1. Introduction
The development of spaceborne remote sensing instruments has allowed for clouds and precipitation to be characterized over vast regions of the world, including areas where measurements are sparse or not easily accessible. They provide a means to evaluate global and regional numerical weather prediction models (e.g., Delanoë et al. 2011; Ori et al. 2020). The Tropical Rainfall Measuring Mission (TRMM; Kummerow et al. 1998) provided the groundwork for observing rainfall over tropical and subtropical regions while the CloudSat mission Core Observatory (Stephens et al. 2002) continues to detect most clouds due to its highly sensitive profiling radar. The Global Precipitation Measurement (GPM) mission (Hou et al. 2014; Skofronick-Jackson et al. 2017) was launched in 2014 to expand the coverage of precipitation to the middle and high latitudes where snowfall is more prevalent. Since approximately 60% of precipitation by mass can be linked to ice processes aloft (A. J. Heymsfield et al. 2020), the global coverage and instrument capabilities of GPM aim to further our understanding of Earth’s water budget. On board the GPM Core Observatory is the first spaceborne Dual-Frequency Precipitation Radar (DPR) that measures the 3D precipitation structure at Ka (35.5 GHz) and Ku (13.5 GHz) bands and provides estimates of the particle size distribution (PSD) from a priori microphysical assumptions such as mass–dimension and projected area–dimension relationships (Skofronick-Jackson et al. 2019). Snowfall properties retrieved from GPM-DPR, however, have known biases due to measurement uncertainties and microphysical assumptions (e.g., Chase et al. 2020).
To address these shortcomings and leverage the utility of dual-frequency radar measurements from space, numerous studies have evaluated how the dual-frequency ratio (DFR), defined as the ratio of the radar reflectivity factor (mm6 m−3) between two wavelengths, is impacted by various microphysical properties. The mass-weighted mean diameter (Dm, defined here as the mean particle diameter after weighting by the ice water contribution for each size bin; e.g., Matrosov 1998; Liao et al. 2016; Duffy et al. 2021), aspect ratio (Matrosov et al. 2005, 2019), and density or degree of riming (e.g., Mason et al. 2019; H. Li et al. 2020; Yu et al. 2021) are known to influence the DFR. Triple-frequency studies have furthered DFR research by investigating how microphysical properties such as Dm and effective bulk density (ρe) relate to DFR between Ku and Ka bands (DFRKu-Ka) and between Ka and W bands (DFRKa-W), or similar combinations of DFR (e.g., Chase et al. 2018; Yin et al. 2017; Tridon et al. 2019; Nguyen et al. 2022). Scattering models of dendrites and unrimed aggregates are shown to occupy specific regions of the DFR phase space compared to spherical and soft spheroid models (e.g., Kneifel et al. 2011; Leinonen et al. 2012; Kulie et al. 2014; Leinonen and Moisseev 2015; Tyynelä and von Lerber 2019), while Kneifel et al. (2015) and other studies have used complimentary in situ observations to show that Dm increases with increasing DFR while higher ρe occurs for lower DFRKu-Ka (or DFRX-Ka).
Dual-frequency measurements are useful for retrieving precipitation properties such as parameters of the PSD, Dm, ice water content (IWC), and snowfall rate. They range in complexity from empirical relationships or lookup tables (e.g., Liao et al. 2016; Ni et al. 2019) to optimal estimation methods (e.g., Grecu et al. 2004, 2011; Leinonen et al. 2018; Tridon et al. 2019; Mroz et al. 2021) and neural networks (Sekelsky et al. 1999; Chase et al. 2021). These retrievals are typically evaluated using in situ measurements collected from field campaigns and are a convenient way of examining the vertical structure of retrieved microphysical properties where in situ measurements are not available.
To further understand the mechanisms associated with the formation, organization, and evolution of banded precipitation structures, the Investigation of Microphysics and Precipitation for Atlantic Coast-Threatening Snowstorms (IMPACTS; McMurdie et al. 2022) is a NASA Earth Venture Suborbital-3 (EVS-3) mission and the first major project in over 30 years to focus on the precipitation processes associated with East Coast winter cyclones. These cyclones often exhibit singular or multiple snowbands (Novak et al. 2004; Ganetis et al. 2018) and can create large snowfall accumulation gradients (e.g., Picca et al. 2014). Measurements of the physical structure of wintertime midlatitude snowstorms at multiple frequencies (X, Ku, Ka, and W bands) together with in situ microphysical measurements pose a unique opportunity to further explore the relationship between ice phase hydrometeors and remote sensing measurements such as DFR. This paper investigates what multifrequency radar measurements can tell us about the microphysics in snowstorms observed during IMPACTS, and how these observations are related to banded precipitation structures. This study develops a methodology for identifying regions of locally enhanced Ku- and Ka-band DFR along in-cloud flight legs and compares in situ microphysical observations within and outside of enhanced DFR regions in clouds (section 3a). Radar measurements from X and W bands are also examined in order to relate our results to past multifrequency radar studies (section 3b). Further, radar retrievals employing a neural network (NN) are used to evaluate how microphysical properties in other regions of the cloud are related to the DFR (section 3c).
2. Data and methodology
a. Radar data
The NASA ER-2 aircraft carried a suite of satellite-simulating passive and active remote sensing instruments (McMurdie et al. 2022). Flying at approximately 20 km above mean sea level (MSL), the radars on board the ER-2 measured the precipitation structure of winter cyclones as the aircraft typically conducted flight legs in a back-and-forth (“racetrack”) pattern or two longer legs at varying orientations resembling a bowtie pattern.
Three radars spanning four (X-, Ku-, Ka-, and W-band) frequencies were pointed at nadir to provide a cross-sectional view of the cloud and precipitation echoes. The ER-2 X-band (9.6 GHz) Radar (EXRAD; G. M. Heymsfield et al. 1996, 2020) consisted of a nadir-pointing beam as well as a conically scanning beam, with only the nadir beam being used for the purposes of this study. The High-Altitude Imaging Wind and Rain Airborne Profiler (HIWRAP; L. Li et al. 2015, 2020) measured the radar backscatter at the Ku- (13.9 GHz) and Ka-band (35.6 GHz) frequencies, while the Cloud Radar System (CRS; McLinden et al. 2021, 2020) provided highly sensitive measurements especially near cloud top at W-band (94.2 GHz) frequency. The HIWRAP data were sampled at 2 Hz with a vertical sampling of 26 m while the EXRAD and CRS data were sampled at 4 Hz with a vertical sampling of 19 and 26 m, respectively.
A series of corrections were made to the radar reflectivity (Ze) to ensure accurate calculation of the DFR. Internal and external calibration methods (McLinden et al. 2021) mitigated erroneous measurements of Ze, with external calibration involving measurements of the ocean normalized radar cross section and intercomparison among the radars. A 1-sigma noise filter was applied to the data to remove pixels primarily above cloud top (Zagrodnik et al. 2019; McLinden et al. 2021). Reflectivity near the echo top was compared between the Ku, Ka, and W bands, and DFR uncertainties for Ku–Ka and Ka–W were estimated to be 0.75 and 1.75 dB, respectively. Because the intrinsic unattenuated DFR is exclusively a function of non-Rayleigh scattering, attempts were made to correct for attenuation effects in Ze as in Chase et al. (2018) so that comparisons with past DFR studies could be made. The two-way path-integrated attenuation (PIA) was computed for water vapor and atmospheric gases following International Telecommunication Union (2013) and using water vapor density interpolated to each radar gate from hourly High-Resolution Rapid Refresh (HRRR; Blaylock et al. 2017) analysis data. The PIA correction at Ku, Ka, and W bands closer to the surface were on the order of 0.1, 0.3, and 0.5 dB, respectively. The W-band Ze was also corrected for attenuation from liquid water content (LWC) as well as extinction from ice scattering. Computing the specific attenuation caused by supercooled liquid water employed use of the coefficients from Meneghini and Kozu (1990) and LWC values (cloud water and rainwater content) interpolated to each radar gate from the HRRR analysis, with PIA corrections generally less than 0.3 dB above the melting layer. The use of the HRRR analysis was favored over radiosondes and in situ measurements of LWC due to its vertical and horizontal coverage where ER-2 radar measurements were made and because of the good agreement exhibited in the temperature and dewpoint profiles where sounding data were available. The relationship between extinction and Ku-band Ze (ZKu) from Kulie et al. (2014) was used to correct the W-band Ze due to ice attenuation effects, with specific attenuation values > 5 dB km−1 possible for ZKu > 30 dB.
b. Microphysics data
The NASA P-3 aircraft housed a suite of instrumentation that collected microphysics and thermodynamic data and flew within the cloud in coordination with the ER-2. The P-3 in situ measurements were often made in “stacked” legs under the ER-2 flight tracks spanning a range of altitudes and temperatures.
In situ microphysical properties such as the sizes, shapes, and distribution of particle sizes were obtained from optical array probes (OAPs). The SPEC, Inc., 2D-Stereo (2D-S; Lawson et al. 2006) and High Volume Precipitation Spectrometer (HVPS; Lawson et al. 1998) produced silhouetted images of individual particles by combining a high-powered laser and array of photodiode receptors with fast-responsive electronics. The University of Illinois–Oklahoma OAP Processing Software (UIOOPS; McFarquhar et al. 2018) was used to provide the particle size and morphological (e.g., area ratio, habit) features needed to derive estimates of the PSD and bulk properties for every second of flight. All 1-Hz PSDs where the 2D-S sampled <30% of the 1-Hz period were removed to ensure sufficient sampling statistics and then averaged every 5 s. A 5-s averaging interval was chosen as it provided better sampling statistics of larger particles (Hallett 2003; McFarquhar et al. 2007) and because the horizontal resolution was roughly in between the typical surface footprint of the Ku-band (1050 m) and Ka-band (420 m, assuming an ER-2 airspeed of 200 m s−1) beams from the HIWRAP radar. Finlon et al. (2020) detail the processing techniques employed, including the criteria used to identify particles for inclusion in the PSD. Particles with a maximum dimension between 100 μm and 1.4 mm for the 2D-S and between 1.4 mm and 3 cm for the HVPS were used for the composite N(D) for four of the five flights used in this study. The lower size limit was chosen such that uncertainties in the N(D) that resulted from ambiguities in the probe’s depth of field were minimized (Baumgardner and Korolev 1997; Jackson et al. 2012). The 1.4 mm probe cutoff was found to minimize the N(D) ratio between the 2D-S and HVPS for all 5-s samples with temperature T < 0°C among four of the five coordinated flights. An issue with the 2D-S necessitated exclusive use of the HVPS for the 25 February 2020 flight, with the N(D) and bulk microphysical properties estimated for D ≥ 0.4 mm. Given that the bulk properties described below are primarily impacted by the larger particles, comparison of microphysical observations between 25 February and the other flights remains appropriate for the purposes of this study.
c. Construction of collocated datasets
Sampling the same or similar region of the cloud from remote sensing and microphysical perspectives was achieved through flight coordination of a cloud-penetrating P-3 aircraft and the ER-2 satellite-simulating platform above the storm. Use of a radar matching algorithm based on that described in Chase et al. (2018) and Ding et al. (2020) permitted calculation of the radar Ze in the vicinity of the P-3 aircraft for each 5-s in situ PSD during flight. The 30 nearest neighbor radar gates within 4 km of the P-3 location were efficiently mapped using a k–d tree searching technique (Oliphant 2007) and a Barnes (1964) interpolation technique applied to these gates to compute a spatially weighted Ze for each 5-s collocated point. To address hydrometeor transport over longer time scales and maintain representativeness between the two datasets, the matched Ze had to occur within 3 min temporally when considering different airspeeds between the aircraft.
The aforementioned distance criteria differ from previous studies (Chase et al. 2018; Finlon et al. 2019; Ding et al. 2020) due to a more limited spatial coverage from the nadir-pointing radars and occasional misalignment between the ER-2 and P-3 aircraft. An incremental spatial autocorrelation analysis using the global Moran’s I autocorrelation index (Moran 1950) was conducted on the Dm derived from the PSDs (in situ perspective) and the matched ZKu (remote sensing perspective) to determine the optimal distance threshold that was not too restrictive while ensuring confidence in the correlation between radar and microphysical properties. The Moran’s I test, which evaluates how similar values of a variable are within a defined distance, was performed with 999 permutations on a range of distance thresholds varying from 2 to 10 km in increments of 0.5 km for each flight leg and the z score determined to provide a standardized comparison of significance by distance. The Moran’s I z score for Dm and ZKu is shown as a function of distance for a coordinated flight leg on 25 January 2020 in Figs. 1a and 1b. While all local maxima in the z score represent distances where spatial processes promoting clustering are most pronounced, the first local maxima is considered the optimal distance threshold (Figs. 1a,b, blue diamond) as it indicates the smallest distance where significant clustering occurs (Jossart et al. 2020). The mean optimal distance threshold among all flight legs used in this study, signified by the first peak in the z score for each leg, was 4.2 and 4.6 km for Dm and ZKu, respectively. A distance threshold of 4 km was chosen for this study to represent a more restrictive constraint for relating microphysical observations to the radar measurements.
Moran’s I z score of (a) Dm and (b) ZKu as a function of the distance threshold for the 2031–2040 UTC 25 Jan flight leg (Table 1), with the optimal distance threshold denoted as a blue diamond. (c) Cumulative frequency of the distance between the P-3 and ER-2 aircrafts among all coordinated observations.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
To mitigate contamination of the radar gates impacted by the direct sampling of the P-3 aircraft itself, Ze associated with the top 5% of spectrum width values from each flight leg where a P-3 signal existed were ignored as they were found to have local maxima in the backscatter signal compared to the ambient cloud. Matched Ze were also masked if the standard deviation among nearby pixels was >5 dBZe to reduce the likelihood of including observations affected by signal noise or weak signal, particularly near cloud edges. Among the 2328 coincident points with valid radar gates within 4 km spatially and 3 min temporally from the P-3 aircraft, 67%, 81%, and 91% of the observations had radar gates that were at most 1, 2, and 3 km from the P-3 location (Fig. 1c) while 33% and 66% of the observations occurred when the ER-2 and P-3 aircraft were within 1 and 2 min of each other. Table 1 details the flight segments used, number of coincident observations, and meteorological context for each flight considered in this study.
Summary of events, time segments, and number of collocated observations where coordinated flight operations were conducted. Asterisks denote UTC date + 1.
d. Identifying regions of enhanced DFR
To investigate whether the DFR of certain thresholds were related to precipitation structures, a variable DFR threshold was used. The technique used to determine the variable threshold is analogous to Wang et al. (2020), which examined the prominence in the along-track Ze to identify generating cells near cloud top. In this study, the radar measured DFRKu-Ka was matched to the P-3 location and all peaks in the along-track DFR signal with prominence ≥ 2 dB were first considered using the SciPy signal processing module (https://docs.scipy.org/doc/scipy/reference/signal.html). More prominent peaks signified segments of the cloud where the DFR signal stood out more compared to surrounding regions at the same altitude. A sensitivity analysis using different prominent thresholds revealed either too many or too few peaks in the along-track DFR upon visual inspection of the DFR cross sections. Observations on either side of each DFR peak (the peak width) were also considered as part of the enhanced DFR region if the DFR was at least 40% of the peak prominence (the relative peak height). Another sensitivity analysis using other percentages of the peak prominence yielded regions of enhanced DFR that were either too wide and potentially encompassed smaller neighboring peaks in the along-track DFR or too narrow and unable to adequately capture the entire increase then decrease in the along-track DFR.
3. Results
a. Regions of enhanced DFR at flight level
This subsection highlights the microphysical properties observed both within and outside of enhanced DFR regions as observed at the P-3 flight level. Analysis is first presented in the context of two flight legs followed by a statistical analysis among all coordinated flights.
1) 7 February case
The 7 February 2020 event featured a rapidly deepening cyclone affecting portions of the mid-Atlantic and New England. The merger of two jet streaks resulted in a strong upper-level jet with 250-hPa level winds exceeding 100 m s−1 (Fig. 2a) along the East Coast of the United States. A deep, large-scale, and negatively tilted trough with its axis positioned over eastern Michigan, Ohio, and West Virginia provided the cyclonic vorticity advection needed to deepen the cyclone (Fig. 2b). Warm air advection located over eastern New York, northern New Jersey, and New England at 850 hPa resulted in T > 0°C in many of those areas (Fig. 2c). Frontogenesis was present in the regions of warm air advection as well as over southern and central New York. The sub-980-hPa surface cyclone and associated frontal structure was complex (Fig. 2d), with a surface cold front propagating eastward toward the Atlantic Ocean and a stationary boundary extending northeast over Long Island and Cape Cod. Meanwhile, a secondary cold front from the north was moving southward over northern New York.
ERA5 analysis of (a) 250-hPa wind speed (shaded) and streamlines, (b) 500-hPa vorticity (shaded) and geopotential height (dm, black contours), (c) 850-hPa temperature advection (shaded), 0°C isotherm, and frontogenesis in 2 K (100 km)−1 (3 h)−1 increments, and (d) 2-m temperature (shaded), mean sea level pressure (hPa, black contours), and analyzed fronts adapted from the Weather Prediction Center (https://www.wpc.ncep.noaa.gov/) valid at 1500 UTC 7 Feb 2020. Solid red box in (a) denotes inset region for (c) and (d) and dashed red box indicates region for Fig. 3.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
The ER-2 and P-3 aircraft flew in coordinated, west-to-east-oriented flight legs with in situ data collected from constant-altitude passes ranging from 3.3 to 5.2 km MSL. The flight leg discussed below (1556–1613 UTC) consisted of collocated measurements from roughly Syracuse to Saratoga Springs, New York (Fig. 3). Infrared (IR) brightness temperature from band 13 (10.3 μm wavelength) of the GOES-16 mesoscale sector data (Brodzik 2020a) shows that the western three-quarters of the flight leg contained colder cloud tops (Fig. 3a) and were associated with deeper clouds from the ER-2 radar cross sections (Fig. 4). The exception was a linear feature with warmer cloud tops around 75.75°W that corresponded to a narrow region of lower echo-top heights from the radar cross sections around 30 km from the westernmost end of the cross section.
View of (a) infrared (IR) brightness temperature from band 13 of the GOES-16 mesoscale sector 1 and (b) 2 km MSL mosaic of Ze from NEXRAD radars valid at 1603 UTC 7 Feb 2020. The ER-2 flight track from 1556 to 1613 UTC is shown as a black line, and the magenta ellipse in (b) denotes a region of enhanced Ze objectively identified from the Ganetis et al. (2018) technique.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
HIWRAP cross sections of (a) ZKu, (b) ZKa, and (c) dual-frequency ratio between Ku and Ka band (DFRKu-Ka) for the 1508–1613 UTC 7 Feb 2020 flight leg (black line in Fig. 3). The purple line here denotes the P-3 flight track where collocated measurements were made, and the blue box in (c) is the region of prominently higher DFRKu-Ka determined from the methodology in section 2d. The magenta bars at 2 km MSL correspond to the bounded region of higher reflectivity in Fig. 3b and the dotted contours in (c) are isotherms in 5°C increments. The purple diamonds labeled I–III denote the regions highlighted in Figs. 5 and 6.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
The 2-km gridded Next Generation Weather Radar (NEXRAD) reflectivity mosaic (Brodzik 2020b) shows a linear region of enhanced Ze reminiscent of a banded precipitation structure near the middle of the ER-2 flight leg (75°W) as well as a radar bright band, defined by the correlation coefficient ρhv < 0.98 (not shown), near the east end of the flight track (Fig. 3b). The western feature of interest was collocated with a local minimum in cloud-top brightness temperature and consistent with a greater cloud depth. The existence of and properties associated with banded precipitation structures motivated the use of an automated approach to identify such features from the 2-km reflectivity mosaic. Similar to Ganetis et al. (2018), regions of contiguous radar pixels containing Ze above the 90th percentile of values in the domain and ρhv ≥ 0.98 were identified and an ellipse fit to each region using the Scikit-image measure module (https://scikit-image.org/docs/dev/api/skimage.measure.html). One of the bounded regions, shown as an ellipse in Fig. 3b, contained a width (128 km) that was too broad to be considered a snowband following the definition outlined in Novak et al. (2004) and Ganetis et al. (2018). Nonetheless, the linear region of higher Ze near the center of the ER-2 flight leg appeared to influence the multifrequency radar observations shown in Fig. 4 and discussed below.
Cross sections of ZKu, ZKa, and the DFR between Ku and Ka band (DFRKu-Ka) from the HIWRAP radar highlight the fine-scale precipitation structures observed for the 1556–1613 UTC ER-2 flight leg in Fig. 4. Radar echoes near cloud top (7–8 km MSL, consistent with lidar measurements of cloud-top altitude) resembled that of generating cells, particularly at higher frequencies (Fig. 4b), as ZKu and ZKa increased at varying rates with respect to depth below cloud top. This resulted in increasing DFR farther below cloud top, with DFR = 5 dB first observed between −15° and −10°C (3–5 km MSL) across most of the flight leg (Fig. 4c). The blue box encompassing the P-3 flight track in Fig. 4c denotes a segment along the P-3 track where prominently higher DFR (section 2d) was observed.
Figure 5a shows the DFRKu-Ka matched to the P-3 location (dotted curve) corresponding to the purple line in Fig. 4. Within this flight segment was an 8-km region where the along-track DFR was at least 40% of the peak prominence of 6.9 dB. Although the precipitation feature associated with this DFR peak may have moved in the two minutes it took for both aircraft to observe the same feature, causing a slight “lag” in the PSD-estimated DFR signal in this region, the Ku- and Ka-band DFR estimated from the PSDs (section 2a) correlated well with the DFR values measured by HIWRAP for this flight leg with a Pearson correlation coefficient r of 0.88. Values of Dm directly west and east of the enhanced DFR region were around 3 mm and peaked to around 7 mm within the region of enhanced DFR (Fig. 5b). Larger Dm within this enhanced DFR region were correlated with the presence of particles exceeding 2 cm and a reduction in ρe.
Along-track (a) DFRKu-Ka estimated from the in situ PSD (solid) and matched from HIWRAP measurements (dotted), (b) number distribution function N(D) (shaded), mass-weighted mean diameter Dm (solid) and effective bulk density ρe (dotted), (c) liquid-equivalent normalized intercept parameter Nw, and (d) ice water content IWC for the P-3 segment shown in Fig. 4. The criteria used to classify regions of enhanced DFR are outlined in (a).
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
Figure 6 qualitatively shows the differences in particle habits and sizes from the HVPS within and outside of the enhanced DFR region (vertical purple lines in Fig. 5). Particles within the region of enhanced DFR (Fig. 6b; labeled II in Fig. 5) were visually larger, with 35% of particles for D ≥ 1.4 mm constituting of dendrites and aggregates from the UIOOPS habit classification. In contrast, regions to the west and east denoted by I and III in Fig. 5 contained particles that are visually smaller with only 9% of particles larger than 1.4 mm classified as dendrites or aggregates (Figs. 6a,c). The presence of aggregation within the enhanced DFR region was further evident by a reduction in the N(D) in the 1.5–5 mm size range and in the Nw (Figs. 5b,c). While the higher Dm and particle images support the evidence of larger aggregates within the region of enhanced DFR, IWC was reduced within this region (Fig. 5d). Lower values of IWC were possibly explained by the lower ρe and Nw observed with larger, less dense particles that likely contributed a greater fraction to the bulk ice mass. The mass per unit volume may have also been locally lower in regions where the larger particles exhibited greater fall speeds and were therefore less concentrated within the sample volume. Lower values of IWC along other segments of the flight leg (e.g., at 87 km) suggested that the correlation between IWC and the DFR may not have been mutually dependent. Further discussion regarding this relationship is found in section 3a(3).
Representative particle images from the HVPS corresponding to the region labeled (a) I, (b) II, and (c) III in Figs. 4 and 5.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
2) 5 February case
The 5 February 2020 event involved a developing cyclone over the Lower Mississippi valley with precipitation occurring in the Midwest United States. A digging positively tilted trough traversed eastward across the Rocky Mountains and by 2100 UTC 5 February it featured two distinct shortwaves over west Texas and Nebraska (Figs. 7a,b). Although the center of surface low pressure was located well south of where science operations were conducted, a warm front overrunning preexisting cold air over portions of Missouri, Illinois, and Indiana provided advection of warm air aloft and localized regions of frontogenesis (Fig. 7c). The 1002-hPa surface low was situated over northeast Louisiana at 2100 UTC, with a surface temperature gradient focused along the surface warm front extending over the Tennessee–Kentucky border (Fig. 7d).
As in Fig. 2, but valid at 2100 UTC 5 Feb 2020. Dashed red box in (a) indicates region for Fig. 8.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
GOES-16 mesoscale sector IR brightness temperature shows slightly warmer cloud tops along the northwestern quarter of the ER-2 flight track corresponding to areas of lower Ze from the NEXRAD reflectivity mosaic (Fig. 8; Brodzik 2020c). Two linear areas of enhanced Ze, denoted by the solid and dashed ellipses in Fig. 8b, met the definition of snowbands along the 2145–2209 UTC ER-2 flight track. These regions also corresponded to larger ZKu and DFRKu-Ka in the lowest 3 km in Figs. 9a and 9c. Precipitation within a region of taller cloud tops 75–100 km from the start of the flight leg appeared to aid in the Ze enhancement at S and Ku bands, whereas non-Rayleigh scattering at Ka band (Fig. 9b) contributed to the two regions of higher DFRKu-Ka.
As in Fig. 3, but valid at 2157 UTC 5 Feb. The ER-2 flight track from 2145 to 2209 UTC is shown as a black line, and the magenta ellipses are the identified snowbands.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
As in Fig. 4, but for the 2145–2209 UTC 5 Feb 2020 flight leg (black line in Fig. 8). Magenta ellipses correspond to the snowbands highlighted in Fig. 8.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
As the P-3 aircraft sampled the snowbands at an altitude of 3.3 km (−7° ≤ T ≤ −3°C), a series of six enhanced DFR regions were present at the P-3 flight level (Fig. 9c). Similar to the 7 February flight leg [section 3a(1)], the larger signatures in DFRKu-Ka originated above the aircraft within the −15° ≤ T ≤ −10°C layer. The width of the enhanced DFR regions ranged between 4.9 and 10.4 km, with the mean width of 7.2 km and was similar to the 8-km-wide enhanced DFR region observed in the 7 February flight leg.
Figure 10 provides a microphysical analysis for a segment of the P-3 flight track. The DFR estimated by the PSDs correlated well (r = 0.79) with that measured by HIWRAP at the same location. Within the enhanced DFR regions, larger Dm of up to 5 mm coincided with the presence of larger particles in the N(D). In addition, the ρe was either low (<0.03 g cm−3) or decreased within each region. Figure 11 contains representative particle images from the HVPS that illustrate the visual differences in the sizes and habits within (labeled II in Fig. 10) and outside of (I and III in Fig. 10) the enhanced DFR regions. Habit classification of HVPS images >1.4 mm determined that 33% were either dendrites or aggregates within the enhanced DFR region and only 13% for regions I and III outside of the enhanced DFR region. Although inhomogeneities in the precipitation structure from the NEXRAD plan view and HIWRAP cross section (Figs. 8 and 9) can explain the fluctuation in the Nw through this segment of the cloud, values of Nw decreased once the P-3 penetrated each region of enhanced DFR (Fig. 10c). While the inhomogeneities in the precipitation structures may explain some of the IWC variability at the P-3 altitude, there exists a lack of consistent relative change in the IWC as the P-3 enters each region of enhanced DFR (Fig. 10d).
As in Fig. 5, but for the P-3 flight segment outlined in Fig. 9.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
As in Fig. 6, but for the regions I–III labeled in Figs. 9 and 10.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
3) Statistical analysis
To evaluate differences in the microphysical properties within and outside of regions of enhanced DFR at the P-3 flight level, PSDs and derived bulk microphysical properties were partitioned into these two categories for each coordinated flight. Figure 12 provides the mean N(D), represented by the average N(D) among the n 5-s periods as indicated in the legend of each panel, for observations within the enhanced DFR region (solid black lines) and for the remaining periods (dashed black lines). Among all flights, the mean PSD is broader and there is a greater concentration of larger particles within regions of enhanced DFR (Fig. 12f). Differences in the N(D) among observations within and outside of the enhanced DFR regions were most notable for the 5, 7, and 25 February events where measurements were primarily made in the northwest sector of the cyclones (Table 1). Here, the mean N(D) within the enhanced DFR regions was at least one order of magnitude greater than the remaining observations for D ≥ 1 cm (Figs. 12c–e). Less notable differences in the PSDs between the partitioned groups for the 25 January and 1 February events (Figs. 12a,b) are explained by the sampling strategy performed and the meteorological context. Measurements were made across a warm occluded front for the 25 January event while the P-3 sampled a developing warm oceanic frontal system over the Atlantic Ocean on 1 February.
Mean number distribution function N(D) within (solid black) and outside of enhanced DFR (dashed black) for the (a) 25 Jan, (b) 1 Feb, (c) 5 Feb, (d) 7 Feb, and (e) 25 Feb 2020 events and for (f) all coordinated flights. Gamma distributions fit to the mean N(D) are shown as blue lines. Number of 5-s collocated observations for each partitioned category are given in the legend.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
Comparison of the intercept N0, shape μ, and slope Λ gamma parameters fit to the mean N(D) for within (outside of) regions of enhanced DFR for coordinated flights where 2D-S and HVPS data are available.
The relationship between a lower μ within regions of enhanced DFR and larger Dm is further supported by previous studies within ice clouds. Borque et al. (2019) show μ decreasing with larger Dm for ice and snow PSDs during the GPM Cold-Season Precipitation Experiment (GCPEx), with μ approaching −1 for a Dm of 3.5 mm. Williams et al. (2014) show a similar relationship between μ and liquid-equivalent Dm based on disdrometer measurements.
Figure 13 shows distributions of HIWRAP ZKu, DFRKu-Ka, Dm, ρe, log10(Nw), and IWC for each of the five flights for regions within (solid) and outside of (dashed) enhanced DFR. The middle line is the median, the box edges are the 25th and 75th percentiles, and the whiskers denote 1.5 times the interquartile range (IQR) from the 25th and 75th percentiles. Statistics for each bulk microphysical property (Table 3) compliment the distributions shown in Figs. 13c–f. The PDmed metric from the table follows Eq. (4), where XDFR and Xother represent the median value from the solid and dashed boxplots (Fig. 13), respectively, for each bulk variable. A Mann–Whitney U test performed on the distributions within and outside of the enhanced DFR regions for each bulk property are also listed in the table, and its p value p computed to evaluate whether the two distributions were statistically different at the 1% confidence level (bold values).
Boxplots of (a) HIWRAP ZKu, (b) DFRKu-Ka, (c) Dm, (d) ρe, (e) log10(Nw), and (f) IWC for regions within (solid) and outside of (dashed) enhanced DFR for each coordinated flight. The middle line denotes the median, the box edges are the 25th and 75th percentiles, and the whiskers signify 1.5 times the interquartile range (IQR) from the 25th and 75th percentiles.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
Median percent difference (PDmed) and p value from a Mann–Whitney U test (parentheses) for Dm, ρe, log10(Nw), and IWC distributions within vs outside of enhanced DFR regions for each coordinated flight. Boldface p values (<0.01) suggest the two distributions are statistically different from one another.
Boxplots of Dm (Fig. 13c) and values of PDmed (Table 3) indicate that Dm is larger within regions of enhanced DFR for all coordinated flights. Differences in the Dm distributions are most notable for the 5, 7, and 25 February events where p < 0.001 suggests a statistically notable difference in Dm for regions within and outside of enhanced DFR. While the median Dm within enhanced DFR regions on 25 January and 1 February were about 25% larger compared to the remaining observations on those days, the Dm distributions were more similar than the other flights where the N(D) varied considerably within and outside of the enhanced DFR regions (Figs. 12c–e) and the DFRKu-Ka was notably higher within regions of enhanced DFR (Fig. 13b).
Like Dm, values of ρe and Nw within enhanced DFR regions were most different for the 5, 7, and 25 February events (Figs. 13d,e; Table 3). Larger Dm and lower ρe and Nw for these events were characterized by PSDs with a lower intercept parameter and consisted of larger aggregates that were less dense than PSDs observed outside of regions with enhanced DFR. Although median values of IWC were marginally greater for four of the five events (Fig. 13f), the p values provided in Table 3 and the relationship between IWC and DFR for the two flight legs discussed in sections 3a(1) and 3a(2) do not suggest a clear association of large IWC within enhanced DFR regions for the cases examined. The negative correlation between Dm and ρe, Nw illustrates the impact that the PSD shape, types of habits, and degree of riming (or lack thereof) have on the IWC.
b. Multifrequency results
While partitioning observations into regions within and outside of enhanced DFR is useful for evaluating microphysical differences associated with specific precipitation structures, there exists utility in understanding how the microphysical properties relate to the magnitude of DFR. The discussion below on DFR–Dm relationships and microphysical properties within a triple-frequency framework aim to relate observations from IMPACTS to past field studies.
Figure 14 shows the mean ρe as a function of DFR measured by radar and Dm for four frequency pairings: Ku–Ka, Ka–W, X–Ka, and X–W. The 50th, 75th, 90th, and 95th percentiles in DFR for each frequency pairing provide context on the distribution spread and where the more extreme values reside. The negative correlation between Dm and ρe irrespective of the frequency pairing in DFR is consistent with PSDs containing larger, less dense aggregates, particularly from observations within enhanced DFR regions (Figs. 13c,d). A notable range in DFR was observed for low Dm, particularly for Figs. 14b and 14d where small aggregates deviate from Rayleigh scattering first at the highest frequency (e.g., W band). Larger frequency differences in the DFR pairing yielded a greater range in the DFR since larger aggregates are less likely to deviate from Rayleigh scattering at lower frequencies (e.g., X band), and explains why the DFRX-W approached 15.7 dB compared to 9.1 dB for DFRKa-W (P95 in Figs. 14b,d). These findings are put in the context of Sekelsky et al. (1999), who indicated that the lower frequency in the DFR pairing impacted the maximum detectable D within a PSD, and that particle sizes of 0.2 ≤ D ≤ 5 mm could be unambiguously detected between Ka and W bands. Their finding is reflected in the DFRKa-W–Dm relationship (Fig. 14b), where the DFRKa-W began to level off for a Dm ≅ 5 mm before slightly decreasing. Put another way, the DFRKa-W signal saturates around 8 dB (90th percentile) which is similar to the findings of Stein et al. (2015) and Ori et al. (2020), who noted DFRKa-W reached a limiting value between 7.5 and 10 dB based on multifrequency radar measurements.
Mean ρe as a function of Dm (both derived from PSDs using a variable m–D relation based on LS15) and the radar observed (a) DFRKu-Ka, (b) DFRKa-W, (c) DFRX-Ka, and (d) DFRX-W. Bins are in 1 dB × 0.25 mm increments and the 50th, 75th, 90th, and 95th percentiles are represented by the horizontal lines. Bins with fewer than five observations are ignored.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
Microphysical observations within a triple-frequency framework are presented in Fig. 15 to evaluate similarities and differences in particle properties between midlatitude winter snowstorms sampled during IMPACTS and past studies. Among all collocated measurements in this study, the majority of observations at lower temperatures (T < −15°C) occurred where DFRKu-Ka < 1.5 dB while higher T were found for larger DFRKu-Ka and DFRKa-W (Fig. 15a). The visual clustering of observations into multiple regions within the DFR plane suggests distinct scattering regimes influenced by different particle habits (Kulie et al. 2014) or microphysical processes such as aggregation and riming. Scattering curves from LS15 using an exponential size distribution with varying inverse scale parameter for three degrees of riming are shown for context, with observations to the left of the 0.0 kg m−2 curve (solid black line) consistent with scattering models of dendrites and unrimed aggregates (e.g., Petty and Huang 2010; Leinonen and Moisseev 2015). Here, DFRKa-W increased and then decreased while DFRKu-Ka increased monotonically with increasing Dm (aka a hook signature).
(a) Scatterplot of temperature as a function of the radar DFRKu-Ka (y axis) and DFRKa-W (x axis) with scattering curves from LS15 at effective liquid water paths (ELWP) of 0.0, 0.2, and 0.5 kg m−2. (b)–(d) Two-dimensional histograms of (b) number of observations, (c) mean Dm, and (d) mean ρe for the same DFR phase space as in (a) with 1 dB bin increments. Bins with fewer than five observations are ignored.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
Figures 15b–d show the number of collocated measurements for 1-dB bins containing at least 5 observations as well as the mean Dm and mean ρe calculated from the PSDs. Nearly half of the observations fell within DFRKu-Ka ≤ 2 dB and DFRKa-W ≤ 5 dB (Figs. 14a,b and 15b). The mean Dm was at a minimum near the origin of the DFR plane, where temperatures were lower, and increased to values approaching 8 mm for the DFR bin corresponding to a DFRKu-Ka of 9–10 dB and a DFRKa-W of 5–6 dB (Fig. 15c). DFR bins with lower ρe (Fig. 15d) resided where Dm was largest, consistent with the presence of aggregates.
Although the relationships between Dm, ρe, and DFR within the triple-frequency framework corroborate results from the Biogenic Aerosols Effects on Clouds and Climate campaign (BAECC; Kneifel et al. 2015) and the Olympic Mountains Experiment (OLYMPEX; Chase et al. 2018), there was a less pronounced hook signature corresponding to dendrite and unrimed aggregate scattering models (e.g., Petty and Huang 2010; Leinonen and Moisseev 2015). While the collocation technique, calibration of the radar Ze, measurement uncertainties, or the attenuation correction (section 2) may explain why a less robust hook signature was observed for the cases examined here, more observations are needed to better evaluate whether the degree of riming (reflected in the ρe observations in Fig. 15d) and the shape of the PSD explain the differences between the IMPACTS observations and past studies. Results from the Radar Snow Experiment (RadSnowExp; Wolde et al. 2019) attributed greater variability in the amount of riming to a hook signature that deviated from scattering models and past campaigns (Nguyen et al. 2022) while Mason et al. (2019) discussed how a broader PSD (smaller μ) can affect the distribution of DFR values and explain a less concave hook signature in the observed triple-frequency framework compared to scattering models employing an exponential PSD (μ = 0). The presence of a smaller μ within regions of enhanced DFR (Table 2) corroborates the findings of Mason et al. (2019) and illustrates the usefulness in evaluating the PSD shape with regard to the observed DFR.
c. Retrieved microphysical properties
One appealing reason to employ radar retrievals is to better understand the microphysical properties and their associated processes where in situ observations are not possible. The radar retrieval of Chase et al. (2021) was applied to the HIWRAP data in this study, with results shown within the northwest quadrant of a developing midlatitude cyclone and assessed within a triple-frequency framework across all coordinated flights.
The radar retrieval uses an NN with three inputs (T, ZKu, and DFRKu-Ka) and six hidden layers featuring eight neurons to compute the Nw (liquid equivalent), Dm, and liquid-equivalent Dm as well as an estimate of the IWC for each radar pixel. The datasets used to train and evaluate the retrieval consisted of airborne radar and in situ microphysical observations from the Midlatitude Continental Convective Clouds Experiment (MC3E; Jensen et al. 2016), GCPEX (Skofronick-Jackson et al. 2015), and OLYMPEX (Houze et al. 2017). Chase et al. (2021) provides more details of the retrieval methodology and evaluation for the aforementioned field observations.
1) 5 February case
The 2149:25–2158:27 UTC 5 February period is used to illustrate how the 2D retrieved properties from the NN relate to the in situ microphysical estimates from the P-3 aircraft (Fig. 16). Retrieved properties were masked below 1 km in case there were contaminated inputs from liquid hydrometeors at and below the melting layer or ground echo present. The top row of Fig. 16 shows the NN retrieved Dm, Nw, and IWC along the ER-2 flight leg, the middle row compares the retrieved properties to those computed from the PSD observations made from the P-3 aircraft, and the bottom row provides a vertical profile of the retrieved properties.
Cross sections of (a) observed DFRKu-Ka and retrieved (b) Dm, (c) Nw, and (d) IWC with the P-3 flight track in black for a segment of the ER-2 flight track in Fig. 9. Values of (e) DFR, (f) Dm, (g) Nw, and (h) IWC from in situ PSDs (black) and from the retrieval [blue, (f)–(h) only]. Vertical profiles of (i) DFR and retrieved (j) Dm, (k) Nw, and (l) IWC with the median as a black line and the IQR shaded red. Mean altitude corresponding to T = −20°, −15°, −10°, and −5°C is provided in (i)–(l).
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
Two broad regions of larger DFRKu-Ka, corresponding to an along-track distance of 80–110 and 120–165 km, are highlighted, and these regions are related to snowbands observed from the NEXRAD Ze (Fig. 8b). Within these regions, larger values of Dm approached 5.5 mm and lower values of Nw approached 103 m−3 mm−1 (Figs. 16b,c). The negative correlation between Dm and Nw is consistent with aggregation as suggested in Figs. 13c and 13e. IWC was also slightly greater in these regions, locally exceeding 0.5 g m−3 (Fig. 16d). The vertical profile of Dm (Nw) shows a steady increase (decrease) from cloud top to an altitude of 1.5 km, corresponding to a maximum in the HIWRAP DFR (Figs. 16i–k). Notable increases in the HIWRAP DFR around −15°C and particularly for T > −5°C corresponded with modest increases in Dm and decreases in Nw (Figs. 16i–k). The recent studies of Ori et al. (2020) and von Terzi et al. (2022) attributed the more efficient ice and snow sticking and enhanced aggregation in these temperature regions (around −15°C corresponding to the dendritic growth layer and close to the melting level) to increases in the DFR and Dm. Retrieved IWC increased from cloud top to about 500 m below the P-3 altitude (2.8 km) before it decreased slightly below this level (Fig. 16l). While the increase in IWC is consistent with particle growth through a combination of deposition and/or riming, lower IWC values below 2.8 km appear to be related to robust aggregation given the dramatic decrease in Nw and a continued increase in Dm.
Since the collocated datasets must be representative of each other for the NN to be properly evaluated, the DFR derived from the P-3 observations were compared to the matched DFR and found to have good correlation (r = 0.79; Figs. 10a and 16e). Overall, the retrieved properties correlate reasonably well with the PSD observations as the P-3 flew in and out of the fine-scale precipitation structures represented by the regions of enhanced DFR discussed in section 3a(2). In particular, the NN and the PSD observations of Dm and IWC were well correlated with r values of 0.74 and 0.82, respectively. Although the retrieval was able to appropriately capture the peaks in Dm along the P-3 flight leg, its underestimation may be explained in part by the representativeness of the PSDs and degree of riming used to train the NN compared to the types and sizes of hydrometeors observed during IMPACTS. This is further explained in the context of all coordinated flights below.
2) Retrieved properties in a triple-frequency framework
Retrieved Dm, Nw, and IWC corresponding to the collocated observations along the P-3 flight tracks for all coordinated flights are placed into the same radar DFR phase space as in Fig. 15 and shown in Fig. 17. Similar to the results of Kneifel et al. (2015) and Chase et al. (2018) as well as the IMPACTS analysis (Fig. 15), the retrieved Dm increased further from the origin of the DFR plane (Fig. 17a) with indications of a hook signature evident for Dm ≥ 5 mm. The minimum in retrieved Nw (Fig. 17c) resided in the same region of the DFR plane where Dm was largest, consistent with aggregation suggested in the P-3 observations (Figs. 13c,e). Conversely, smaller retrieved Dm and greater retrieved Nw for small DFRKu-Ka were related to higher ρe in the P-3 observations (Fig. 15d) and suggested that a higher concentration of smaller, more dense ice crystals, or possibly a greater degree of riming, was present in this region of the DFR plane. Last, mean values of retrieved IWC for each 1 dB bin did not show a clear relationship between IWC and either DFRKu-Ka or DFRKa-W. As discussed earlier, a combination of lower Nw, consistent with aggregation, and lower ρe, associated with dendrites and unrimed aggregates, may have limited the extent of IWC values. Further, the PD in the mean IWC between regions within and outside of enhanced DFR were found to be negligible among the cases examined (Fig. 13f).
Mean NN retrieved (a),(b) Dm, (c),(d) Nw, and (e),(f) IWC for the same DFR phase space in Fig. 15 for the (left) collocated observations and (right) all valid radar pixels based on cross sections from the coordinated flight legs. Bins with fewer than 5000 radar pixels are ignored for (b), (d), and (f).
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
To assess whether the retrieved Dm, Nw, and IWC along the P-3 flight track were representative of other regions of the cloud relative to the measured DFR, the retrieved properties for all valid radar data among the coordinated flight legs were also placed into a triple-frequency framework and shown in Fig. 17. Radar pixels possibly contaminated by ground echo (<1 km MSL), influenced by the melting layer (T ≥ 0°C), and impacted by nonphysical DFR values (e.g., near cloud top) due to slight misalignment of the radar beams or measurement calibration limitations (DFR < −1 dB) were ignored. Similar to the results among collocated observations (Figs. 17a,c,e), Dm increased and Nw decreased further from the origin of the DFR plane while the IWC did not show a clear relationship with DFRKu-Ka or DFRKa-W (Figs. 17b,d,f).
Since neural networks and other machine learning models are not guaranteed to follow expected, known relationships if they are overfit or underfit to the observations, the similarities between the observed and retrieved properties (Figs. 16 and 17) provide added confidence that the NN has learned physically plausible relationships. Furthermore, since the IMPACTS data were not included in the initial formulation of the NN retrieval the results in Figs. 16 and 17 are encouraging and show support for the application of the NN outside of the field campaigns it was built from. Despite the apparent low bias in the retrieved properties based on MPE values for the IMPACTS cases examined in this study, recent efforts comparing the retrieved liquid equivalent precipitation rate (R) from the Chase et al. (2021) NN to existing operational retrievals (e.g., 2C-SNOWPROFILE and 2A.DPR) suggest that the NN retrieved R is the largest in magnitude for snowstorms that GPM-DPR can sample (Chase et al. 2022). Future versions of the NN will aim to improve retrieval skill across more storm environments using PSDs from additional field projects such as IMPACTS.
4. Summary and conclusions
A novel approach to evaluate microphysical properties within regions of prominently higher dual-frequency ratio (DFR) between Ku and Ka bands for constant altitude transects through the cloud was constructed to determine how these properties and their associated processes were affected within and outside of precipitation structures. A collocated dataset containing particle size distributions (PSDs) and bulk properties from in situ measurements collected by the P-3 aircraft and airborne radar data spanning four frequencies (X, Ku, Ka, and W bands) from the ER-2 aircraft was generated to provide information every 5 s for five coordinated flights during the 2020 Investigation of Microphysics and Precipitation for Atlantic Coast-Threatening Snowstorms (IMPACTS) deployment. The novel approach permitted comparison of microphysical properties such as the mass-weighted mean diameter Dm, liquid-equivalent normalized intercept parameter Nw, effective bulk density ρe, and ice water content (IWC) within and outside of notable precipitation features such as snowbands. These results were also presented in DFR–Dm and triple-frequency frameworks to draw conclusions on relationships between the observed microphysical properties and the DFR for multiple frequency pairings. Last, a radar retrieval employing a neural network (NN) was run on all coordinated flight legs between the ER-2 and P-3 aircraft to analyze the retrieved properties throughout the cloud and to evaluate against the in situ observations.
The following are the key findings of this study:
-
Regions of enhanced DFR between Ku and Ka bands along the P-3 flight track had a mean width and standard deviation of 5.4 ± 3.6 km, the PSDs were characterized as broader with a smaller shape parameter, the Dm 58% larger, the ρe 37% lower, and the Nw 74% lower for the events studied.
-
The relationship between IWC and DFR was less prominent compared to other bulk microphysical properties, with the mean IWC within regions of enhanced DFR 0.9% lower than the other observations.
-
Microphysical properties mapped into a triple-frequency framework comprising of radar DFRKu-Ka and DFRKa-W corroborated the findings of past studies and scattering models of Dm that increased further from the origin of the DFR plane, while ρe increased as values of DFRKu-Ka approached 0 dB. A less robust hook signature was present in the IMPACTS analysis compared to some previous studies and scattering models of dendrites and unrimed aggregates.
-
An NN radar retrieval effectively resolved spatial differences in microphysical properties related to processes such as aggregation and precipitation structures manifested in the radar DFR at Ku and Ka bands. The retrieved properties were consistent with the relationships between DFR and the in situ microphysical observations despite a low bias that existed in the retrieved properties for the IMPACTS cases in this study.
This study illustrated that variability in the cloud microphysical properties can be linked to growth processes such as aggregation and ultimately shown in the multifrequency radar measurements as prominently higher DFR. These radar and microphysical signatures are occasionally the result of banded precipitation structures which can have implications for snowfall at the surface. The results herein qualitatively relate signatures in the DFR and in situ cloud microphysical properties to banded precipitation structures exhibiting higher reflectivity closer to the surface. Future work will require considering hydrometeor transport to evaluate these radar and microphysical properties within and outside of snowbands identified objectively from ground-based radars and to cast these findings more directly to nowcasting applications. The evaluation of a novel radar retrieval further extends the capabilities of estimating microphysical properties and inferring precipitation processes where in situ observations are not available within winter midlatitude cyclones.
While the current work reflects an effort to relate microphysical properties from observational and retrieval perspectives to notable precipitation features, incorporating wind synthesis from radar measurements and analyzing high resolution model simulations with output of particle trajectories may be useful to more definitively link precipitation structures at higher altitudes (e.g., fallstreaks) to radar signatures closer to the surface (e.g., snowbands). Planned IMPACTS deployments for the 2022 and 2023 winter seasons should involve measurements across a greater range of temperatures and storm structures to provide more insight into the findings presented here. Future work on extending the NN retrieval to examine global ice and mixed phase processes, including supercooled water and riming, has the potential to improve our understanding of microphysical processes associated with precipitation structures across a wider range of environments.
Acknowledgments.
This research was supported by the NASA Grant 80NSSC19K0338 awarded to the University of Washington. We thank all participants of the IMPACTS project for collecting the data used in this study. The feedback from three anonymous reviewers significantly improved aspects of the manuscript. This paper is dedicated to Dr. Gail Skofronick-Jackson, whose support for IMPACTS and other projects related to precipitation research have been instrumental in advancing the state of spaceborne measurements and ground validation.
Data availability statement.
The IMPACTS data (McMurdie et al. 2019; https://doi.org/10.5067/IMPACTS/DATA101) and the individual datasets cited within this paper can be found at the NASA Global Hydrology Resource Center’s DAAC. The processing routines to read the processed radar and OAP data are available in the data repository associated with this paper (https://github.com/joefinlon/Finlon_et_al_2021_DFR). The NN retrieval can be found with the repository associated with Chase et al. (2021; https://doi.org/10.13012/B2IDB-0791318_V2).
APPENDIX
Defining Parameters for Multiple Degrees of Riming
This study employs radar backscatter cross section σb and particle mass estimates from scattering lookup tables provided by LS15. While three riming scenarios were considered in that study, only aggregates that exclusively experienced the aggregation process prior to being exposed to the entire ELWP (model B; cf. section 2c) were considered. For each riming category and radar frequency (X, Ku, Ka, and W bands), a flexible function was developed using 1D interpolation to assign a σb for each size bin from the supplied PSD (solid curves in Figs. A1a,b). The a/b coefficients [Eq. (1)] were used for each riming category from LS15.
Backscatter cross section as a function of D at different effective liquid water paths (ELWP) from LS15 (solid curves) and interpolated for other ELWP (dotted curves) for (a) Ku- and (b) Ka-band frequencies. (c) Particle mass as a function of D for different mass–dimension relationships, including H10 and a modified version from BF95.
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
To better resolve the spatiotemporal variability in the amount of available supercooled water from in situ observations, additional ELWP values from those included in LS15 [0 (no riming), 0.1, 0.2, 0.5, 1.0, and 2.0 kg m−2] were considered. Since the techniques used to derive scattering properties can be computationally intensive, the σb and m–D coefficients from LS15 were linearly interpolated at an ELWP of 0.05, 0.15, 0.3, and 0.4 kg m−2 for each radar frequency and size bin D (dotted curves in Fig. A1). The interpolated curves preserve the relationship between σb as well as particle mass and ELWP. The black curves in Fig. A1c represent the commonly used m–D relationships from Heymsfield et al. (2010, H10 hereafter) and Brown and Francis [1995, BF95 hereafter; using the size conversion factor of Hogan et al. (2012)] for reference.
To illustrate the sensitivity of derived bulk properties to the estimated degree of riming present, Fig. A2 shows a time series of the best estimates of ELWP and IWC (section 2b) using the ELWP values prescribed in LS15 (blue) and with the inclusion of additional ELWP and m–D parameters (red). Considering additional degrees of riming permitted more sensitivity in the IWC while preserving the overall magnitudes. The additional σb and m–D parameters considered in the minimization procedure (section 2b) yielded an IWC that was 8.4% greater than the standard riming categories for the flight leg shown, and only 3.5% greater among all collocated observations. Even smaller differences were observed for other derived properties, with a mean PD between the two approaches <0.1% for Dm and 1.1% for DFRKu-Ka.
Time series of (a) ELWP and (b) IWC based on best estimates between the Ku- and Ka-band Ze and that estimated from the PSDs (section 2b).
Citation: Journal of the Atmospheric Sciences 79, 10; 10.1175/JAS-D-21-0311.1
REFERENCES
Barnes, S. L., 1964: A technique for maximizing details in numerical weather map analysis. J. Appl. Meteor., 3, 396–409, https://doi.org/10.1175/1520-0450(1964)003<0396:ATFMDI>2.0.CO;2.
Baumgardner, D., and A. Korolev, 1997: Airspeed corrections for optical array probe sample volumes. J. Atmos. Oceanic Technol., 14, 1224–1229, https://doi.org/10.1175/1520-0426(1997)014<1224:ACFOAP>2.0.CO;2.
Blaylock, B. K., J. D. Horel, and S. T. Liston, 2017: Cloud archiving and data mining of High-Resolution Rapid Refresh forecast model output. Comput. Geosci., 109, 43–50, https://doi.org/10.1016/j.cageo.2017.08.005.
Borque, P., K. J. Harnos, S. W. Nesbitt, and G. M. McFarquhar, 2019: Improved parameterization of ice particle size distributions using uncorrelated mass spectrum parameters: Results from GCPEx. J. Appl. Meteor. Climatol., 58, 1657–1676, https://doi.org/10.1175/JAMC-D-18-0203.1.
Brodzik, S., 2020a: GOES IMPACTS, version 1. NASA Global Hydrology Resource Center DAAC, accessed 21 July 2021, https://doi.org/10.5067/IMPACTS/GOES/DATA101.
Brodzik, S., 2020b: NEXRAD Mosaic East IMPACTS, version 1. NASA Global Hydrology Resource Center DAAC, accessed 17 July 2021, https://doi.org/10.5067/IMPACTS/NEXRAD/DATA201.
Brodzik, S., 2020c: NEXRAD Mosaic Midwest IMPACTS, version 1. NASA Global Hydrology Resource Center DAAC, accessed 17 July 2021, https://doi.org/10.5067/IMPACTS/NEXRAD/DATA202.
Brown, P. R. A., and P. N. Francis, 1995: Improved measurements of the ice water content in cirrus using a total-water probe. J. Atmos. Oceanic Technol., 12, 410–414, https://doi.org/10.1175/1520-0426(1995)012<0410:IMOTIW>2.0.CO;2.
Chandrasekar, V., and V. N. Bringi, 1987: Simulation of radar reflectivity and surface measurements of rainfall. J. Atmos. Oceanic Technol., 4, 464–478, https://doi.org/10.1175/1520-0426(1987)004<0464:SORRAS>2.0.CO;2.
Chase, R. J., and Coauthors, 2018: Evaluation of triple-frequency radar retrieval of snowfall properties using coincident airborne in situ observations during OLYMPEX. Geophys. Res. Lett., 45, 5752–5760, https://doi.org/10.1029/2018GL077997.
Chase, R. J., S. W. Nesbitt, and G. M. McFarquhar, 2020: Evaluation of the microphysical assumptions within GPM-DPR using ground-based observations of rain and snow. Atmosphere, 11, 619, https://doi.org/10.3390/atmos11060619.
Chase, R. J., S. W. Nesbitt, and G. M. McFarquhar, 2021: A dual-frequency radar retrieval of two parameters of the snowfall particle size distribution using a neural network. J. Appl. Meteor. Climatol., 60, 341–359, https://doi.org/10.1175/JAMC-D-20-0177.1.
Chase, R. J., S. W. Nesbitt, G. M. McFarquhar, N. B. Wood, and G. M. Heymsfield, 2022: Direct comparisons between GPM-DPR and CloudSat snowfall retrievals. J. Appl. Meteor. Climatol., 61, 1257–1271, https://doi.org/10.1175/JAMC-D-21-0081.1.
Delanoë, J., R. J. Hogan, R. M. Forbes, A. Bodas-Salcedo, and T. H. M. Stein, 2011: Evaluation of ice cloud representation in the ECMWF and UK Met Office models using CloudSat and CALIPSO data. Quart. J. Roy. Meteor. Soc., 137, 2064–2078, https://doi.org/10.1002/qj.882.
Delanoë, J., A. J. Heymsfield, A. Protat, A. Bansemer, and R. J. Hogan, 2014: Normalized particle size distribution for remote sensing application. J. Geophys. Res. Atmos., 119, 4204–4227, https://doi.org/10.1002/2013JD020700.
Ding, S., G. M. McFarquhar, S. W. Nesbitt, R. J. Chase, M. R. Poellot, and H. Wang, 2020: Dependence of mass–dimensional relationships on median mass diameter. Atmosphere, 11, 756, https://doi.org/10.3390/atmos11070756.
Duffy, G., G. McFarquhar, S. W. Nesbitt, and R. Bennartz, 2021: Demonstration of a consistent relationship between dual-frequency reflectivity and the mass-weighted mean diameter in measurements of frozen precipitation from GCPEX, OLYMPEX, and MC3E. J. Atmos. Sci., 78, 2533–2547, https://doi.org/10.1175/JAS-D-20-0174.1.
Finlon, J. A., G. M. McFarquhar, S. W. Nesbitt, R. M. Rauber, H. Morrison, W. Wu, and P. Zhang, 2019: A novel approach for characterizing the variability in mass–dimension relationships: Results from MC3E. Atmos. Chem. Phys., 19, 3621–3643, https://doi.org/10.5194/acp-19-3621-2019.
Finlon, J. A., and Coauthors, 2020: Structure of an atmospheric river over Australia and the Southern Ocean: II. Microphysical evolution. J. Geophys. Res. Atmos., 125, e2020JD032514, https://doi.org/10.1029/2020JD032514.
Ganetis, S. A., B. A. Colle, S. E. Yuter, and N. P. Hoban, 2018: Environmental conditions associated with observed snowband structures within northeast U.S. winter storms. Mon. Wea. Rev., 146, 3675–3690, https://doi.org/10.1175/MWR-D-18-0054.1.
Grecu, M., W. S. Olson, and E. N. Anagnostou, 2004: Retrieval of precipitation profiles from multiresolution, multifrequency active and passive microwave observations. J. Appl. Meteor., 43, 562–575, https://doi.org/10.1175/1520-0450(2004)043<0562:ROPPFM>2.0.CO;2.
Grecu, M., L. Tian, W. S. Olson, and S. Tanelli, 2011: A robust dual-frequency radar profiling algorithm. J. Appl. Meteor. Climatol., 50, 1543–1557, https://doi.org/10.1175/2011JAMC2655.1.
Hallett, J., 2003: Measurement in the atmosphere. Handbook of Weather, Climate, and Water: Dynamics, Climate, Physical Meteorology, Weather Systems, and Measurements, T. D. Potter and B. R. Colman, Eds., Wiley-Interscience, 711–720.
Heymsfield, A. J., C. Schmitt, A. Bansemer, and C. H. Twohy, 2010: Improved representation of ice particle masses based on observations in natural clouds. J. Atmos. Sci., 67, 3303–3318, https://doi.org/10.1175/2010JAS3507.1.
Heymsfield, A. J., C. Schmitt, C. Chen, A. Bansemer, A. Gettelman, P. R. Field, and C. Liu, 2020: Contributions of the liquid and ice phases to global surface precipitation: Observations and global climate modeling. J. Atmos. Sci., 77, 2629–2648, https://doi.org/10.1175/JAS-D-19-0352.1.
Heymsfield, G. M., and Coauthors, 1996: The EDOP radar system on the High-Altitude NASA ER-2 aircraft. J. Atmos. Oceanic Technol., 13, 795–809, https://doi.org/10.1175/1520-0426(1996)013<0795:TERSOT>2.0.CO;2.
Heymsfield, G. M., L. Li, and M. McLinden, 2020: ER-2 X-band Doppler Radar (EXRAD) IMPACTS, version 1. NASA Global Hydrometeorology Resource Center DAAC, accessed 24 March 2021, https://doi.org/10.5067/IMPACTS/EXRAD/DATA101.
Hogan, R. J., L. Tian, P. R. A. Brown, C. D. Westbrook, A. J. Heymsfield, and J. D. Eastment, 2012: Radar scattering from ice aggregates using the horizontally aligned oblate spheroid approximation. J. Appl. Meteor. Climatol., 51, 655–671, https://doi.org/10.1175/JAMC-D-11-074.1.
Hou, A. Y., and Coauthors, 2014: The Global Precipitation Measurement mission. Bull. Amer. Meteor. Soc., 95, 701–722, https://doi.org/10.1175/BAMS-D-13-00164.1.
Houze, R. A., Jr., and Coauthors, 2017: The Olympic Mountains Experiment (OLYMPEX). Bull. Amer. Meteor. Soc., 98, 2167–2188, https://doi.org/10.1175/BAMS-D-16-0182.1.
International Telecommunication Union, 2013: Attenuation by atmospheric gases. International Telecommunication Union Recommendation ITU‐R P.676-10, 24 pp., https://www.itu.int/dms_pubrec/itu-r/rec/p/R-REC-P.676-10-201309-S!!PDF-E.pdf.
Jackson, R. C., G. M. McFarquhar, A. D. Korolev, M. E. Earle, P. S. K. Liu, and R. P. Lawson, 2012: The dependence of ice microphysics on aerosol concentrations in Arctic mixed‐phased stratus clouds during ISDAC and M‐PACE. J. Geophys. Res., 117, D15207, https://doi.org/10.1029/2012JD017668.
Jensen, M. P., and Coauthors, 2016: The Midlatitude Continental Convective Clouds Experiment (MC3E). Bull. Amer. Meteor. Soc., 97, 1667–1686, https://doi.org/10.1175/BAMS-D-14-00228.1.
Jossart, J., S. J. Theuerkauf, L. C. Wickliffe, and J. A. Morris Jr., 2020: Applications of spatial autocorrelation analyses for marine aquaculture siting. Front. Mar. Sci., 6, 806, https://doi.org/10.3389/fmars.2019.00806.
Kneifel, S., M. S. Kulie, and R. Bennartz, 2011: A triple-frequency approach to retrieve microphysical snowfall parameters. J. Geophys. Res., 116, D11203, https://doi.org/10.1029/2010JD015430.
Kneifel, S., A. von Lerber, J. Tiira, D. Moisseev, P. Kollias, and J. Leinonen, 2015: Observed relations between snowfall microphysics and triple-frequency radar measurements. J. Geophys. Res. Atmos., 120, 6034–6055, https://doi.org/10.1002/2015JD023156.
Kulie, M. S., M. J. Hiley, R. Bennartz, S. Kneifel, and S. Tanelli, 2014: Triple-frequency radar reflectivity signatures of snow: Observations and comparisons with theoretical ice particle scattering models. J. Appl. Meteor. Climatol., 53, 1080–1098, https://doi.org/10.1175/JAMC-D-13-066.1.
Kummerow, C., W. Barnes, T. Kozu, J. Shiue, and J. Simpson, 1998: The Tropical Rainfall Measuring Mission (TRMM) sensor package. J. Atmos. Oceanic Technol., 15, 809–817, https://doi.org/10.1175/1520-0426(1998)015<0809:TTRMMT>2.0.CO;2.
Lawson, R. P., R. E. Stewart, and L. J. Angus, 1998: Observations and numerical simulations of the origin and development of very large snowflakes. J. Atmos. Sci., 55, 3209–3229, https://doi.org/10.1175/1520-0469(1998)055<3209:OANSOT>2.0.CO;2.
Lawson, R. P., D. O’Connor, P. Zmarzly, K. Weaver, B. Baker, Q. Mo, and H. Jonsson, 2006: The 2D-S (Stereo) probe: Design and preliminary tests of a new airborne, high-speed, high-resolution particle imaging probe. J. Atmos. Oceanic Technol., 23, 1462–1477, https://doi.org/10.1175/JTECH1927.1.
Leighton, H., R. Black, X. Zhang, F. D. Marks, and S. G. Gopalakrishnan, 2020: Ice particle size distributions from composites of microphysics observations collected in tropical cyclones. Geophys. Res. Lett., 47, e2020GL088762, https://doi.org/10.1029/2020GL088762.
Leinonen, J., and D. Moisseev, 2015: What do triple-frequency radar signatures reveal about aggregate snowflakes? J. Geophys. Res. Atmos., 120, 229–239, https://doi.org/10.1002/2014JD022072.
Leinonen, J., and W. Szyrmer, 2015: Radar signatures of snowflake riming: A modeling study. Earth Space Sci., 2, 346–358, https://doi.org/10.1002/2015EA000102.
Leinonen, J., S. Kneifel, D. Moisseev, J. Tyynelä, S. Tanelli, and T. Nousiainen, 2012: Evidence of nonspheroidal behavior in millimeter-wavelength radar observations of snowfall. J. Geophys. Res., 117, D18205, https://doi.org/10.1029/2012JD017680.
Leinonen, J., and Coauthors, 2018: Retrieval of snowflake microphysical properties from multifrequency radar observations. Atmos. Meas. Tech., 11, 5471–5488, https://doi.org/10.5194/amt-11-5471-2018.
Li, H., J. Tiira, A. von Lerber, and D. Moisseev, 2020: Towards the connection between snow microphysics and melting layer: Insights from multifrequency and dual-polarization radar observations during BAECC. Atmos. Chem. Phys., 20, 9547–9562, https://doi.org/10.5194/acp-20-9547-2020.
Li, L., and Coauthors, 2015: The NASA High-Altitude Imaging Wind and Rain Airborne Profiler (HIWRAP). IEEE Trans. Geosci. Remote Sens., 54, 298–310, https://doi.org/10.1109/TGRS.2015.2456501.
Li, L., M. McLinden, and G. M. Heymsfield, 2020: High Altitude Imaging Wind and Rain Airborne Profiler (HIWRAP) IMPACTS, version 1. NASA Global Hydrometeorology Resource Center DAAC, accessed 29 September 2021, https://doi.org/10.5067/IMPACTS/HIWRAP/DATA101.
Liao, L., R. Meneghini, A. Tokay, and L. F. Bliven, 2016: Retrieval of snow properties for Ku- and Ka-band dual-frequency radar. J. Appl. Meteor. Climatol., 55, 1845–1858, https://doi.org/10.1175/JAMC-D-15-0355.1.
Mascio, J., G. M. McFarquhar, T. Hsieh, M. Freer, A. Dooley, and A. J. Heymsfield, 2020: The use of gamma distributions to quantify the dependence of cloud particle size distributions in hurricanes on cloud and environmental conditions. Quart. J. Roy. Meteor. Soc., 146, 2116–2137, https://doi.org/10.1002/qj.3782.
Mason, S. L., R. J. Hogan, C. D. Westbrook, S. Kneifel, D. Moisseev, and L. von Terzi, 2019: The importance of particle size distribution and internal structure for triple-frequency radar retrievals of the morphology of snow. Atmos. Meas. Tech., 12, 4993–5018, https://doi.org/10.5194/amt-12-4993-2019.
Matrosov, S. Y., 1998: A dual-wavelength radar method to measure snowfall rate. J. Appl. Meteor., 37, 1510–1521, https://doi.org/10.1175/1520-0450(1998)037<1510:ADWRMT>2.0.CO;2.
Matrosov, S. Y., A. J. Heymsfield, and Z. Wang, 2005: Dual-frequency radar ratio of nonspherical atmospheric hydrometeors. Geophys. Res. Lett., 32, L13816, https://doi.org/10.1029/2005GL023210.
Matrosov, S. Y., M. Maahn, and G. de Boer, 2019: Observational and modeling study of ice hydrometeor radar dual-wavelength ratios. J. Appl. Meteor. Climatol., 58, 2005–2017, https://doi.org/10.1175/JAMC-D-19-0018.1.
McFarquhar, G. M., M. S. Timlin, R. M. Rauber, B. F. Jewett, J. A. Grim, and D. P. Jorgensen, 2007: Vertical variability of cloud hydrometeors in the stratiform region of mesoscale convective systems and bow echoes. Mon. Wea. Rev., 135, 3405–3428, https://doi.org/10.1175/MWR3444.1.
McFarquhar, G. M., T. Hsieh, M. Freer, J. Mascio, and B. F. Jewett, 2015: The characterization of ice hydrometeor gamma size distributions as volumes in N0–λ–μ phase space: Implications for microphysical process modeling. J. Atmos. Sci., 72, 892–909, https://doi.org/10.1175/JAS-D-14-0011.1.
McFarquhar, G. M., J. A. Finlon, D. M. Stechman, W. Wu, R. C. Jackson, and M. Freer, 2018: University of Illinois/Oklahoma optical array probe (OAP) processing software. Zenodo, https://doi.org/10.5281/zenodo.1285969.
McLinden, M., L. Li, and G. M. Heymsfield, 2020: Cloud Radar System (CRS) IMPACTS, version 1. NASA Global Hydrometeorology Resource Center DAAC, accessed 24 March 2021, https://doi.org/10.5067/IMPACTS/CRS/DATA101.
McLinden, M., L. Li, G. M. Heymsfield, M. Coon, and A. Emory, 2021: The NASA/GSFC 94 GHz airborne solid-state Cloud Radar System (CRS). J. Atmos. Oceanic Technol., 38, 1001–1017, https://doi.org/10.1175/JTECH-D-20-0127.1.
McMurdie, L. A., G. Heymsfield, J. E. Yorks, and S. A. Braun, 2019: Investigation of Microphysics and Precipitation for Atlantic Coast-Threatening Snowstorms (IMPACTS) Collection. NASA EOSDIS Global Hydrology Resource Center DAAC, accessed 24 September 2020, https://doi.org/10.5067/IMPACTS/DATA101.
McMurdie, L. A., and Coauthors, 2022: Chasing snowstorms: The Investigation of Microphysics and Precipitation for Atlantic Coast-Threatening Snowstorms (IMPACTS) campaign. Bull. Amer. Meteor. Soc., 103, E1243–E1269, https://doi.org/10.1175/BAMS-D-20-0246.1.
Meneghini, R., and T. Kozu, 1990: Spaceborne Weather Radar. Artech House, 212 pp.
Moisseev, D. N., and V. Chandrasekar, 2007: Examination of the μ–Λ relation suggested for drop size distribution parameters. J. Atmos. Oceanic Technol., 24, 847–855, https://doi.org/10.1175/JTECH2010.1.
Moran, P., 1950: Notes on continuous stochastic phenomena. Biometrika, 37, 17–23, https://doi.org/10.1093/biomet/37.1-2.17.
Mroz, K., A. Battaglia, C. Nguyen, A. Heymsfield, A. Protat, and M. Wolde, 2021: Triple-frequency radar retrieval of microphysical properties of snow. Atmos. Meas. Tech., 14, 7243–7254, https://doi.org/10.5194/amt-14-7243-2021.
Nguyen, C. M., M. Wolde, A. Battaglia, L. Nichman, N. Bliankinshtein, S. Haimov, K. Bala, and D. Schuettemeyer, 2022: Coincident in situ and triple-frequency radar airborne observations in the Arctic. Atmos. Meas. Tech., 15, 775–795, https://doi.org/10.5194/amt-15-775-2022.
Ni, X., C. Liu, and E. Zipser, 2019: Ice microphysical properties near the tops of deep convective cores implied by the GPM dual-frequency radar observations. J. Atmos. Sci., 76, 2899–2917, https://doi.org/10.1175/JAS-D-18-0243.1.
Novak, D. R., L. F. Bosart, D. Keyser, and J. S. Waldstreicher, 2004: An observational study of cold season–banded precipitation in northeast U.S. cyclones. Wea. Forecasting, 19, 993–1010, https://doi.org/10.1175/815.1.
Oliphant, T., 2007: Python for scientific computing. Comput. Sci. Eng., 9, 10–20, https://doi.org/10.1109/MCSE.2007.58.
Ori, D., V. Schemann, M. Karrer, J. D. Neto, L. von Terzi, A. Seifert, and S. Kneifel, 2020: Evaluation of ice particle growth in ICON using statistics of multi-frequency Doppler cloud radar observations. Quart. J. Roy. Meteor. Soc., 146, 3830–3849, https://doi.org/10.1002/qj.3875.
Petty, G. W., and W. Huang, 2010: Microwave backscatter and extinction by soft ice spheres and complex snow aggregates. J. Atmos. Sci., 67, 769–787, https://doi.org/10.1175/2009JAS3146.1.
Picca, J. C., D. M. Schultz, B. A. Colle, S. Ganetis, D. R. Novak, and M. J. Sienkiewicz, 2014: The value of dual-polarization radar in diagnosing the complex microphysical evolution of an intense snowband. Bull. Amer. Meteor. Soc., 95, 1825–1834, https://doi.org/10.1175/BAMS-D-13-00258.1.
Sekelsky, S. M., W. L. Ecklund, J. M. Firda, K. S. Gage, and R. E. McIntosh, 1999: Particle size estimation in ice-phase clouds using multifrequency radar reflectivity measurements at 95, 33, and 2.8 GHz. J. Appl. Meteor., 38, 5–28, https://doi.org/10.1175/1520-0450(1999)038<0005:PSEIIP>2.0.CO;2.
Skofronick-Jackson, G., and Coauthors, 2015: Global Precipitation Measurement Cold Season Precipitation Experiment (GCPEX): For measurement’s sake, let it snow. Bull. Amer. Meteor. Soc., 96, 1719–1741, https://doi.org/10.1175/BAMS-D-13-00262.1.
Skofronick-Jackson, G., and Coauthors, 2017: The Global Precipitation Measurement (GPM) mission for science and society. Bull. Amer. Meteor. Soc., 98, 1679–1695, https://doi.org/10.1175/BAMS-D-15-00306.1.
Skofronick-Jackson, G., M. Kulie, L. Milani, S. J. Munchak, N. B. Wood, and V. Levizzani, 2019: Satellite estimation of falling snow: A Global Precipitation Measurement (GPM) Core Observatory perspective. J. Appl. Meteor. Climatol., 58, 1429–1448, https://doi.org/10.1175/JAMC-D-18-0124.1.
Stein, T. H. M., C. D. Westbrook, and J. C. Nicol, 2015: Fractal geometry of aggregate snowflakes revealed by triple-wavelength radar measurements. Geophys. Res. Lett., 42, 176–183, https://doi.org/10.1002/2014GL062170.
Stephens, G. L., and Coauthors, 2002: The CloudSat mission and the A-Train: A new dimension of space-based observations of clouds and precipitation. Bull. Amer. Meteor. Soc., 83, 1771–1790, https://doi.org/10.1175/BAMS-83-12-1771.
Testud, J., S. Oury, R. A. Black, P. Amayenc, and X. Dou, 2001: The concept of “normalized” distribution to describe raindrop spectra: A tool for cloud physics and cloud remote sensing. J. Appl. Meteor., 40, 1118–1140, https://doi.org/10.1175/1520-0450(2001)040<1118:TCONDT>2.0.CO;2.
Tridon, F., and Coauthors, 2019: The microphysics of stratiform precipitation during OLYMPEX: Compatibility between triple-frequency radar and airborne in situ observations. J. Geophys. Res. Atmos., 124, 8764–8792, https://doi.org/10.1029/2018JD029858.
Tyynelä, J., and A. von Lerber, 2019: Validation of microphysical snow models using in-situ, multi-frequency and dual-polarization radar measurements in Finland. J. Geophys. Res. Atmos., 124, 13 273–13 290, https://doi.org/10.1029/2019JD030721.
Ulbrich, C. W., 1983: Natural variations in the analytical form of the raindrop size distribution. J. Appl. Meteor. Climatol., 22, 1764–1775, https://doi.org/10.1175/1520-0450(1983)022<1764:NVITAF>2.0.CO;2.
von Terzi, L., J. Dias Neto, D. Ori, A. Myagkov, and S. Kneifel, 2022: Ice microphysical processes in the dendritic growth layer: A statistical analysis combining multi-frequency and polarimetric Doppler cloud radar observations. Atmos. Chem. Phys., 22, 11 795–11 821, https://doi.org/10.5194/acp-22-11795-2022.
Wang, Y., and Coauthors, 2020: Microphysical properties of generating cells over the Southern Ocean: Results from SOCRATES. J. Geophys. Res. Atmos., 125, e2019JD032237, https://doi.org/10.1029/2019JD032237.
Williams, C. R., and Coauthors, 2014: Describing the shape of raindrop size distributions using uncorrelated raindrop mass spectrum parameters. J. Appl. Meteor. Climatol., 53, 1282–1296, https://doi.org/10.1175/JAMC-D-13-076.1.
Wolde, M., and Coauthors, 2019: Radar Snow Experiment for future precipitation mission (RadSnowExp). Living Planet Symp. 2019, Milan, Italy, European Space Agency.
Yin, M., G. Liu, R. Honeyager, and F. J. Turk, 2017: Observed differences of triple-frequency radar signatures between snowflakes in stratiform and convective clouds. J. Quant. Spectrosc. Radiat. Transfer, 193, 13–20, https://doi.org/10.1016/j.jqsrt.2017.02.017.
Yu, T., V. Chandrasekar, H. Xiao, and S. S. Joshil, 2021: Snowfall estimation using dual-wavelength radar during the Pyeongchang 2018 Olympics and Paralympic Winter Games. J. Meteor. Soc. Japan, 99, 67–77, https://doi.org/10.2151/jmsj.2021-004.
Zagrodnik, J. P., L. A. McMurdie, R. A. Houze Jr., and S. Tanelli, 2019: Vertical structure and microphysical characteristics of frontal systems passing over a three-dimensional coastal mountain range. J. Atmos. Sci., 76, 1521–1546, https://doi.org/10.1175/JAS-D-18-0279.1.
