Polarimetric Radar Observation of the Melting Layer in a Winter Precipitation System Associated with a SouthCoast Cyclone in Japan

In this study, we describe the spatial distribution of the melting layer (ML) in a winter stratiform precipitation system associated with a south-coast cyclone (SCC) on 30 January 2015 over the Kanto Plain, Japan, using an X-band polarimetric radar at Funabashi operated by the Ministry of Land, Infrastructure, Transport and Tourism (MLIT). The detailed horizontal distribution of surface precipitation types based on weather reports from citizens provided by Weathernews Inc. (WNI reports) was also investigated in relation to the ML structure. Surface precipitation in the Kanto region started with rain and then changed to snow around Tokyo. According to WNI reports, a large dry snow area formed around Tokyo by 0900 Japan Standard Time (JST; UTC + 9 hours), whereas surface rainfall continued in the southeast of the Kanto Plain (most of Chiba and the southern part of Kanagawa). A boundary line between the surface dry snow and rain areas became clear in the eastern part of Kanagawa and the northwestern part of Chiba. This boundary then gradually moved inland. Polarimetric ML signatures suggesting the presence of melting snow were continuously observed above the rainfall area in the southeast of the Kanto Plain. The polarimetric ML signatures, on the other hand, approached the ground near the surface dry snow–rain boundary while the surface snowfall was predominant around Tokyo. During the mature snowfall period around Tokyo, the ML vertically extended below 1 km above sea level (ASL) near the surface dry snow–rain boundary, which indicates the presence of a local horizontal temperature gradient and a surrounding ~ 0°C near-isothermal layer. It is suggested that this vertically extending ML coincided with the edge of a cold air mass in the lower atmosphere, which often forms during snowfall associated with SCCs in the Kanto region.


Introduction
The Kanto region in Japan typically receives little snow, especially in the plain part (i.e., the Kanto Plain).However, snowfall events do occur about 10 times a year, sometimes resulting in the accumulation of snow on the ground.Even a few centimeters of snow can have a significant impact on traffic, logistics, and the daily lives of people in this region because the urban infrastructure is not fully prepared for snowfall.Techniques to monitor the area of snowfall and distinguish it from the area of rainfall are therefore required to adapt social activities in the region to the occasional snowfall.
Snowfall events on the Kanto Plain are often associated with winter extratropical cyclones (Yamamoto 1984), so-called south-coast cyclones (SCCs), that pass by the southern coast of Japan (e.g., Takano 2002).Snowfall from SCCs commonly occurs in the stratiform precipitation region on the north side of a warm front (Yamamoto 1984;Tomiyama 2001;Araki et al. 2015).Whether snow particles aloft arrive at the ground without melting depends mainly on the temperature in the lower atmosphere.One of the conventional indicators of surface snowfall over the plain region is a temperature below −6°C at 850 hPa (~ 1.5 km above sea level (ASL)).However, higher temperatures, up to ~ 0°C at 850 hPa, can accompany snowfall on the Kanto Plain during SCCs (Yamamoto 1984;Tomiyama 2001).A more important factor for snowfall in this region is the vertical temperature profile below 850 hPa.The formation of shallow cold air masses over the Kanto Plain likely contributes to vertical thermodynamic profiles that favor snowfall at the ground.Northerly advection of cold air at low levels and/or diabatic cooling related to the melting and evaporation of precipitation particles have been considered as possible causes of shallow cold air-mass formation (Yamamoto 1984;Tomiyama 2001).However, the observation of these thermodynamic processes with sufficient spatiotemporal resolution is difficult, and this is the main limitation in real-time monitoring and forecasting of snowfall in this region.
One of the dominant approaches to detect the spatial distribution of precipitation types is polarimetric radar observation.Generally, polarimetric signatures such as differential reflectivity (Z DR ), the correlation coefficient between horizontal and vertical polarization signals ( ρ hv ), and the specific differential phase (K DP ) do not show a crucial difference between dry snow and rain areas in regions with moderate values of radar reflectivity (Z h ) (e.g., Ryzhkov et al. 2005).However, some polarimetric measurements show robust signatures that indicate the presence of a melting layer (ML; e.g., Zrnić et al. 1993;Brandes and Ikeda 2004;Tabary et al. 2006;Giangrande et al. 2008;Shusse et al. 2011), which is usually located between dry snow and rain areas and is associated with melting aggregated snow in stratiform precipitation.For example, ρ hv decreases markedly in regions with melting hydrometeors with a wide distribution of axis ratios, shapes, and canting angles.A bright band (BB) in Z h and a peak of Z DR in their vertical distributions are also common features of an ML in stratiform precipitation.Thus, developing a method to distinguish areas of snowfall from those of rainfall by taking advantage of polarimetric ML signatures is warranted.
Some previous studies have proposed algorithms for automatic ML detection using polarimetric radar measurements.For example, Brandes and Ikeda (2004) estimated the freezing (0°C) level with a horizontal spacing of 5 km using statistically modeled vertical profiles of polarimetric ML signatures of Z h , ρ hv , and the linear depolarization ratio (LDR).Tabary et al. (2006) proposed an ML identification algorithm based solely on the profiles of ρ hv , which assumes the height and thickness of the ML to be uniform within the radar observation area.Giangrande et al. (2008) and Boodoo et al. (2010) estimated the ML top and bottom as a function of azimuth using radial profiles of Z h , Z DR , and ρ hv .These algorithms are premised on some degree of horizontal uniformity of ML.However, spatially variable ML structures such as multiple MLs (Ikeda et al. 2005) and rapidly changing ML heights (Boodoo et al. 2010) have been observed in association with complex thermodynamic structures during frontal passages.The Kanto Plain is surrounded by mountains on the north and west sides and by the sea on the south and east sides (Fig. 1a).Snowfall on the Kanto Plain accompanied by SCCs is thought to depend on the local-scale distribution of the thermodynamic field, such as a shallow cold air mass formed on the plain, as mentioned above.However, the spatial distributions of ML structures in snowfall systems in this region have not been reported.It is important to know the spatiotemporal distribution of the ML to understand snowfall processes.In addition, this information will be indispensable to efforts to establish automatic ML detection technologies.Observational knowledge of ML structures in the winter precipitation systems associated with SCCs must be enhanced by conducting individual case studies.
One difficulty in the detection of polarimetric ML signatures in winter precipitation systems arises from their proximity to the ground.An ML that is close to the ground is often missed at far radar ranges, because it tends to be located below the height at the lowest elevation angle.Roughness of the vertical resolution of observation data and beam broadening at far ranges are also inconvenient for detecting thin MLs.In addition, ground-clutter contamination is likely to occur at low altitudes.In Japan, an operational X-band polarimetric radar network has been deployed by the Ministry of Land, Infrastructure, Transport and Tourism (MLIT) since 2010 (Maesaka et al. 2011;Godo et al. 2014) as part of the eXtended RAdar Information Network (XRAIN).Several XRAIN X-band polarimetric radars are densely installed in the Kanto region.The possibility that these radars might capture ML structures in winter precipitation systems associated with SCCs is worthy of exploration.It is also worth verifying the potential that the spatial distribution of ML observed by polarimetric radars can be useful for discriminating between snowfall and rainfall areas on the ground.
The purpose of this paper is to describe the spatial distribution of the ML, which changed dramatically in a winter precipitation system associated with an SCC on 30 January 2015 over the Kanto Plain, using an X-band polarimetric radar.Weather reports from citizens, including categories associated with surface precipitation type, provided by Weathernews Inc. (WNI) are utilized to verify the winter ML detection by the polarimetric radar and to evaluate the potentially discriminating surface precipitation types by polarimetric ML signatures.

Observation and utilization data
In this study, an X-band polarimetric radar operated by MLIT as part of XRAIN located at Funabashi (hereinafter the Funabashi radar) is used to investigate the ML structure in the precipitation system on 30 January 2015.The Funabashi radar was chosen to capture clear images of the spatial distribution of the ML during the snowfall period.Figure 1b shows the location (35°41′45″N, 140°0′26″E; altitude: 86.5 m ASL) and observation area of the Funabashi radar.The Funabashi radar covers a vast area of the Kanto Plain (Figs. 1a, b).The system characteristics of the Funabashi radar are listed in Table 1.The Funabashi radar was operated with 15 plan position indicator (PPI) scans at 12 elevation angles of 0.7°, 1.6°, 2.7°, 3.8°, 5.1°, 6.5°, 8.1°, 9.9°, 11.9°, 14.2°, 16.9°, and 20.0° every 5 min.The PPI scans alternately contained the second-and third-lowest elevation angle per  scans were conducted by the Funabashi radar, the radar data are shown in "pseudo-RHI" displays, constructed by extracting the corresponding azimuth angle from a series of PPI scans with different elevation angles in 5 min in the analysis.In drawing pseudo-RHI displays, observation data at a certain elevation represents the half area of its upper and lower elevation intervals.For the data of the lowest (highest) elevation angle, twice the area of the half of its upper (lower) elevation interval is applied.
For surveying the spatial distribution of ML in stratiform rainfall, Z h alone provides useful information to some extent as mentioned in the Introduction.However, a BB signature in Z h tends to become weaker when the ML is accompanied by relatively strong rainfall, such as embedded convective rain, beneath it (Shusse et al. 2011).ρ hv shows a more robust signa- ture against melting snow particles.Although other polarimetric parameters such as Z DR and differential propagation phase (Φ DP ) may also be useful, their contribution to ML identification is less than that of ρ hv , as discussed in previous studies (Brandes and Ikeda 2004;Tabary et al. 2006).Therefore, ρ hv is used in addition to Z h in this study.Z h is computed using the dielectric factor of water.ρ hv was corrected for the signal-to-noise ratio (SNR) using the method of Shusse et al. (2009).
Precipitation intensity data observed by the Japan Meteorological Agency (JMA) radars at 2 km ASL with 1 km horizontal resolution are used to show a wide-area distribution of precipitation associated with the SCC.The surface temperature and horizontal wind during the event were investigated using Automated Meteorological Data Acquisition System (AMeDAS) data every 10 min.The precipitation amount, relative humidity, visual weather observations, and snow depth at the Tokyo AMeDAS site (see Fig. 1b for the location), the JMA Tokyo Regional Headquarters, were also used in the analysis.An upper-air sounding launched at Tateno (see Fig. 1b for the location) was used to examine the ambient conditions of the target area.Surface precipitation types were confirmed by weather reports provided by WNI (WNI reports).The WNI reports were sent from WNI's "Weather Supporters".They post current weather conditions and their geolocation through the WNI's app.Although the categories of WNI reports include sunny, cloudy, rain, wet snow, dry snow, and graupel, only the reports related to the precipitation types, that is, (1) rain, (2) wet snow, (3) dry snow, and (4) graupel, were used in the analysis.These precipitation categories of the WNI reports are subdivided according to their intensity and/ or way of falling.However, these subcategories were not considered in this study.

Environment and surface weather conditions
On 30 January 2015, precipitation events over Japan were dominated by an extratropical cyclone moving east-northeastward along the southern coast of Japan.The surface weather map at 0900 Japan Standard Time (JST; UTC + 9 hours) on 30 January 2015 shows a low-pressure center and associated fronts on the south of the Japanese archipelago (Fig. 2).The temperature at 850 hPa from the upper-air sounding at Tateno was −6.1°C at 2100 JST on 29 January.It rose to −2.3°C at 0900 JST on 30 January when snowfall was observed at the Tokyo AMeDAS site, which is thought to have been caused by the approach of the low-pressure center (Fig. 2).
Figure 3 shows the temporal change in the horizontal distribution of precipitation intensity at 2 km ASL observed by the JMA radars.The precipitation echo started to expand over the Kanto region at around 0200 JST as the warm front approached and continued until the evening of that day (Figs.3a -d).
At the Tokyo AMeDAS site, rain initiation was recorded by visual observations at 0240 JST (Fig. 4).Transient wet snow and dry snow were recorded from 0325 to 0520 JST, dry snow was recorded from 0520 to 1045 JST, wet snow was recorded from 1045 to 1205 JST, and then rain was recorded after 1205 JST.The surface temperature at the Tokyo AMeDAS site was 4.5 -4.6°C before the rainfall initiation at 0240 JST and began to decrease at around 0250 JST (Fig. 4).The surface temperature showed a rapid drop of 3.0°C from 0240 JST (4.5°C) to 0440 JST (1.5°C) and reached the lowest value at 0730 JST (0.6°C).
Temperatures of ≤ 0.7°C persisted from 0630 to 1120 JST and then gradually increased.The maximum snow depth at the Tokyo AMeDAS site was 3 cm at 1100 JST, approximately coincident with the maximum 1 h precipitation of 3.5 mm from 1000 to 1100 JST and the transition from dry snow to wet snow at 1045 JST (Fig. 4).Relative humidity was ~ 55 % before the initiation of surface rainfall and increased rapidly to > 90 % at 0500 JST as surface temperatures decreased (Fig. 4).
Figure 5 shows the horizontal distribution of WNI reports at 0600, 0900, and 1200 JST.There were numerous reports of rain, wet snow, and dry snow, with few reports of graupel in the Kanto region on 30 January 2015.At 0600 JST, many reports of dry snow had already spread to most of Tokyo and the northern part of Kanagawa, but some reports of rain and wet snow were also found within the area of primarily dry snow reports (Fig. 5a).Rain reports dominated most parts of Chiba and the southern part of Kanagawa.At 0900 JST, the position of the southeastern edge of the dry snow reports was almost the same as that at 0600 JST (Fig. 5b).Meanwhile, the number of rain and wet snow reports in the area where dry snow reports were predominant was significantly lower than that at 0600 JST, and the boundary line between rain and dry snow areas became clearer.This means that a stable snowy area had formed in and around Tokyo by 0900 JST.Several reports of wet snow were seen near the boundary between rain and dry snow areas at 0900 JST.Then, the position of the boundary gradually moved inland.The rain reports had dominated the surrounding area of the Tokyo AMeDAS site by 1200 JST (Fig. 5c).This is consistent with the visual weather observations at the Tokyo AMeDAS site shown in Fig. 4. From the distributions of WNI reports in Fig. 5 and the AMeDAS observation data in Fig. 4, the snowfall around Tokyo on that day was considered to be in an early stage at 0600 JST, in a mature stage at 0900 JST, and in a later stage at 1200 JST. Figure 6 shows the horizontal distribution of surface temperature and horizontal wind from AMeDAS.Surface temperature distributions did not differ much among 0600, 0900, and 1200 JST.The temperature in the snowfall area around Tokyo was approximately 0 -1°C at these times (Fig. 5).The horizontal wind direction was continuously dominated by northerly to north-northwesterly winds during the snowfall period in this area.
The Funabashi radar was located near the boundary line between surface rain and dry snow areas during this snowfall period (Figs.1b, 5), enabling the capturing of clear images of the spatial distribution of the ML.In the following section, the spatial distribution of ML structures during this snowfall period in the precipitation system is investigated using the polarimetric radar measurements of the Funabashi radar.

Radar signature of the ML
The PPI displays of Z h and ρ hv at 1.6° elevation ob- served by the Funabashi radar at 0600, 0900, and 1200 JST are presented in Fig. 7 to show the overall spatial distribution of the ML during the snowfall period around Tokyo.At 0600 JST, a BB signature in Z h was seen at a range of ~ 20 km to the south and southwest of the radar, indicating that it was raining on the ground in this area, and much closer to the north and northwest of the radar (Fig. 7a).The local minimum value of ρ hv in each azimuth direction was observed slightly inside the Z h maximum in the BB signature (Fig. 7b), which is consistent with the characteristics of the ML signatures dominated by melting aggregated snow in stratiform precipitation systems (e.g., Brandes and Ikeda 2004;Shusse et al. 2011).The difference in the distance of the polarimetric ML signatures in Z h and ρ hv for each azimuth direction on the PPI displays indicates the changes of the ML height around the radar observation site.These changes became more pronounced at 0900 JST.The polarimetric ML signature in Z h was observed at a range of 20 -40 km from the northeast through the south to the west of the radar (Fig. 7c), and that in ρ hv was observed at a range of 20 km or more in the region (Fig. 7d).This indicates that the ML height at 0900 JST was higher than that at 0600 JST there.On the other hand, the ML signatures on the northwest side of the radar were distributed in an almost straight line and were closer to the radar on the PPI displays.The position of this line-shaped ML signature on the PPI displays at 0900 JST roughly coincided with the boundary between surface rain and dry snow areas shown in Fig. 5b.This suggests that the ML on the northwestern side of the radar, where the elevation of the ML was lower than in other regions, was nearly connected to the surface.At 1200 JST, the ML signatures in Z h and ρ hv showed a nearly circular shape and were observed at farther distances than those at 0900 JST in all azimuth directions on the PPI displays at 1.6° elevation, which indicates that the ML had increased in altitude around the Funabashi radar site.
The pseudo-RHI displays of Z h and ρ hv around the radar site are shown in Fig. 8 and reveal the vertical structure of the ML in a direction from the southeast to the northwest, almost perpendicular to the boundary between the surface dry snow and rain reporting areas, at 0600, 0900, and 1200 JST.At 0600 JST, the Z h BB signature was observed slightly above 0.5 km ASL at a distance of approximately 0 -20 km to the southeast of the radar and near the ground at a distance of 0 -5 km to the northwest of the radar (Fig. 8a).The region with ρ hv ≤ 0.94, which is highly likely to be composed of wet snow (e.g., Kouketsu et al. 2015), around the Z h BB signature clearly shows that the height of the ML was nearly 0.5 km ASL to the southeast of the radar and gradually got lower toward the northwest and appeared to have reached the surface just around the radar site (Figs.8a, b).These polarimetric ML signatures near the surface were roughly coincident with the position of the southeastern edge of the dry snow reporting area shown in Fig. 5a.The WNI reports confirm that rain and dry snow dominated near the ground on the southeast and northwest sides of the grounding location of the ML indicated by the pseudo-RHI displays at 0600 JST, respectively (Figs. 5a, 8a, b).Some reports of rain and wet snow in the area of primarily dry snow reports at 0600 JST (Fig. 5a), on the northwest side of the grounding location of the ML (Figs. 8a, b), suggest that some smaller snow particles were melted, but most of the snow particles remained unmelted in the immediate vicinity of the ground with temperatures slightly above 0°C (Fig. 6a).Note that the dispersion in ρ hv in the area of weak echo at far range is caused by low SNR (Figs. 7,8).A slightly low ρ hv of the lowest elevation angle is caused by ground-clutter contamination (Fig. 8).At 0900 JST, the height of ML signatures in Z h and ρ hv over the radar site rose to ~ 1 km ASL (Figs. 8c,  d), which is indicative of increasing temperatures at higher altitudes, as detailed later in Section 5.1.It is noted that the polarimetric ML signatures extended toward the ground so that the ML was almost upright at 1 km ASL or less around the grounding location at 0900 JST.The position where the ML reached the surface moved only a little during 0600 -0900 JST and was 5 -10 km northwest of the radar site at 0900 JST, coincident with the boundary between surface rain and dry snow reporting areas shown in Fig. 5b.At 1200 JST, the height of ML signatures further increased to ~ 1.4 km ASL (Figs. 8e, f).The grounding position of the ML moved northwestward and was at ~ 30 km northwest of the radar on the pseudo-RHI displays, which corresponds to the boundary between surface rain and dry snow reporting areas at 1200 JST (Fig. 5c).

Spatiotemporal variation of the ML during the snowfall period
As described in Section 4, polarimetric ML signatures suggesting the presence of melting aggregated snow were continuously observed above the rainfall area in the southeast of the Kanto Plain and descended toward the ground around the boundary between surface rain and dry snow areas during the snowfall period on 30 January. Figure 9 summarizes the temporal change in the spatial ML distribution, which is depicted on the basis of the region with ρ hv ≤ 0.94, in a direction perpendicular to the boundary between the surface dry snow and rain reporting areas, during the snowfall period.The average height of the ML above the rainfall area gradually increased from ~ 0.5 km ASL at 0600 JST to ~ 1.4 km ASL at 1200 JST (Fig. 8), which indicates an increase in temperature in the upper atmosphere.In contrast, no marked increase in surface temperature was observed at the AMeDAS sites in the rainfall area during the period (Fig. 6).For example, the surface temperature at the Chiba AMeDAS site, which was located in the rainfall area, was 3.1°C at 0600 JST, 2.4°C at 0900 JST, and 2.5°C at 1200 JST.It is reasonable to think that the rise in temperature in the upper atmosphere that caused the elevation of the ML was due to the approach of the low-pressure center and its associated warm front.Another remarkable feature is the ML structure, which was almost upright below 1 km ASL at 0900 JST during the mature period of snowfall near the boundary line between surface rain and dry snow areas.In other words, a locally deep ML was observed near the boundary, suggesting that the slow melting of snow particles occurred in a near-isothermal layer with temperatures only slightly above 0°C.Generally, MLs in stratiform rainfall spread in the horizontal direction because of the relatively uniform thermodynamic fields with a vertical temperature gradient.In the current case, the vertically extending ML at 0900 JST implies the presence of an apparent local horizontal temperature gradient without a significant vertical temperature gradient below 1 km ASL.Marked horizontal temperature gradients can occur around warm frontal surfaces.Some observations of multiple MLs have been reported in such synoptic situations (Ikeda et al. 2005;Boodoo et al. 2010).In this case, however, multiple structure of the ML was not observed during the event.Horizontal temperature gradients at low altitudes can also occur inside and around locally formed cold air masses, which commonly form during snowfall associated with SCCs over the Kanto Plain (Yamamoto 1984;Tomiyama 2001).Unfortunately, there is no observation data to verify the presence of cold air masses at low altitudes for this event.We speculate, however, that the formation of a stable snowy area in and around Tokyo from the early stage to the mature stage of the snowfall (Fig. 5) is one indication of the cold air mass on the inland side of the vertically extending ML.It is reasonable to consider that the vertically extending ML coincided with the edge region of this cold air mass.

Possibility for monitoring boundaries between
surface dry snow and rain areas with polarimetric radars Surface precipitation types have often been diagnosed on the basis of surface meteorological elements when creating and/or forecasting their distribution maps.In particular, some algorithms using surface temperature and relative humidity have been proposed to discriminate among rain, dry snow, and wet snow (Matsuo et al. 1981).However, surface precipitation types also depend on the atmospheric conditions above, not only on surface meteorological elements (e.g., Matsuo et al. 1985).Thus, precipitation-type discriminations that are strongly dependent on surface meteorological elements are not always accurate.
In this study, the spatial distribution of an ML associated with the winter SCC over the Kanto Plain observed in the vicinity of an operational X-band polarimetric radar was demonstrated.The position where the ML descended toward the ground was nearly coincident with the boundary between surface dry snow and rain areas, as shown in Figs. 5, 7, and 8.These results suggest that the detection of the spatial distribution of MLs, such as the spatial structure of grounding ML, using polarimetric radars is an effective means to monitor the surface snowfall and rainfall areas for winter stratiform precipitation systems in this region.In addition, there is no clear difference in polarimetric measurements such as Z DR , ρ hv , and K DP between stratiform rain and snow regions.Therefore, many of the previously proposed methods for hydrometeor classification using polarimetric radars relied on external data such as temperature from upper-air sounding observations or from numerical modeling to distinguish stratiform rain and snow (e.g., Kouketsu et al. 2015).The combination of ML detection techniques based solely on polarimetric radar information and some simple methods to judge rain and snow sides, like higher and lower temperature sides, can contribute to the development of practical algorithms of hydrometeor classification for winter stratiform precipitation systems.
When adapting the results of precipitation-type identification by polarimetric radar observations to surface precipitation types, the altitude difference between the radar observation and the ground may pose a problem.However, operational X-band polarimetric radars are densely arranged throughout Japan (Maesaka et al. 2011;Godo et al. 2014) and are likely to have the ability to detect MLs near the ground.In the future, it would be beneficial to develop a method to automatically differentiate rain and dry snow areas using ML signatures observed by the operational polarimetric radar network.

Summary
In this study, we described the spatial distribution of the melting layer (ML) in a winter stratiform precipitation system associated with a south-coast cyclone (SCC) on 30 January 2015 over the Kanto Plain, using the Funabashi radar, which is an X-band polarimetric radar operated by the Ministry of Land, Infrastructure, Transport and Tourism (MLIT).The detailed horizontal distribution of surface precipitation types mainly based on the weather reports provided by Weathernews Inc. (WNI reports) was also investigated to confirm the spatial distribution of the ML and to evaluate the potentially discriminating surface precipitation types by polarimetric ML signatures.
Stratiform precipitation started to expand over the Kanto region at around 0200 JST.Surface precipitation around Tokyo started with rain before 0300 JST, which then changed to transient wet snow and dry snow.By 0900 JST, a stable snowy area had formed in and around Tokyo.A boundary line between this dry snow area and the rain area in the southeast of the Kanto Plain, extending from the southwest to the northeast, became clear in the eastern part of Kanaga-wa and the northwestern part of Chiba.The boundary line between the surface rain and dry snow areas then gradually moved inland.
Polarimetric ML signatures suggesting the presence of melting aggregated snow were continuously observed above the rainfall area in the southeast of the Kanto Plain and approached the surface near the boundary between the surface rain and dry snow areas during the snowfall period.The typical height of the ML above the rainfall area gradually increased from ~ 0.5 km ASL at 0600 JST to ~ 1.4 km ASL at 1200 JST along with the approach of a low-pressure center and its associated warm front.
A remarkable feature of the ML at 0900 JST during the mature snowfall period was its vertical extension below 1 km ASL near the boundary between the surface rain and dry snow areas.This ML structure indicates the presence of a local horizontal temperature gradient and a near-isothermal layer with temperatures of ~ 0°C surrounding the vertically extending ML.It is suggested that the vertically extending ML was associated with a cold air mass in the lower atmosphere, which often forms during snowfall associated with SCCs over the Kanto Plain, and was located at the edge of the cold air mass in the lower atmosphere.
This study demonstrates that an operational X-band polarimetric radar in the Kanto region captured the spatial distribution of the ML, which changed dramatically in the winter stratiform precipitation system associated with an SCC.It is also suggested that the detection of the ML using operational polarimetric radars is effective for determining surface rainfall and snowfall areas.The utilization of dense weather reports from citizens was very useful to support this polarimetric ML observation.In the future, it would be beneficial to develop automatic detection algorithms for winter MLs and techniques for monitoring snowfall in the Kanto region using the dense polarimetric radar observation network.Furthermore, future efforts to derive a more detailed three-dimensional ML structure in the winter precipitation system utilizing the polarimetric radar network are expected to lead to the elucidation of the unique snowfall process in this region.
tion Exploration System.The Weather Reports were provided by Weathernews Inc.

Fig. 1 .
Fig. 1.Locations of observation sites.(a) Map of the Kanto Plain.The gray shading shows the altitude of topography.The broken-line square indicates the area in Fig. 1b.(b) Location of the X-band polarimetric radar site at Funabashi (+), the AMeDAS sites at Tokyo and Chiba (•), and the upper-air sounding station at Tateno (■).The circle encloses the observation area of the radar at Funabashi (80 km in radius).The location of this map is shown in Figs.1a and 3.

Fig. 7 .
Fig. 7. PPI displays at 1.6° elevation showing (a) Z h and (b) ρ hv at 0600 JST, (c) Z h and (d) ρ hv at 0900 JST, and (e) Z h and (f) ρ hv at 1200 JST.The solid lines indicate the locations of the pseudo-RHI displays (azimuth directions of 150° and 330°) in Fig. 8. Range rings are every 20 km.

Fig. 8 .
Fig. 8. Pseudo-RHI displays along azimuth 330° showing (a) Z h and (b) ρ hv at 0600 JST, (c) Z h and (d) ρ hv at 0900 JST, and (e) Z h and (f) ρ hv at 1200 JST.Locations of the pseudo-RHI displays are shown in Fig. 7.The contours of ρ hv of 0.94 around the polarimetric ML signatures are also shown in the left panels (a, c, e) for reference.

Fig. 9 .
Fig. 9. Spatial ML distributions at 0600 JST (green), 0900 JST (red), and 1200 JST (yellow) in the direction almost perpendicular to the edge of the surface dry snow area.The ML distributions are traced on the basis of the regions with ρ hv ≤ 0.94.The level of 850 hPa observed at Tateno at 0900 JST (1.48 km ASL) is also shown.

Table 1
. System characteristics of the MLIT X-band polarimetric radar at Funabashi.minute.Although no range height indicator (RHI)