Meteor. Soc., https://doi.org/10.1175/AMSMONOGRAPHS-D-16-0006.1.
Kristovich, D. A. R., and N. F. Laird, 1998: Observations of widespread lake-effect cloudiness: Influences of lake surface temperature and upwind conditions. Wea. Forecasting, 13, 811–821, https://doi.org/10.1175/1520-0434(1998)013<0811:OOWLEC>2.0.CO;2.
Kristovich, D. A. R., and Coauthors, 2017: The Ontario Winter Lake-effect Systems field campaign: Scientific and educational adventures to further our knowledge and prediction of lake-effect storms. Bull. Amer. Meteor. Soc., 98, 315–332, https://doi.org/10.1175/BAMS-D-15-00034.1.
Laird, N. F., N. D. Metz, L. Gaudet, C. Grasmick, L. Higgins, C. Loeser, and D. A. Zelinsky, 2017: Climatology of cold season lake-effect cloud bands for the North American Great Lakes. Int. J. Climatol., 37, 2111–2121, https://doi.org/10.1002/joc.4838.
Levin, Z., A. Teller, E. Ganor, and Y. Yin, 2005: On the interactions of mineral dust, sea-salt particles, and clouds: A measurement and modeling study from the Mediterranean Israeli Dust Experiment campaign. J. Geophys. Res., 110, D20202, https://doi.org/10.1029/2005JD005810.
Luo, G., and F. Yu, 2011: Simulation of particle formation and number concentration over the eastern United States with the WRF-Chem + APM model. Atmos. Chem. Phys., 11, 11 521–11 533, https://doi.org/10.5194/acp-11-11521-2011.
Markowski, P., and Y. Richardson, 2010: Mesoscale Meteorology in Midlatitudes. Wiley-Blackwell, 430 pp.
Marshall, J. S., R. C. Langille, and W. M. Palmer, 1947: Measurement of rainfall by radar. J. Meteor., 4, 186–192, https://doi.org/10.1175/1520-0469(1947)004<0186:MORBR>2.0.CO;2.
Meyers, M. P., P. J. DeMott, and W. R. Cotton, 1992: New primary ice-nucleation parameterizations in an explicit cloud model. J. Appl. Meteor., 31, 708–721, https://doi.org/10.1175/1520-0450(1992)031<0708:NPINPI>2.0.CO;2.
Murray, B. J., S. L. Broadley, T. W. Wilson, J. D. Atkinson, and R. H. Wills, 2011: Heterogeneous freezing of water droplets containing kaolinite particles. Atmos. Chem. Phys., 11, 4191–4207, https://doi.org/10.5194/acp-11-4191-2011.
NOAA, 2013: Average GLSEA surface water temperature data. Great Lakes Environmental Research Laboratory, accessed 20 March 2018, https://coastwatch.glerl.noaa.gov/ftp/glsea/avgtemps/2013/glsea-temps2013_1024.dat.
Phillips, V. T. J., L. J. Donner, and S. T. Garner, 2007: Nucleation processes in deep convection simulated by a cloud-system-resolving model with double-moment bulk microphysics. J. Atmos. Sci., 64, 738–761, https://doi.org/10.1175/JAS3869.1.
Phillips, V. T. J., P. J. DeMott, and C. Andronache, 2008: An empirical parameterization of heterogeneous ice nucleation for multiple chemical species of aerosol. J. Atmos. Sci., 65, 2757–2783, https://doi.org/10.1175/2007JAS2546.1.
Phillips, V. T. J., and Coauthors, 2017: Ice multiplication by breakup in ice–ice collisions. Part II: Numerical simulations. J. Atmos. Sci., 74, 2789–2811, https://doi.org/10.1175/JAS-D-16-0223.1.
Phillips, V. T. J., S. Patade, J. Gutierrez, and A. Bansemer, 2018: Secondary ice production by fragmentation of freezing drops: Formulation and theory. J. Atmos. Sci., 75, 3031–3070, https://doi.org/10.1175/JAS-D-17-0190.1.
Prenni, A. J., and Coauthors, 2007: Can ice-nucleating aerosols affect Arctic seasonal climate? Bull. Amer. Meteor. Soc., 88, 541–550, https://doi.org/10.1175/BAMS-88-4-541.
Pruppacher, H. R., and J. D. Klett, 1997: Microphysics of Clouds and Precipitation. Kluwer Academic, 954 pp.
Ryzhkov, A. V., P. Zhang, H. D. Reeves, M. Kumjian, T. Tschallener, S. Troemel, and C. Simmer, 2016: Quasi-vertical profiles—A new way to look at polarimetric radar data. J. Atmos. Oceanic Technol., 33, 551–562, https://doi.org/10.1175/JTECH-D-15-0020.1.
Solomon, A., G. Feingold, and M. D. Shupe, 2015: The role of ice nuclei recycling in the maintenance of cloud ice in Arctic mixed-phase stratocumulus. Atmos. Chem. Phys., 15, 10 631–10 643, https://doi.org/10.5194/acp-15-10631-2015.
Stout, G. E., and E. A. Mueller, 1968: Survey of relationships between rainfall rate and radar reflectivity in the measurement of precipitation. J. Appl. Meteor., 7, 465–474, https://doi.org/10.1175/1520-0450(1968)007<0465:SORBRR>2.0.CO;2.
Sulia, K. J., and J. Y. Harrington, 2011: Ice aspect ratio influences on mixed-phase clouds: Impacts on phase partitioning in parcel models. J. Geophys. Res., 116, D21309, https://doi.org/10.1029/2011JD016298.
Sulia, K. J., and M. R. Kumjian, 2017a: Simulated polarimetric fields of ice vapor growth using the adaptive habit model. Part I: Large-eddy simulations. Mon. Wea. Rev., 145, 2281–2302, https://doi.org/10.1175/MWR-D-16-0061.1.
Sulia, K. J., and M. R. Kumjian, 2017b: Simulated polarimetric fields of ice vapor growth using the adaptive habit model. Part II: A case study from the FROST experiment. Mon. Wea. Rev., 145, 2303–2323, https://doi.org/10.1175/MWR-D-16-0062.1.
Sulia, K. J., J. Y. Harrington, and H. Morrison, 2013: A method for adaptive habit prediction in bulk microphysical models. Part III: Applications and studies within a two-dimensional kinematic model. J. Atmos. Sci., 70, 3302–3320, https://doi.org/10.1175/JAS-D-12-0316.1.
Sulia, K. J., H. Morrison, and J. Y. Harrington, 2014: Dynamical and microphysical evolution during mixed-phase cloud glaciation simulated using the bulk adaptive habit prediction model. J. Atmos. Sci., 71, 4158–4180, https://doi.org/10.1175/JAS-D-14-0070.1.
Welsh, D., B. Geerts, X. Jin, P. T. Bergmaier, J. R. Minder, W. J. Steenburgh, and L. S. Campbell, 2016: Understanding heavy lake-effect snowfall: The vertical structure of radar reflectivity in a deep snowband over and downwind of Lake Ontario. Mon. Wea. Rev., 144, 4221–4244, https://doi.org/10.1175/MWR-D-16-0057.1.
Wright, D. M., D. J. Posselt, and A. L. Steiner, 2013: Sensitivity of lake-effect snowfall to lake ice cover and temperature in the Great Lakes region. Mon. Wea. Rev., 141, 670–689, https://doi.org/10.1175/MWR-D-12-00038.1.
Yu, F., and G. Luo, 2009: Simulation of particle size distribution with a global aerosol model: Contribution of nucleation to aerosol and CCN number concentrations. Atmos. Chem. Phys., 9, 7691–7710, https://doi.org/10.5194/acp-9-7691-2009.
