1. Introduction
Measuring clouds to understand the involved ice particle growth processes (Shupe et al. 2008; Kalesse et al. 2016) is still a challenge because of the small temporal and spatial scales involved. Ground-based radar measurements are widely used for such observations (Kollias et al. 2007; Shupe et al. 2008). Nowadays their advanced capabilities make the observation and study of microphysical processes within cloud and precipitation systems possible.
One approach for improving the understanding of cloud particle growth processes is following a population of particles from their generation through their different stages of development until they evaporate or fall as precipitation on the ground (Pruppacher and Klett 1996).
This can be done by tracking fall streak structures within radar measurements (Marshall and Gunn 1954; Yuter and Houze 2003; Kalesse et al. 2016). Yuter and Houze (2003) defined a fall streak signature as a manifestation of an inhomogeneity in the microphysical structure of a cloud system. To be observed, the relative size and number of precipitation particles within the fall streak need to be sufficient such that their radar reflectivity stands out as a local maximum from the immediate background reflectivities. Such fall streak structures are visible in radar reflectivity range–height indicator (RHI) scans or time–height plots, when the thermodynamical conditions in so-called generating cells lead to a continuous and homogeneous production of particles (Heymsfield 1975b; Pruppacher and Klett 1996). Figure 1 depicts such a fall streak structure (dark blue area from top to bottom), with the particles being generated near cloud top (Marshall 1953; Heymsfield 1975a).
(a) Theoretical sketch of radar reflectivity time–height plots of observed fall streak signatures within a precipitating cloud system. Sketch shows observations taken by two different profiling radars (radars 1 and 2) that are measured at different places (distance
Citation: Journal of Atmospheric and Oceanic Technology 34, 4; 10.1175/JTECH-D-16-0117.1
Following the generating level concept, Marshall (1953) and Browne (1952) were the first to analyze and later retrieve the shape and structure of a fall streak within radar measurements, which is similar to the one featured in Fig. 1a. By varying the input parameters of the fall streak calculation, some analysis of the particle population was also performed. Marshall (1953) related the broadening in fall streaks to the size sorting of the crystals depending on some size–fall speed relationships. The width of the fall streak was then correlated to the size range of the particle size distribution of the analyzed particle population. Independent of Marshall (1953), the same microphysical relations were found by Browne (1952). The predominant influence on the fall streak shape was found to be the horizontal wind structure within the cloud system (Marshall 1953; Marshall and Gunn 1954). This was shown by Marshall and Gunn (1954), where they demonstrated that slope changes in fall streaks are related to changes in the horizontal wind field within the cloud system.
To be able to analyze particle growth processes using fall streak signatures, some assumptions have to be made. First of all, it is assumed that particles generate continuously and homogeneously within the generating cell. Second, the dynamical and microphysical conditions of the cloud system are homogeneous and stable over time. In such a way, it is possible to translate fall streak signatures based on Eulerian observations provided by the 2D time plot of the radar to Lagrangian-based “pseudo” particle trajectory. Figure 1a shows the same fall streak signature at two different times obtained by profiling radars, located in the line of wind direction with known distance
Marshall and Gunn (1954) linked the fall streak in clouds to observed microphysical changes in precipitation pattern. This was the first attempt to correlate precipitation patterns with the ice particle growth found in the cloud aloft. This approach was followed by Yuter and Houze (2003) and Mittermaier et al. (2004) to link precipitation pattern to the particle formation processes aloft. Mittermaier et al. (2004) used the fall streak structures to improve the forecast of precipitation patterns at the ground. This was also considered for improving the validation of the rain estimates with rain gauges at ground level so that the fit between precipitation peaks in the radar data and rain gauges can be enhanced. Other microphysical studies have been performed where different particle populations and their different microphysical processes were tracked along the streak. Yuter and Houze (2003) focused on the link of particle formation and the resulting rain intensity, while Kalesse et al. (2016) focused on riming processes within winter precipitation. Observation results of both studies were compared to 1D column models to see whether the models are able to reproduce the observations. In both cases, the models were able to reproduce the processes, although Kalesse et al. (2016) stressed that more observations are needed to minimize the initialization settings of the model. The fall streak concept was also used to create inhomogeneity in modeled cirrus cloud field (Hogan and Kew 2005). This was done to find out what influence those inhomogeneities in clouds have on radiative transfer simulations. The result is that 3D effects can significantly affect the radiation budget and that within global climate models a parameterization adjustment might be useful. All these papers point out the potential and the possibilities of using fall streaks for further microphysical analysis. It is, however, worth stressing that all applications rely on additional wind information, a chosen relation between particle size and fall velocity, and an assumption on the generation level height. Because of a lack of horizontal wind field information, analysis of fall streaks is limited to situations where dynamical conditions are simple and stable over time (Marshall 1953; Marshall and Gunn 1954; Kalesse et al. 2016).
In this paper, a novel definition of fall streaks based on particle dynamic rather than on microphysical contrast is used. The shape of the fall streak is, indeed, mainly influenced by the cloud dynamic, which does not necessarily have to follow an enhanced or outstanding reflectivity pattern (i.e., homogeneous cloud conditions). To represent fall streaks for different cloud situations, we base our definition on the path of a particle population obtained from the observation of its own motion. If the exact cloud dynamic is known, then it is possible to retrieve fall streaks according to the individual particle motions for each time step of the radar measurement, as seen by the white dotted lines in Fig. 1a. Note that because of this definition, features like the width of the fall streak patterns cannot be taken into account. Following the concept based on individual particle motions, the definition aims at the microphysical process understanding of the tracked particle population falling through the cloud system rather than analyzing the size sorting of the near–cloud top generated particles.
In this paper, we introduce an automatic fall streak retrieval based on single-Doppler measurements, taken with the TU Delft–operated Transportable Atmospheric Radar (TARA). From this radar, the full 3D wind vector per sampling volume can be retrieved, thanks to its three-beam configuration (Unal et al. 2012). Furthermore, the high resolution of 3D wind information provided by TARA makes it possible to retrieve fall streaks at high temporal resolution, offering more insights into the growth processes occurring in complex, local, and inhomogeneous cloud conditions. A better representation of the diversity of the fall streaks within a selected time frame is, in this way, achieved. Finally, fall streaks are retrieved based on measurements of a single instrument so that fewer assumptions for the algorithm, compared to previous techniques, are required. After introducing the data and the radar system in section 2, the paper gives an overview of the proposed retrieval technique in section 3. The limitations and requirements of the retrieval are discussed in section 4, and section 5 shows preliminary retrieval results.
2. Dataset and instrument
a. TARA and Composition of Clouds with Extended Polarization Techniques campaign dataset
The results and retrieval developments are based on measurements performed with TARA (Heijnen et al. 2000). TARA is a frequency-modulated continuous wave (FM-CW) S-band radar profiler that has Doppler and fully polarimetric capabilities.
Data measured during the Analysis of the Composition of Clouds with Extended Polarization Techniques (ACCEPT) campaign is used to illustrate the fall streak algorithm. The measurements were performed from October to November 2014 at the Cabauw Experimental Site for Atmospheric Research (CESAR), the Netherlands. TARA was measuring collocated with an extended setup of the Leipzig Aerosol and Cloud Remote Observations System (LACROS; Bühl et al. 2013). The aim of ACCEPT is to understand the microphysical processes involved in mixed-phase clouds at high resolution. One focus is to improve the understanding of ice crystal formation at the top of single-layer mixed-phase clouds (Myagkov et al. 2016). A second focus is to improve the understanding of ice particle growth when ice crystals fall through such liquid layers embedded within the cloud systems.
To observe the variety in size and shape, and the different phases of the involved hydrometeors, a synergy of instruments was used. TARA measured in parallel with the vertically pointing Ka-band cloud radar Millimeter Wave Radiometer (MIRA; Görsdorf et al. 2015) to obtain ice crystal information within the cloud being probed. Adding a high-frequency radar (MIRA,
b. Use of TARA as wind profiler: The wind retrieval
The 3D wind field can be retrieved because of the unique three-beam configuration of TARA (Unal et al. 2012). Using the Doppler spectra information of the three beams—main beam and two offset beams—the horizontal wind velocity
Specifications of TARA during the ACCEPT campaign. HH = horizontal transmit and horizontal receive. VV; vertical transmit and vertical receive. HV = vertical transmit and horizontal receive.
3. The fall streak retrieval technique
The aim of the algorithm is to retrieve and analyze the microphysical evolution along fall streaks of a particle population from the particle generation till they reach the bottom of the cloud system.
In comparison with the common definition of fall streaks based on microphysical contrast, it can be seen that the fall streaks retrieved with the new method follow in some cases the enhanced reflectivity filaments (0249 and 0250 UTC at 3 km in Fig. 1b). Consistency is therefore found between the two definitions when a reflectivity contrast is observed.









Schematics of how the retrieval estimates the best start time for the fall streak retrieval for (a) a stratiform cloud and (b) a precipitating cloud case. In (a) the retrieval does not adjust the averaging time window. For a raining case, the algorithm adjusts the position of the averaging window. Therefore,
Citation: Journal of Atmospheric and Oceanic Technology 34, 4; 10.1175/JTECH-D-16-0117.1
Flowchart of the fall streak retrieval algorithm. Boxes 1–4 show the general retrieval routines with the needed input and output variables for each step. Box E deals with the estimation of the best averaging window within a fixed time frame to retrieve multiple fall streaks.
Citation: Journal of Atmospheric and Oceanic Technology 34, 4; 10.1175/JTECH-D-16-0117.1
Sketches illustrate the basic concept behind the calculation of (a)
Citation: Journal of Atmospheric and Oceanic Technology 34, 4; 10.1175/JTECH-D-16-0117.1
a. Step 1: Scaling of the wind profiles






b. Step 2: Retrieving the start time of the averaging window
The focus of the retrieval algorithm is to obtain the fall streak structure within the cloud. In case of a raining system, the cloud is defined as the part above the melting layer. Otherwise, the retrieval starts at the radar-detected cloud bottom; see examples in Fig. 2. The higher variability of the wind in the cloud compared to the precipitation part of the cloud system makes it necessary to average the wind to get homogeneous wind profiles. To do so, the averaging window can be optimized in terms of location and averaging time.






The start point of the averaging window is obtained by defining the intersection of
As seen in Fig. 5a, which shows the summation of
Profiles of the (a) two parts of the fall streak retrieval and (b) corresponding wind profiles. In (a) the summed up dynamical contribution
Citation: Journal of Atmospheric and Oceanic Technology 34, 4; 10.1175/JTECH-D-16-0117.1
c. Step 3: Elevation contribution





d. Step 3: Dynamical contribution








The first term in Eq. (5) describes the travel time (the time particle population needs to fall through one radar height bin). Figure 4b illustrates that
The second term of Eq. (5) calculates the advection or relative horizontal displacement of fall streaks, which depends on the difference of
Last, in Eq. (5) a turbulence contribution
If in Eq. (5) singularities occur [
In comparison with the existing theory of Browne (1952) and Marshall and Gunn (1954), and following representations (Hogan and Kew 2005), no generation level at cloud top and no microphysical assumptions are used. A bottom-up approach is chosen instead, due to lower accuracy of the wind retrieval in the far range of the radar beam. So, the reference point of the retrieval is as close to the ground as possible where we expect accurate dynamical information.
e. Step 4: Bottom-up summation of Δtα and Δtdyn











This procedure strongly depends on the accuracy of retrieved winds, which may decrease at far ranges. This decrease is caused, on the one hand, by a decrease of the radar sensitivity and, on the other hand, by the increasing physical separation of the three-beam resolution volumes of the wind profiler with increasing range (height). With the bottom-up approach, errors are reduced, and no generation level has to be assumed. This procedure relates the retrieved fall streak to the t0,bin(z0), which is closest to the bottom. But to get appropriate results for the fall streak within the cloud part, which is the aim of this retrieval, the position of the averaging window plays an important role. How to retrieve the right start point for different cases is explained in the next section.
4. Discussion: Limitations and requirements of the retrieval technique
a. Retrieved 3D wind
Thanks to its three-beam configuration, TARA can retrieve the full 3D wind field continuously at high resolution within the cloud systems, taking into account Doppler information within each beam. However, the wind retrieval decreases in quality in the region where the signal-to-noise-ratio (SNR) is low or when turbulence is present, or the wind field is inhomogeneous. Turbulence is often the highest at the cloud edges because of the de-entrainment and entrainment of air. Low SNR is also expected near the cloud top because there the particles are small in size and their concentration is low. Furthermore, the radar sensitivity is decreasing with increasing range (height). At cloud base for nonprecipitating clouds, similar effects occur, like the entrainment of dry air leading to turbulence and evaporation—the latter results in an SNR decrease. Because of the bottom-up structure of the algorithm, choosing
Low SNR is detected using the linear depolarization ratio (Ldr). This reflectivity ratio, cross-polar to copolar, can be seen as the noise-to-signal ratio (NSR, dB) in the cloud part because the cross-polar measurement within that area is below the noise level. Therefore, this radar observable provides an estimation of the SNR, which is used to ensure the quality of the data. A threshold of
Broadening of the Doppler spectra within the three beam can also lead to biased wind information. The broadening of the Doppler spectrum can be caused by turbulence or by horizontal wind shear. In cases of wind shear, the broadening within the three-beam Doppler spectra might not be evenly spread. Furthermore, the spectra might be skewed toward the smaller particles, because of the higher inertia of the bigger particles (Oude Nijhuis et al. 2016). Large particles are less prone to follow the airstream flow compared to the smaller particles. Therefore, the calculated mean Doppler velocity of that volume shifts toward lower or larger fall speeds.
Last, the final result of the retrieval can also be influenced by the multimodality of the spectra. Several microphysical processes can affect the particle population monitored along the fall streak. This sometimes can lead to the formation of a secondary particle population (new particle formation or particle growth), which in turn can lead to the secondary fall velocity mode in the Doppler spectra. Such a secondary mode in the spectrum causes a broadening, affecting the calculated mean Doppler velocity and therefore the retrieved fall streaks. The algorithm is at the moment not capable of handling such a case.
Averaging is used to improve the wind estimates. It is worth mentioning that the averaging window size has an influence on the shape of the resulting fall streak.
b. Averaging window size
The slope and shape of the retrieved fall streaks depend on the retrieved wind input and its averaging time. Therefore, the choice of the average window for the wind average is the main influence of the fall streak shape. For a case study, the appropriate wind averaging can be estimated by a variation of the averaging time for a fixed start time. In the case of retrieving all the fall streaks within a fixed period, the averaging window has to be fixed. This fixed averaging window is selected in a way that guarantees the best representation of the dynamical conditions during that period. In the following, a statistical method is proposed to estimate the best averaging window within a fixed time frame using a bias analysis of the retrieved fall streaks (step E in Fig. 3).










Figure 6 shows the results of such an analysis for two cases. Both plots show the analysis for a 10-min period with a 1-min time difference between the different
Normalized bias of the retrieved fall streaks with respect to different selections of the average window sizes within a fixed time frame. (a) Mean normalized bias profile over
Citation: Journal of Atmospheric and Oceanic Technology 34, 4; 10.1175/JTECH-D-16-0117.1
Overview of a stratiform cloud case of ACCEPT observed 1640–1710 UTC 12 Oct 2014. (a) Reflectivity overlaid with eight retrieved fall streaks, separated with a 1-min time difference from 1657 to 1704 UTC. (b),(c) Plots of the vertical and horizontal Doppler velocities, respectively. (d) Reflectivity overlaid with the retrieved supercooled liquid water signatures from lidar (black; de Boer et al. 2009). (e) Doppler spectrum width.
Citation: Journal of Atmospheric and Oceanic Technology 34, 4; 10.1175/JTECH-D-16-0117.1
Figure 6b shows results for a precipitating and dynamically more complex case. In the rain part, at heights below
c. 3D structure of fall streaks
Using a profiling radar system, only 2D information (time–height) of a cloud system is obtained. Therefore, it is not possible to get any information of the 3D fall streak structures. The fall streaks are retrieved based on the assumption of a homogeneous cloud system (concerning dynamics and microphysical processes). This assumption makes it possible to study the microphysical evolution along the fall streak rearranged data.
The projection of the horizontal wind on the line of sight of the radar is made. For cases where the wind direction smoothly changes in the cloud system, the retrieval can be applied. For cases where a sharp shear of the wind direction within a small height can be seen, the microphysical continuity within the fall streak cannot be assumed.
5. Results
Several fall streaks retrieved for two different cloud systems are presented in this section. Figures 7d and 8d show TARA measurements of a stratiform and a raining cloud system, respectively, obtained during the ACCEPT campaign. The results of the fall streak retrieval for those two cloud cases are depicted in Figs. 7a and 8a, respectively.
Overview of a raining cloud case of ACCEPT observed 0240–0300 UTC 16 Oct 2014. (a) Reflectivity overlaid with 10 retrieved fall streaks separated by 1-min time difference from 0250 to 0259 UTC. (b),(c) Plots of the vertical and horizontal Doppler velocities, respectively. (d) Reflectivity overlaid with the retrieved supercooled liquid water signatures from lidar (black; de Boer et al. 2009). (e) Doppler spectrum width.
Citation: Journal of Atmospheric and Oceanic Technology 34, 4; 10.1175/JTECH-D-16-0117.1
a. Stratiform cloud case: 12 October 2014
Figure 7 represents TARA measurements of a stratiform cloud of an approaching warm front on 12 October 2014, at the Cabauw Experimental Site for Atmospheric Research (CESAR).
Figure 7a depicts eight retrieved fall streaks using a 6-min averaged wind profile and an approximate
The mean
In Fig. 7d no signature of supercooled liquid water is retrieved within the cloud system when using the method of de Boer et al. (2009). The liquid water path (LWP) measurements from the microwave radiometer show values below
b. Precipitating case: 16 October 2014
Figures 8 and 9 show a cloud system that was related to occlusion front overpasses at the CESAR observatory in the night of 16 October 2014. This example shows a precipitating cloud with an embedded liquid layer so that the benefit of analyzing growth processes related to supercooled liquid water using fall streak correction can be examined.
(a) Vertical spectrogram of Doppler spectra at 0251 UTC. (b) Fall streak–corrected spectrogram at
Citation: Journal of Atmospheric and Oceanic Technology 34, 4; 10.1175/JTECH-D-16-0117.1
1) Retrieved fall streaks
Figure 8a features the result of 10 different retrieved fall streaks from 0250 to 0259 UTC. All fall streaks are retrieved with 1.5-m averaged wind profiles with a 1-min interval. The wind profile averaging time is smaller than the one in the case in section 5a because the dynamical conditions are less homogeneous. The mean
2) Spectrograms
Spectrograms are used to identify and analyze changes in the ice particle microphysics and to link them to the presence of supercooled liquid water. Spectrograms provide Doppler spectra information per height bin at each time step (spectral reflectivity vs Doppler velocity and height, elevation of
Figure 9a shows a spectrogram as a vertical profile at 0251 UTC, while Fig. 9b is the fall streak–corrected spectrogram for the same starting time (along the light blue line in Fig. 8a). In the example presented in Fig. 9, three specific features can be identified (see arrows and letters in Fig. 9).
The first region features a spectral broadening between 3.5 and 4 km, where multimodality can be identified in Fig. 9b. This broadening is not correlated with any dynamical effect from Fig. 8 and therefore changes of the particle microphysics are taken into account. As indicated with arrows, two different particle modes, A and B, can be identified and separated in the spectrogram between 4 and 3.5 km. Between 3.8 and 3.5 km a third particle mode, C, occurs that has even higher Doppler velocities with respect to the other two modes. Other studies showed similar multimodal structures of Doppler spectra when particles fall through a liquid layer and start to rime (Kalesse et al. 2016; Oue et al. 2015). Therefore, we assume that a supercooled liquid layer is present in that region with riming processes involved. Because measurements are done under
In Fig. 9b a second region, indicated with a D, is visible between 2.6 and 3.1 km. As for the first region, no significant wind shear can be detected in Figs. 8b and 8c. The increase of reflectivity at the same heights in Fig. 8 indicates microphysical changes that lead to a broadening of the spectra. At the bottom of the broadening signature at
The third feature is the increase of reflectivity within the rain pattern below the melting layer (ML; see Fig. 9). This feature is well correlated with the lowest particle growth processes, D, detected in the cloud and is discussed above. The vertical profile spectrogram shows a weaker correlation of this cloud-to-rain conversion related to the detected particle growth processes.
Summarizing, it is shown in this section that there is a strong potential for studying microphysical processes of a particle population along its path from cloud top to the bottom using the fall streak–corrected spectrograms. In this example, enhanced cloud particle growth due to the presence of a liquid layer increases rainfall intensity. This type of pattern can be identified several times in the time–height plots in Fig. 8. Further analysis shows that this signature is correlated with the small pattern of upward motion in the vertical Doppler velocity field at around
6. Conclusions
In this paper, a new algorithm to retrieve fall streaks within a radar time–height plot is presented. The aim is to study the microphysical process evolution of a single particle population through its fall. Fall streaks are based on the assumption that particle populations are constantly and homogeneously generated at cloud top. Under such hypothesis, the fall streak signature contains all evolution states of the tracked particle population and therefore their microphysical changes can be observed.
The unique aspect of the retrieval is that it relies on genuine high-resolution wind information obtained with the Transportable Atmospheric Radar (TARA), avoiding any assumption on the wind field from other sensors or models that are known to drastically affect the accuracy of the retrieval. The high spatial and temporal observations provided by TARA can be employed to retrieve fall streaks in the case of dynamically stable stratiform cloud cases (section 5a) as well as for precipitating and dynamically more complex cases (section 5b).
Several steps are taken into account to guarantee the quality of the input velocity field, based on adaptive averaged windows. The presented case studies suggest that the retrieval can produce robust results for stratiform clouds and raining cases. Regarding microphysical process studies and particle growth due to supercooled liquid water presence, the spectrograms in Fig. 9 display clearly the advantage of using the fall streak–corrected spectrogram instead of the vertical profile spectrogram. Using the fall streak–corrected spectrogram, the identified signatures can be linked to coherent microphysical processes. The coherent features observed can also be linked and better compared to the structures that are visible in the corresponding time–height plots.
Summarizing the presented retrieval technique provides the first algorithm for fall streaks that is completely independent of additional wind information, a prescribed relation between particle size and fall velocity, and particle generation level height. Such fall streaks offer a completely new perspective and the potential to study cloud microphysics through process evolution analyses of the tracked particle populations.
Acknowledgments
The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013): People, ITN Marie Curie Actions Programme (20122016) in the frame of ITaRS under Grant Agreement 289923. The ACCEPT campaign was partly funded by ACTRIS Research Infrastructure Project by the European Union’s Horizon 2020 research and innovation programme under Grant Agreement 654169 and previously under Grant Agreement 262254 in the Seventh Framework Programme (FP7/2007–2013). The authors acknowledge also the cooperation of institutes (TROPOS, KNMI, LMU) and companies (METEK GmbH) during ACCEPT. We highly appreciated S. Kneifel for the discussions about the presented work.
APPENDIX
Variables List of the Retrieval Technique
Table A1 is placed on the previous page.
List of variables.
REFERENCES
Althausen, D., R. Engelmann, H. Baars, B. Heese, A. Ansmann, D. Müller, and M. Komppula, 2009: Portable Raman lidar PollyXT for automated profiling of aerosol backscatter, extinction, and depolarization. J. Atmos. Oceanic Technol., 26, 2366, doi:10.1175/2009JTECHA1304.1.
Browne, I. C., 1952: Precipitation streaks as a cause of radar upper bands. Quart. J. Roy. Meteor. Soc., 78, 590–595, doi:10.1002/qj.49707833809.
Bühl, J., and Coauthors, 2013: LACROS: The Leipzig Aerosol and Cloud Remote Observations System. Remote Sensing of Clouds and the Atmosphere XVIII; and Optics in Atmospheric Propagation and Adaptive Systems XVI, A. Comeron et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 8890), 889002, doi:10.1117/12.2030911.
de Boer, G., E. W. Eloranta, and M. D. Shupe, 2009: Arctic mixed-phase stratiform cloud properties from multiple years of surface-based measurements at two high-latitude locations. J. Atmos. Sci., 66, 2874, doi:10.1175/2009JAS3029.1.
Görsdorf, U., V. Lehmann, M. Bauer-Pfundstein, G. Peters, D. Vavriv, V. Vinogradov, and V. Volkov, 2015: A 35-GHz polarimetric Doppler radar for long-term observations of cloud parameters—Description of system and data processing. J. Atmos. Oceanic Technol., 32, 675–690, doi:10.1175/JTECH-D-14-00066.1.
Heijnen, S. H., L. P. Ligthart, and H. W. J. Russchenberg, 2000: First measurements with TARA; an S-band transportable atmospheric radar. Phys. Chem. Earth, 25B, 995–998, doi:10.1016/S1464-1909(00)00140-4.
Heymsfield, A., 1975a: Cirrus uncinus generating cells and the evolution of cirriform clouds. Part I: Aircraft observations of the growth of the ice phase. J. Atmos. Sci., 32, 799–808, doi:10.1175/1520-0469(1975)032<0799:CUGCAT>2.0.CO;2.
Heymsfield, A., 1975b: Cirrus uncinus generating cells and the evolution of cirriform clouds. Part II: The structure and circulations of the cirrus uncinus generating head. J. Atmos. Sci., 32, 809–819, doi:10.1175/1520-0469(1975)032<0809:CUGCAT>2.0.CO;2.
Hogan, R. J., and S. F. Kew, 2005: A 3D stochastic cloud model for investigating the radiative properties of inhomogeneous cirrus clouds. Quart. J. Roy. Meteor. Soc., 131, 2585–2608, doi:10.1256/qj.04.144.
Kalesse, H., W. Szyrmer, S. Kneifel, P. Kollias, and E. Luke, 2016: Fingerprints of a riming event on cloud radar Doppler spectra: Observations and modeling. Atmos. Chem. Phys., 16, 2997–3012, doi:10.5194/acp-16-2997-2016.
Kollias, P., and B. Albrecht, 2010: Vertical velocity statistics in fair-weather cumuli at the ARM TWP Nauru Climate Research Facility. J. Climate, 23, 6590–6604, doi:10.1175/2010JCLI3449.1.
Kollias, P., E. E. Clothiaux, M. A. Miller, B. A. Albrecht, G. L. Stephens, and T. P. Ackerman, 2007: Millimeter-wavelength radars: New frontier in atmospheric cloud and precipitation research. Bull. Amer. Meteor. Soc., 88, 1608, doi:10.1175/BAMS-88-10-1608.
Kumjian, M. R., S. A. Rutledge, R. M. Rasmussen, P. C. Kennedy, and M. Dixon, 2014: High-resolution polarimetric radar observations of snow-generating cells. J. Appl. Meteor. Climatol., 53, 1636–1658, doi:10.1175/JAMC-D-13-0312.1.
Marshall, J. S., 1953: Precipitation trajectories and patterns. J. Meteor., 10, 25–29, doi:10.1175/1520-0469(1953)010<0025:PTAP>2.0.CO;2.
Marshall, J. S., and K. L. S. Gunn, 1954: Measurement of snow parameters by radar. J. Meteor., 9, 322–327, doi:10.1175/1520-0469(1952)009<0322:MOSPBR>2.0.CO;2.
Mittermaier, P. M., J. R. Hogan, and J. A. Illingworth, 2004: Using mesoscale model winds for correcting wind-drift errors in radar estimates of surface rainfall. Quart. J. Roy. Meteor. Soc., 130, 2105–2123, doi:10.1256/qj.03.156.
Myagkov, A., P. Seifert, U. Wandinger, J. Bühl, and R. Engelmann, 2016: Relationship between temperature and apparent shape of pristine ice crystals derived from polarimetric cloud radar observations during the ACCEPT campaign. Atmos. Meas. Tech., 9, 3739–3754, doi:10.5194/amt-9-3739-2016.
Oude Nijhuis, A. C. P., F. J. Yanovsky, O. A. Krasnov, C. M. H. Unal, H. W. J. Russchenberg, and A. Yarovoy, 2016: Assessment of the rain drop inertia effect for radar based turbulence intensity retrievals. Int. J. Microwave Wireless Technol., 8, 835–844, doi:10.1017/S1759078716000660.
Oue, M., M. R. Kumjian, Y. Lu, Z. Jiang, E. E. Clothiaux, J. Verlinde, and K. Aydin, 2015: X-band polarimetric and Ka-band Doppler spectral radar observations of a graupel-producing Arctic mixed-phase cloud. J. Appl. Meteor. Climatol., 54, 1335–1351, doi:10.1175/JAMC-D-14-0315.1.
Pruppacher, H. R., and D. J. Klett, 1996: Microphysics of Clouds and Precipitation. Atmospheric and Oceanographic Sciences Library, Vol. 18, Springer, 976 pp.
Shupe, M. D., and Coauthors, 2008: A focus on mixed-phase clouds: The status of ground-based observational methods. Bull. Amer. Meteor. Soc., 89, 1549–1562, doi:10.1175/2008BAMS2378.1.
Unal, C. M. H., 2015: High-resolution raindrop size distribution retrieval based on the Doppler spectrum in the case of slant profiling radar. J. Atmos. Oceanic Technol., 32, 1191–1208, doi:10.1175/JTECH-D-13-00225.1.
Unal, C. M. H., and D. N. Moisseev, 2004: Combined Doppler and polarimetric radar measurements: Correction for spectrum aliasing and nonsimultaneous polarimetric measurements. J. Atmos. Oceanic Technol., 21, 443, doi:10.1175/1520-0426(2004)021<0443:CDAPRM>2.0.CO;2.
Unal, C. M. H., Y. Dufournet, T. Otto, and H. Russchenberg, 2012: The new real-time measurement capabilities of the profiling TARA radar. Proc. Seventh European Conf. on Radar in Meteorology and Hydrology (ERAD 2012), Toulouse, France, Météo-France, 199 SP. [Available online at http://www.meteo.fr/cic/meetings/2012/ERAD/short_abs/SP_388_sh_abs.pdf.]
Yuter, S. E., and R. A. Houze, 2003: Microphysical modes of precipitation growth determined by S-band vertically pointing radar in orographic precipitation during MAP. Quart. J. Roy. Meteor. Soc., 129, 455–476, doi:10.1256/qj.01.216.