Observational Study on the Supercooled Fog Droplet Spectrum Distribution and Icing Accumulation Mechanism in Lushan, Southeast China

ADVANCES IN ATMOSPHERIC SCIENCES, VOL. 36, JANUARY 2019, 29–40

• Original Paper • Observational Study on the Supercooled Fog Droplet Spectrum Distribution and Icing Accumulation Mechanism in Lushan, Southeast China

Tianshu WANG1, Shengjie NIU∗1,2,3, Jingjing LU¨ 1,4, and Yue ZHOU5

1School of Atmospheric Physics, Nanjing University of Information Science and Technology, Nanjing 210044, China 2Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science and Technology, Nanjing 210044, China 3Nanjing Tech University, Nanjing 211816, China 4Key Laboratory for Aerosol–Cloud–Precipitation of China Meteorological Administration, Nanjing University of Information Science and Technology, Nanjing 210044, China 5Wuhan Regional Climate Center, Wuhan 430074, China

(Received 2 March 2018; revised 29 May 2018; accepted 9 July 2018)

ABSTRACT A fog monitor, hotplate total precipitation sensor, weather identifier and visibility sensor, ultrasonic wind speed meter, an icing gradient observation frame, and an automated weather station were involved in the observations at the Lushan Meteorological Bureau of Jiangxi Province, China. In this study, for the icing process under a cold surge from 20–25 January 2016, the duration, frequency, and spectrum distribution of agglomerate fog were analyzed. The effects of rain, snow, and supercooled fog on icing growth were studied and the icing and meteorological conditions at two heights (10 m and 1.5 m) were compared. There were 218 agglomerate fogs in this icing process, of which agglomerate fogs with durations less than and greater than 10 min accounted for 91.3% and 8.7%, respectively. The average time interval was 10.3 min. The fog droplet number concentration for sizes 2–15 µm and 30–50 µm increased during rainfall, and that for 2–27 µm decreased during snowfall. Icing grew rapidly (1.3 mm h−1) in the freezing rain phase but slowly (0.1 mm h−1) during the dry snow phase. Intensive supercooled fog, lower temperatures and increased wind speed all favored icing growth during dry snow (0.5 mm h−1). There were significant differences in the thickness, duration, density, and growth mechanism of icing at the heights of 10 m and 1.5 m. Differences in temperature and wind speed between the two heights were the main reasons for the differences in icing conditions, which indicated that icing was strongly affected by height.

Key words: cold surge, microstructure of supercooled fog, icing gradient observation, growth rate of icing Citation: Wang, T. S., S. J. Niu, J. J. Lu,¨ and Y. Zhou, 2019: Observational study on the supercooled fog droplet spectrum distribution and icing accumulation mechanism in Lushan, Southeast China. Adv. Atmos. Sci., 36(1), 29–40, https://doi.org/ 10.1007/s00376-018-8017-6.

1. Introduction systems, physical mechanisms, monitoring, and early- warning methods. Wang (2011) analyzed the temporal and The phenomenon in which glaze and rime condense or spatial distributions of ice-freezing days in China from 1954 wet snow freezes on wires is known as wire icing (China to 2009 and pointed out that China’s heavy icing areas were Meteorological Administration, 1979), which can seriously mainly in northern Xinjiang, southern Shaanxi, the central damage the normal operation of transmission lines and af- part of northeastern China, the eastern part of northern China, fect productivity and daily life. From January to February Qinling, northeastern Yunnan, Guizhou, and other places, pri- in 2008, four large-scale freezing rain and snow weather marily from November to late March. Li et al. (2015) used events in southern China caused huge economic losses, caus- the observational data of China’s civil aviation airports from ing widespread concern in the research community. 2011 to 2013 to analyze temporal and spatial distributions Many in-depth studies have been conducted regarding and the weather conditions of freezing rain, freezing drizzle, icing in terms of the temporal and spatial distribution of and freezing fog. They pointed out that freezing rain, freez- freezing weather disasters, circulation conditions, weather ing drizzle, and freezing fog appeared in January with frequencies as high as 55%, 67%, and 38%, respectively. Other ∗ Corresponding author: Shengjie NIU studies (Ding et al., 2008; Wang et al., 2008a, 2008b; Yang Email: [email protected] et al., 2008) have pointed out that a large area of freezing

© Institute of Atmospheric Physics/Chinese Academy of Sciences, and Science Press and Springer-Verlag GmbH Germany, part of Springer Nature 2019 30 SUPERCOOLED FOG AND ICING IN LUSHAN VOLUME 36 weather is usually accompanied by a unique circulation pat- lyzed the features of fog droplet and precipitation particle tern. The center of the polar vortex located in the northern spectra under three kinds of weather. Their results showed part of the eastern hemisphere, a quasi-steady blocking con- that the average concentration of fog droplets was 224 cm−3 dition at mid and high latitudes, and an active southern branch on rainy days, 181 cm−3 on rainy and snowy days, and 139 of circulation in the low-latitude area were the main synop- cm−3 on foggy days. Niu et al. (2012) studied the microphys- tic causes for freezing rain and snow disasters in southern ical characteristics of fog droplet and raindrop spectra during China at the beginning of 2008. Zeng et al. (2008), Li et al. the formation of icing by using the observation data of ic- (2009), and Tao et al. (2012) studied the stratification char- ing at Enshi Radar Station in Hubei Province of China in the acteristics of freezing rain and found that a stable, thicker winters of 2008/09 and 2009/10. They pointed out that there melting layer was the direct cause for freezing rain encom- was a positive correlation (0.62) between liquid water con- passing a wide area. The main mechanisms for freezing rain tent in the rain/fog and icing growth rate. Zhou et al. (2013) formation were ice-crystal melting and supercooled warm- studied the effect of freezing drizzle on icing growth during rain processes (Tao et al., 2012). Zhou et al. (2012) and Liu fog processes and pointed out that the direct contribution of and Niu (2016) studied real-time icing observational data of freezing drizzle to ice thickness was 14.5%. The inclusion of 500-kV high-voltage transmission lines in the Central China microphysical processes in these studies helps us further un- Power Grid in Hubei Province in the winters of 2008 and derstand wire icing accretion, as opposed to the initial studies 2009. Both temperatures for icing formation and shedding that only considered meteorological conditions. were lower than the temperature thresholds on the test cable. We conducted meteorological condition and microphysi- Icing formed easily when the temperature was about −2◦C, cal observations of icing accretion from December 2015 to relative humidity was greater than 95%, and wind speed was February 2016 at the Lushan Meteorological Bureau Ob- 0–1 m s−1. Jiang (1984) established the relationship between servatory, Southeast China. To study the effect of different the growth rate of icing and meteorological elements by using heights on icing, we included icing gradient observations. icing accretion observation data from the Lushan Cloud Test This study is a comprehensive analysis of the distribution of Station in China in 1978–81. Tan (1982) found that the ratio the supercooled fog droplet spectrum and the mechanism of of ice thickness at two different heights was a power func- icing growth in the icing process from 20–25 January 2016. tion of the ratio of the two heights. Jones (1998) proposed a model for freezing rain ice load, with parameters such as precipitation rate and wind speed. Makkonen (1989) proposed a 2. Observation instruments and data wet snow accretion model by using the wet-bulb temperature as the index to judge the occurrence of wet snow, consider- The observatory is located at the Lushan Meteorolog- ing visibility as an input. Other studies have applied and im- ical Bureau, Jiangxi Province, Southeast China (29.58◦N, proved this model (Makkonen and Wichura, 2010; Nygaard 115.98◦E; elevation: 1164.5 m). Lushan is situated be- et al., 2013). Drage and Hauge (2008) combined the MM5 tween Poyang Lake (China’s largest freshwater lake) and the model with an ice accretion model to simulate ice accretion Yangtze River. So, it is rich in moisture resources, and glaze, on the west coast of Norway. Musilek et al. (2009), Pytlak et rime, snow, and other weather phenomena often appear when al. (2010), and Hosek et al. (2011) combined the WRF model cold air intrudes in winter. with an ice accretion model to develop an ice accretion fore- The data measured by instruments and used in this study casting system. are listed in Table 1. The sensor head of the TPS-3100 hot- Luo et al. (2008) studied the microphysical characteris- plate total precipitation sensor has double discs, which pro- tics of cloud and fog during an icing period by analyzing the vide precipitation rate measurements of rain and snow as part cloud droplet spectrum and conventional meteorological data of the automatic weather observation. Its detection range is in a freezing area of Guizhou Province, China. They found 0–50 mm h−1 and the temporal resolution is 1 min. The OWI- that cloud droplets greater than 14 µm were an important fac- 430 weather identifier and visibility sensor measures precip- tor for icing growth. Jia et al. (2010) studied a mixed icing itation by detecting the optical irregularity induced by par- process on the basis of glaze conditions at Enshi Radar Sta- ticles falling through a beam of partially coherent infrared tion in China from 25 February to 4 March 2009 and ana- light, and it calculates visibility by determining the amount

Table 1. List of instruments used.

Instrument Model Manufacturer Measurement

Fog droplet spectrometer FM-100 DMT, USA Size range 2–50 µm, sampling frequency 1 Hz Hotplate total precipitation sensor TPS-3100 YES, USA Precipitation rate Weather identifier and visibility sensor OWI-430 OS, USA Visibility, weather phenomena Temperature and humidity probe HMP155 VAISALA, Finland Temperature, humidity Blade type anemometer Model 05103 R. M. YOUNG, USA Wind speed and direction Longwave and shortwave radiation meter CNR2 Kipp & Zonen, Netherland Short-wave and long-wave radiation JANUARY 2019 WANG ET AL. 31 of forward scattered light by particles. Its detection range for vortices, which were located to the east of Lake Baikal and visibility is 0.001∼10 km. The FM-100 fog droplet spectrom- over the ocean to the northeast of Japan. The low-pressure eter has been widely used in ice and freezing fog observations center to the east of Lake Baikal extended a horizontal trough (Gultepe et al., 2011, 2016). There is a 10-m tower on the westward. The horizontal trough rotated toward the south west side of the observatory, and wires were laid on the 10-m from 0800 LST 23 January to 0800 LST 24 January. The tower and on the 1.5-m icing frames under the tower. Icing shear line at 850 hPa shifted from northwest to southeast from growth was observed. There were sets of wires in the east– 20–23 January and brought rain and snow to the areas south west, north–south, and northeast–southwest directions on the of the Yangtze River. The surface level was controlled by a tower. Limited by observation and severe weather conditions, high-pressure system. With a cold front in front of the high- only the ice thickness data of the northeast–southwest wire pressure system shifting southward, the cold air affected most were used to study the icing accretion on the tower, repre- of the areas south of the Yellow River. The high-pressure cen- senting the average icing conditions of the wires (along east– ter weakened and split at 0800 LST 25 January, and the cold west and north–south). The icing frame under the tower was surge process ended. The icing process lasted for 102 h at the laid in accordance with “ground meteorological observation observation site of the Lushan Meteorological Bureau. norms”: one group was oriented along the east–west direc- The atmospheric stratification at the observation station tion, and the other group was oriented along the north–south during the icing process was obtained from the NCEP reanal- direction. The wire diameter is 26.8 mm. Ice diameter a (the ysis data (not shown). The temperature at 850 hPa was 0◦C maximum of accumulated ice on the cut surface perpendicu- from 0800 LST 19 to 0800 LST 20 January, before the cold lar to the wire, the wire diameter is included) and ice thick- front crossing. Relative humidity was below 90%, and the ness b (the maximum of the accumulated ice perpendicular to wind direction was mainly southeasterly and southwesterly the ice diameter on the cut surface of the wire) were measured from 900 to 500 hPa. When the cold front crossed from 20– hourly (China Meteorological Administration, 1979). 23 January, the low-level system was primarily composed of Through the observation during a cold surge, a more com- the northeasterly wind, and the wind speed increased. The plete and higher temporal resolution fog droplet spectrum, high-level system was primarily composed of the southwest- visibility, weather phenomena, and conventional meteorolog- erly wind. Temperatures dropped below 0◦C, and relative hu- ical elements could be obtained. For this study, we also used midity increased to greater than 90% from 900 to 500 hPa. MICAPS and NCEP reanalysis data. There was no warm layer above the observation site during The wire diameter is φ, the equivalent ice diameter is D, the rain and snow process (from 0815 LST 20 to 1000 LST the equivalent ice thickness is W, and the calculation formu- 23 January), as the stratification exhibited “a single layer las are as follows: structure” (Li et al., 2009). Primarily, a northwesterly wind √ resided from 900 to 500 hPa after 0000 LST 24 January. D = ab ; (1) The transport of cold air further reduced the temperature, but D − φ the weather turned out to be fine and the relative humidity W = . (2) 2 dropped below 50%. Cooling and humidification were conducive to icing growth and maintenance. The ice thickness mentioned later is the equivalent ice Rain and snow were sustained for a long time at the thickness. During an icing process, the ice on blade type Lushan Observatory during the cold surge. Rainfall and anemometer was removed from time to time, which resulted snowfall durations were 14 h and 60 h, respectively, and the in zero wind speed. Such(zero) data was removed during data entire process resulted in a 15◦C cooling. As shown in Fig. processing. Sometimes, non-zero wind data might be col- 1a, rain began at 0815 LST 20 January, and the precipitation lected when ice was not completely removed, which means rate (rain and snow were both measured using the hotplate that the observed wind speed data could be less than what the total precipitation sensor, collectively referred to as the pre- actual wind speed was. However, this is inevitable in freezing cipitation rate) began to decrease after 1800 LST 20 January. weather, and wind speed data still have a reference value. It started to snow at 2200 LST 20 January, and it snowed intermittently from 21–23 January, stopping at 1000 LST 23 −1 3. Cold surge and microstructure of super- January. The maximum precipitation rate was 6.1 mm h (at 1357 LST 20 January) during the entire process, with an cooled fog average of 0.2 mm h−1. The surface snow depth (Fig. 1b) Pressure ridges over the northern Pacific, northern At- continued to rise from 1000 LST 21 January, reached a max- lantic, and Taymyr Peninsula all extended poleward at 500 imum of 15 cm at 0500 LST 23 January, and then continued hPa at 0800 LST 19 January 2016. The Asian polar vortex to decline after the snow stopped, until the snow melted on strengthened and was stretched to the south, while the east- 25 January. ern Asian region underwent an inverted “Ω” circulation pat- Rain and snow occurred along with the emergence of tern. A pressure ridge over the Taymyr Peninsula joined with fog. Visibility (Fig. 1c) continuously declined from 5.9 km one over the northern Pacific at 0800 LST 21 January, and at 0800 LST 20 January. It was below 1 km at 1255 LST 20 a high-pressure center formed north of eastern Siberia. The January, which indicated the beginning of fog. The fog pro- Asian polar vortex stretched to the south, dividing into two cess ended when visibility rose above 1 km at 1039 LST 23 32 SUPERCOOLED FOG AND ICING IN LUSHAN VOLUME 36 °C

Fig. 1. Temporal variations (LST) of (a) precipitation rate, (b) snow depth, (c) visibility, (d) temperature at the 1.5-m height T1.5, (e) fog droplet number concentration N, (f) fog droplet mean diameter Dave, and (g) fog liquid water content (LWC).

January. The temperature at the 1.5-m height, T1.5 (Fig. 1d), glomerate fogs. Agglomerate fogs with durations less than declined below zero at 1636 LST 20 January; afterwards, 5 min had the highest appearance frequency (78.4%). Ag- the fog became supercooled. Fog appeared intermittently, glomerate fogs with durations less than and greater than 10 during which the fog droplet number concentration (N; Fig. min accounted for frequencies of 91.3% and 8.7%, respec- 1e), mean fog droplet diameter (Dave; Fig. 1f), and fog liq- tively. Consistent with the statistical results of Hodges and Pu uid water content (LWC; Fig. 1g) all fluctuated significantly, (2016), the frequency of agglomerate fogs decreased rapidly with mean values of 52 cm−3, 4.1 µm, and 0.01 g m−3, re- with increasing duration. As the correlations of agglomerate spectively. LWC was less than previous observed results of fog duration with N, Dave, and LWC were low, we speculate mountain fog (Wu et al., 2007) and wire icing microphysics that the duration of agglomerate fog was controlled by the (Luo et al., 2008; Jia et al., 2010; Niu et al., 2012). This weather system. Figure 2b shows the frequency distribution might be due to the discontinuity of this fog process and to of time intervals between agglomerate fogs, with a frequency scouring of raindrops and snow particles on fog droplets, of 88.0% for a period of 0 to 10 min, a maximum time inter- among others. According to the criterion of N larger than 10 val of 436.2 min, and an average time interval of 10.3 min. cm−3 and LWC larger than 0.001 g m−3 (Lu et al., 2013), 218 The fog from 1740–1911 LST 22 January (90.6 min) had the agglomerate fogs were identified in this fog process. Figure longest duration; its average N, Dave, LWC, and peak diame- 2a shows the frequency distribution of the durations of ag- ter were 131 cm−3, 5.2 µm, 0.02 g m−3, and 4.7 µm, respec- JANUARY 2019 WANG ET AL. 33

Fig. 2. (a) Frequency distribution of fog durations, (b) frequency distribution of fog time intervals, and (c) fitting of K-M distribution of fog during 1740–1911 LST 22 January. tively. Figure 2c shows that the average droplet spectrum of crystals. The decrease in number concentration of large fog the longest fog was a bimodal distribution with two peak di- droplets in 27–50 µm particles was not significant; this was ameters at 4.9 µm and 8.9 µm, which satisfied the K-M distri- because the ice density was smaller—the ice floated when it bution with a goodness of fit of R2 = 0.92. These agglomerate was close to the large fog droplets and moved with the flow fog durations, frequency, time intervals, and other statistical field around water droplets. As a result, collision efficiency results provide the basic data for developing landscape mete- was reduced (Yang et al., 2011). When the precipitation rate orology. was 2–3 mm h−1, the spectral pattern of fog droplets was sig- Rain and snow appeared at the same time with fog and nificantly changed. impacted the fog droplet spectrum to some degree. How- ever, current research on the microphysical characteristics of rain fog and snow fog is limited. Figure 3 shows fog droplet 4. Analysis of icing accumulation under three spectra under different precipitation rates. At the time of rain- types of precipitation fall (Fig. 3a), the number concentration increased with an increase in precipitation rate for 2–15 µm and 30–50 µm parti- Under the influence of rain, snow, and fog, icing appeared cles. When rain was relatively heavy, evaporation increased on ice frames at both 10 and 1.5 m. Because of significant relative humidity, which was beneficial to the growth of fog icing growth at the 10-m height, the influence of the three droplets. At the same time, the collision and break of rain- kinds of precipitation on icing growth was studied using 10- drops also caused the number concentration to rise. When m height ice thickness data. the precipitation rate was 3–4 mm h−1, the spectral pattern Figure 4 shows the temporal evolution of elements such of fog droplets significantly changed. Figure 3b shows that as ice thickness at the 10-m height. According to the varia- the number concentration decreased with an increase in pre- tion in ice thickness, an icing process can be divided into the cipitation rate for 2–27 µm particles during snowfall. When icing preparation period, growth period, maintenance period, ice crystals and fog droplets coexisted, moisture spread from and fall-off period. Table 2 shows the ice thickness and me- water droplets to ice crystals (ice crystal effect) (Yang et al., teorological elements for the four periods at the 10-m height. 2011; Zhou et al., 2016), which was not conducive to the Figure 4b shows that temperature at the 10-m height, T10, growth of fog droplets. At the same time, the number concen- was slightly higher than zero (average 0.5◦C) during the ic- tration of fog droplets decreased due to the coagulation of ice ing preparation period. With the occurrence of precipitation,

Table 2. Average ice thickness, temperature, relative humidity, and wind velocity in each period of wire icing at the 10-m height.

Ice thickness T10 Relative humidity Wind velocity (mm) (◦C) (%) (m s−1) Preparation period 0000–1659 LST 20 January (17.0 h) — 0.5 71.8 0.7 Growth period 1700 LST 20 to 1041 LST 23 January (65.7 h) 7.9 −6.3 95.5 3.1 Maintenance period 1042–2359 LST 23 January (13.3 h) 18.0 −9.3 82.3 2.3 Fall-oﬀ period 0000 LST 24 to 0000 LST 25 January (24.0 h) 13.2 −12.9 71.1 3.3 34 SUPERCOOLED FOG AND ICING IN LUSHAN VOLUME 36

Fig. 3. Average fog droplet spectra under diﬀerent precipitation rates: (a) rain; (b) snow. °C

Fig. 4. Temporal variations (LST) of (a) ice thickness (black) and icing growth rate (red), (b) Tempera- ture T10 (black) and relative humidity (red), (c) wind velocity, (d) solar radiation, (e) precipitation rate, (f) fog droplet number concentration N, (g) fog liquid water content (LWC), and (h) fog droplet mean diameter Dave, in the period of wire icing at the 10-m height. JANUARY 2019 WANG ET AL. 35 relative humidity increased from 20% to above 90%, and the precipitation type was freezing rain, and the relative humid- average wind speed was 0.7 m s−1. Icing began to appear at ity (97.5%) was the highest for the three phases. In addition, 1700 LST 20 January. The T10 dropped to below zero and supercooled fog appeared. The fog droplet spectrum had a was −1.8◦C at the beginning of icing formation. The average single peak distribution in the rain phase (1700–2200 LST 20 temperature in the growth period was −6.3◦C, while relative January; Fig. 5a), with a peak at 2.8 µm. This phase was humidity was steady in the 81.7%–98.7% range. The wind sustained for 5 h, during which time fog appeared (visibility speed was significantly greater than that in the icing prepa- less than 1 km) for 3.5 h, accounting for 70.0% of the total ration period, with a maximum value of 9.1 m s−1, while ice length of time; this was the largest ratio for the three phases. thickness increased to 17.9 mm. According to the observa- Compared with fog droplets, the collection efficiency of wire tion, snow stopped and fog dissipated at about 1100 LST 23 for raindrops was higher (Lamraoui et al., 2014), so the ic- January, as the maintenance period began. Solar radiation in- ing growth rate (1.3 mm h−1) was the largest in this phase, creased from 7.0 W m−2 at 0806 LST 23 January to 695.0 W compared to the latter two phases without freezing rain. m−2 at 1217 LST 23 January, which prompted the increase The precipitation type in snow phase 1 was mainly snow. ◦ ◦ in T10 from −10.6 C to −5.9 C, followed by a gradual de- The temperature described in previous studies of the wet crease to the lowest value of −15.5◦C (0756 LST 24 January). snow icing process was roughly 0◦C(Makkonen, 1989; Changing with T10 in sync, relative humidity dropped from Makkonen and Wichura, 2010), but the average temperature 98.0% to 67.3% and then rose to 89.7%. Compared to the in this phase was −5.7◦C on the 10-m tower. Moreover, the wind speed in the growth period, the wind speed in the main- stratification analysis in section 3 shows that the entire atmo- tenance period decreased somewhat, with an average speed sphere above the observation site was below 0◦C and without of 2.3 m s−1. Ice thickness was steady and fluctuated slightly, a melting layer during snowfall. Therefore, the snowfall in with an average of 18.0 mm. In the fall-off period, solar radi- this phase was dry snow. The precipitation rate (0.4 mm h−1) ation peaked at 615.0 W m−2 (1157 LST) on 24 January; in was the highest for the three phases, but the surface of the the afternoon, relative humidity reached a minimum of 45.7% snow particles did not melt, so the sticking efficiency (ratio of ◦ (1422 LST), and T10 had a peak of −10.1 C (1433 LST). This the flux density of the particles that stick to the object to the increased the wire surface temperature, causing icing to di- flux density of the particles that hit the object) of wire was minish. The average wind speed in this period was 3.3 m s−1, lower (Makkonen, 2000). The wind speed (2.5 m s−1) and oc- which was larger than that in the maintenance period, accel- currence frequency of fog (16.9%) in snow phase 1 were also erating the icing fall-off (Zhou et al., 2012). the lowest for the three phases. Figures 4f, g, and h show that −3 −3 The growth period of icing on the tower was sustained for N, LWC, and Dave were less than 1 cm , 0.01 g m , and 3 65.7 h. To eliminate the influence of factors such as measure- µm, respectively, at 0800–1200 LST 21 January and 1500– ment errors on the calculation of the icing growth rate, the 1700 LST 21 January, which resulted in the lowest average −3 −3 five-point moving average of ice thickness was used to calcu- LWC (0.006 g m ), N (31 cm ), and Dave (3.4 µm) for the late the hourly icing growth rate, as shown in Fig. 4a. Accord- three phases, failing to provide rich moisture for the growth ing to the precipitation type and the growth of icing, the icing of icing. Figure 5 shows the evolution of the fog droplet growth period was divided into the rain phase, snow phase spectrum in snow phase 1. Compared with the rain phase, the 1, and snow phase 2 (Fig. 4a). Average ice thickness, growth number concentration increased in every spectral range dur- rate of ice, T10, relative humidity, wind velocity, precipita- ing 2200 LST 20 to 0500 LST 21 January (Fig. 5b). However, tion rate, and the microphysics of fog droplets in each phase the average icing growth rate decreased to 0.2 mm h−1 be- were calculated (Table 3). The fog droplet spectral evolution cause freezing rain switched to dry snow. During 0500–1400 was also analyzed (see Fig. 5). The three phases were com- LST 21 January (Fig. 5c), the number concentration was the pared in terms of meteorological conditions and fog droplet lowest for all stages in every spectral range except for 33.5– microphysical characteristics. 50.0 µm, and the average icing growth rate decreased to ◦ −1 The average T10 during the rain phase was −4.3 C, the −0.2 mm h . During 1400 LST 21 to 0800 LST 22 January

Table 3. Average ice thickness, growth rate of icing, temperature, relative humidity, wind velocity, precipitation rate, fog duration, and microphysics of fog droplets in each phase of growth period of wire icing at the 10-m height.

Ice Icing Relative Wind Fog thickness growth rate T10 humidity velocity Precipitation duration N Dave LWC (mm) (mm h−1) (◦C) (%) (m s−1) rate (mm h−1) (h) (cm−3) (µm) (g m−3) Rain 1700–2200 LST 20 Jan- 3.6 1.3 −4.3 97.5 2.6 0.3 3.5 43 4.2 0.006 uary (5.0 h) Snow 1 2200 LST 20 to 1200 6.3 0.1 −5.7 95.3 2.5 0.4 6.1 31 3.4 0.006 LST 22 January (38.0 h) Snow 2 1200 LST 22 to 1041 11.8 0.5 −7.8 95.5 4.2 0.2 10.4 83 5.1 0.012 LST 23 January (22.7 h)

36 SUPERCOOLED FOG AND ICING IN LUSHAN VOLUME 36

0 2

1 the dry snow phase. The intensive occurrence of supercooled

a fog, lower temperatures, and increased wind speed acceler-

3 )

1 0 1

m ated icing growth in the dry snow process.

(

1 0 0

i e

t a

r 5. Analysis of icing accumulation at two diﬀer-

1 0 1

n ent heights

1 0 2 o

c 5.1. Icing at the 1.5-m height

b e

0 3

1 Figure 6 shows the evolution of ice thickness, tempera- m

ture (T1.5), relative humidity, wind speed, and solar radiation N u

0 4 1 at the 1.5-m height. Icing appeared from 2300 LST 20 Jan-

uary until it fell oﬀ the wire at 1200 LST 23 January. Average

5 1 0 1 5 2 0 2 5 3 0 3 5 4 0 4 5 5 0

0 ice thickness on the east–west and north–south wires were 0.5

D i a m e t e r ( m ) and 0.9 mm, respectively, which differed by 44.4%. To study the effect of wind direction on icing growth, the wind direc- Fig. 5. Average fog spectra in each stage of the icing process: tion during icing was statistically analyzed. Wind direction (a) 1700–2200 LST 20 January; (b) 2200 LST 20 to 0500 LST was mainly northeasterly during the icing process. The fre- 21 January; (c) 0500–1400 LST 21 January; (d) 1400 LST 21 quency of the wind direction (with respect to the north–south to 0800 LST 22 January; (e) 0800–1200 LST 22 January; and direction) at greater than 45◦ was 52.9%, and that at less than (f) 1200 LST 22 to 1041 LST 23 January. 45◦ was 47.1%. When the wind direction and the wire were perpendicular, the yield was largest (Luo et al., 2008). Be- (Fig. 5d), the number concentration increased again in ev- cause the frequency of wind direction (with respect to the ery spectral range, and the average icing growth rate rose to north–south direction) at greater than 45◦ was higher, the ice 0.1 mm h−1. During 0800–1200 LST 22 January (Fig. 5e), thickness on the north–south wire was thicker than that on the the number concentration decreased for particles smaller than east–west wire. 27.5 µm and increased for particles larger than 27.5 µm; the Ice thickness at the 1.5-m height showed significant di- average icing growth rate was still 0.1 mm h−1 in this stage. urnal variation. Ice thickness increased at night (2000–0800 The average icing growth rate was 0.1 mm h−1 in snow phase LST) and decreased during the day (0800–2000 LST). Ice 1, icing grew slowly, and the ice thickness remained around thickness showed minima (0.4 and 0.3 mm on the east–west 6.3 mm. wire; 0.7 and 0.6 mm on the north–south wire) around 1600 The precipitation type in snow phase 2 was also dry snow. LST 21 January and 1400 LST 22 January. Variation in Its precipitation rate (0.2 mm h−1) was lower than that in ice thickness is generally sensitive to variations in temper- snow phase 1, but its icing growth rate (0.5 mm h−1) was ature (Zhou et al., 2012) and relative humidity during the day higher than that in snow phase 1. In this phase, icing grew fast when ice thickness is relatively thinner. At 1203 LST 21 Jan- again. Because the adhesion effect of dry snow particles was uary and 1221 LST 22 January, solar radiation had signifi- −2 weak, a higher precipitation rate in snow phase 1 could not cant peaks of 114.0 and 346.0 W m , respectively; T1.5 had effectively accelerate icing growth. The wind speed rapidly peaks of −2.3◦C and −2.5◦C, respectively; relative humidity increased to 9.1 m s−1 after 1200 LST 22 January in snow minima were 83.6% and 81.7%, respectively; while the ice phase 2 (Fig. 4c). The average wind speed in snow phase 2 thickness reached its minima at these times. was the highest (4.2 m s−1) for the three phases, which raised the collision efficiency (the ratio of the flux density of the 5.2. Comparisons of icing growth and meteorological ele- particles that hit the object to the maximum flux density) of ments at two different heights fog droplets and snow particles to the wire (Makkonen, 2000; Figure 7 and Table 4 compare the ice thickness and me- ◦ Davis et al., 2014). The T10 (−7.8 C) in snow phase 2 was the teorological elements at the two heights. The differences in lowest for the three phases, which was favorable for the freez- icing at the two heights were mainly manifested in four as- ing of fog droplets (Lamraoui et al., 2014). The occurrence pects. First, the ice thickness revealed a wide difference, with frequency of fog in snow phase 2 (45.8%) was higher than maximum ice thickness at 10 and 1.5 m of 20.7 and 1.2 mm, that in snow phase 1; the LWC (0.012 g m−3), N (83 cm−3), respectively; the latter was 5.8% of the former. Second, the and Dave (5.1 µm) in snow phase 2 were all the highest for the durations of icing were quite different. The durations were three phases. Figure 5f also shows that the number concen- 102 and 61 h, respectively; icing at 1.5 m only appeared in tration of the fog droplet spectrum increased in all spectral the growth period of 10-m icing. Third, the icing density re- ranges in snow phase 2 (1200 LST 22 to 1041 LST 23 Jan- vealed wide differences; there was translucent solid ice with uary) compared with the previous stage (0800–1200 LST 22 larger density at 1.5 m and white sponge-like ice with lower January), which provided a large number of supercooled liq- density at 10 m. Fourth, there was a difference in the ic- uid droplets for icing growth. ing growth mechanism. The accumulation rate of icing at 10 Icing grew rapidly in the freezing rain phase but slowly in m was closely related to factors such as the precipitation rate JANUARY 2019 WANG ET AL. 37 °C

Fig. 6. Temporal variations (LST) of (a) ice thicknesses (black) and diﬀerence of ice thicknesses between two directions ∆W1 (red), (b) temperature T1.5, (c) relative humidity, (d) wind velocity, and (e) solar radiation in the period of wire icing at the 1.5-m height.

Table 4. Average ice thickness, temperature, and wind velocity in each period of wire icing at the two heights.

Ice thickness Temperature Wind velocity (mm) (◦C) (m s−1) Preparation period 0000–1659 LST 20 January (17.0 h) 10 m — 0.5 0.7 1.5 m — 1.0 0.7 Growth period 1700 LST 20 to 1041 LST 23 January (65.7 h) 10 m 7.9 −6.3 3.1 1.5 m E-W 0.5 −5.2 1.6 N-S 0.9 Maintenance period 1042–2359 LST 23 January (13.3 h) 10 m 18.0 −9.3 2.3 1.5 m — −9.1 1.6 Fall-off period 0000 LST 24 to 0000 LST 25 January (24.0 h) 10 m 13.2 −12.9 3.3 1.5 m — −12.6 2.1 and microphysical characteristics of supercooled fog. The ice ticular, the maximum temperature difference appeared during thickness at 1.5 m was sensitive to daily variations in temper- the icing growth period (Fig. 7e), and the average tempera- ature and relative humidity, indicating diurnal variation. tures differed by 1.1◦C. The lower temperature on the tower The reasons for these differences are analyzed next. On was favorable for freezing on a wire of fog droplets and rain- the one hand, the differences in ice thickness and durations drops (Lamraoui et al., 2014). On the other hand, average were due to differences in temperature. Figure 7c shows the wind speeds were 3.1 and 1.6 m s−1 on and under the tower, temporal variations of air and ground temperatures of each respectively, in the icing growth period. The wind speed on height from 0000 LST 20 to 0000 LST 25 January. In gen- the tower was much larger than that under the tower, espe- eral, the range of ground temperatures was less than that of air cially from 1200 LST 22 to 1000 LST 23 January (Fig. 7e), temperatures. The temperature decreased with height during during which icing on the tower accumulated rapidly to 17.9 icing. Caused by the height difference, the temperature on the mm. Wind speed was an important factor in icing growth tower of T10 was less than that under the tower of T1.5. In par- (Nygaard et al., 2013), which raised the collision efficiency 38 SUPERCOOLED FOG AND ICING IN LUSHAN VOLUME 36 °C °C

Fig. 7. Temporal variations (LST) of (a) ice thickness at 10-m height and difference of ice thicknesses between two heights ∆W2, (b) ice thickness at 1.5-m height, (c) temperature at 10-m height T10, temperature at 1.5-m height T1.5, grass temperature Ts, ground temperature Td, 5-cm ground temperature Td5, 10-cm ground temperature Td10, 15-cm ground temperature Td15, and 20-cm ground temperature Td20, (d) wind velocities at two different heights, and (e) wind velocity difference at two different heights ∆V and temperature difference between the two heights ∆T in the period of wire icing. of fog droplets and snow particles on the wire (Davis et al., serve meteorological factors, such as precipitation type, pre- 2014). The difference in density was caused by the difference cipitation rate, and visibility, continuously. Through this ob- in temperature; ice density was higher when the temperature servation, we were able to obtain valuable data. was higher, and it was lower when the temperature was lower A cold surge was the background of this icing process. (Luo et al., 2008). The difference in the icing growth mech- The entire atmospheric layer was below 0◦C during the icing anism was also due to differences in temperature and wind process and there was no warm layer. Continuous reduction speed. The environment of low temperature and strong wind in temperature during this process turned rain into snow, and caused fog droplets and snow particles to rapidly freeze on fog appeared intermittently. Freezing rain, snow, and super- the wire. However, it was difficult for fog droplets and snow cooled fog provided rich moisture sources for icing growth. particles to freeze on the wire under the tower, so the ice There were 218 agglomerate fogs, of which durations of thickness at 1.5 m was thinner and more vulnerable to am- less than and greater than 10 min accounted for 91.3% and bient temperature and humidity. Environmental conditions 8.7%, respectively, and the average time interval was 10.3 were different at different heights during the same icing pro- min. The duration of agglomerate fog, time intervals, and cess, resulting in large differences in factors such as thick- other statistical features provided basic data for developing ness, duration, density, and growth mechanism. landscape meteorology. At the time of rainfall, the number concentration increased with an increase in precipitation rate 6. Conclusions for 2–15 µm and 30–50 µm particles, and the number concentration decreased with an increase in the precipitation rate A hotplate total precipitation sensor and weather identi- for 2–27 µm particles during snowfall. fier and visibility sensor, which are rarely used for domestic The icing observation at 10 m on the tower was more like meteorological observations, were used in this study to ob- the situation of actual high-voltage transmission lines. The JANUARY 2019 WANG ET AL. 39 icing growth period at 10 m was divided into three phases: of Applied Meteorology and Climatology, 53(2), 262–281, rain phase, snow phase 1, and snow phase 2. The main pre- https://doi.org/10.1175/JAMC-D-13-09.1. cipitation type in the rain phase was freezing rain, and the ic- Ding, Y. H., Z. Y. Wang, Y. F. Song, and J. Zhang, 2008: Causes ing growth rate (1.3 mm h−1) in the rain phase was the largest of the unprecedented freezing disaster in January 2008 and for the three phases. The precipitation type in snow phase 1 its possible association with the global warming. Acta Me- was dry snow, when icing grew slowly with a growth rate teorologica Sinica, 66(5), 808–825, https://doi.org/10.3321/ j.issn:0577-6619.2008.05.014. (in Chinese with English ab- of 0.1 mm h−1. The precipitation type in snow phase 2 was stract) also dry snow, with intensive occurrence of supercooled fogs, Drage, M. A., and G. Hauge, 2008: Atmospheric icing in a coastal ◦ lower temperatures (−7.8 C on average), and increased wind mountainous terrain. Measurements and numerical simula- −1 speed (4.2 m s on average), which accelerated the icing tions, a case study. Cold Regions Science and Technology, growth rate (0.5 mm h−1) in this dry snow process. 53(2), 150–161, https://doi.org/10.1016/j.coldregions.2007. The maximum ice thickness at 10 and 1.5 m was 20.7 and 12.003. 1.2 mm, respectively, while the duration was 102 and 61 h, Gultepe, I., G. A. Isaac, R. M. Rasmussen, and K. Ungar, 2011: A respectively. The density of icing at 10 m was lower than freezing fog/drizzle event during the FRAM-S project. SAE that at 1.5 m, and the accumulation rate of icing on the tower Technical Papers 2011-38-0028, https://doi.org/10.4271/ was closely related to the precipitation rate and microphys- 2011-38-0028. ical characteristics of supercooled fogs. The ice thickness Gultepe, I., and Coauthors, 2016: An overview of the MATER- HORN fog project: Observations and predictability. Pure at 1.5 m was sensitive to daily variations in temperature and Appl. Geophys., 173(9), 2983–3010, https://doi.org/10.1007/ relative humidity, indicating diurnal variation. In the icing s00024-016-1374-0. growth period, the mean temperatures at the 10-m and 1.5-m Hodges, D., and Z. X. Pu, 2016: The climatology, frequency, and ◦ ◦ heights were −6.3 C and −5.2 C, respectively, and the mean distribution of cold season fog events in northern Utah. Pure wind speeds were 3.1 and 1.6 m s−1, respectively. The differ- Appl. Geophys., 173, 3197–3211, https://doi.org/10.1007/ ences in temperature and wind speed were the main reasons s00024-015-1187-6. for the differences in thickness, duration, density, and growth Hosek, J., P. Musilek, E. Lozowski, and P. Pytlak, 2011: Forecast- mechanism of icing at the two heights, which indicated that ing severe ice storms using numerical weather prediction: The the icing thickness, duration, density, and growth mechanism March 2010 Newfoundland event. Natural Hazards and Earth were all closely related to height. We suggest that icing gra- System Science, 11(2), 587–595, https://doi.org/10.5194/ dient observations be conducted, the height of icing observa- nhess-11-587-2011. Jia, R., S. J. Niu, and Li, R., 2010: Observational study on mi- tion frames should be raised, more comparative observation crophysical characteristics of wire icing in west Hubei. Sci- studies of the relationship between different heights and ic- entia Meteorologica Sinica, 30(4), 481–486, https://doi.org/ ing growth should be conducted to make observation results 10.3969/j.issn.1009-0827.2010.04.008. (in Chinese with En- closer to the icing on actual high-voltage transmission lines, glish abstract) and new and more realistic icing observation methods should Jiang, Z. F., 1984: The growth rate of ice accretion on power line. be established. Chinese Science Bulletin, 29(4), 501–504. This work mainly analyzed observed data and discussed Jones, K. F., 1998: A simple model for freezing rain ice loads. At- the ice accumulation process with as much detail as possible. mospheric Research, 46(1–2), 87–97, https://doi.org/10.1016/ Further observational results show that the sticking efficiency S0169-8095(97)00053-7. of snow particles has a significant impact on icing growth Lamraoui, F., G. Fortin, R. Benoit, J. Perron, and C. Masson, 2014: rate, which will be considered in the next wire-icing numeri- Atmospheric icing impact on wind turbine production. Cold Regions Science and Technology, 100, 36–49, https://doi.org/ cal simulation. We will be considering collaborations in sim- 10.1016/j.coldregions.2013.12.008. ulation research, so experts who are interested in the observed Li, D. W., Q. Qiao, and T. Wei, 2009: Analyses on the freezing data of this icing process may contact the corresponding au- rain and snow weather circulation and vertical structure of thor to discuss further research. southern China in early 2008. Plateau Meteorology, 28(5), 1140–1148. (in Chinese with English abstract) Acknowledgements. This research was supported by the Na- Li, J., X. L. Guo, X. N. Zhou, and X. Y. Li, 2015: Characteristics tional Natural Science Foundation of China (Grant Nos. 41775134, of freezing rain, freezing drizzle, and freezing fog in China 41375138, 41505121, 41675132 and 41675136) and Graduate from 2011 to 2013. Chinese Journal of Atmospheric Sci- Student Innovation Plan for the Universities of Jiangsu Province ences, 39(5), 1038–1048, https://doi.org/10.3878/j.issn.1006- (KYCX18 1010). 9895.1501.14239. (in Chinese with English abstract) Liu, X. J., and S. J. Niu, 2016: Analysis of the causes for two high-voltage wire icing cases. Journal of the Meteorological REFERENCES Sciences, 36(2), 230–235, https://doi.org/10.3969/2014jms. 0101. (in Chinese with English abstract) China Meteorological Administration, 1979: Specifications for Lu, C. S., S. J. Niu, Y. G. Liu, and A. M. Vogelmann, 2013: Empir- Surface Meteorological Observation. Beijing: Meteorologi- ical relationship between entrainment rate and microphysics cal Press, 186 pp. (in Chinese) in cumulus clouds. Geophys. Res. Lett., 40(10), 2333–2338, Davis, N., A. N. Hahmann, N.-E. Clausen, and M. Zagar,ˇ 2014: https://doi.org/10.1002/grl.50445. Forecast of icing events at a wind farm in Sweden. Journal Luo, N., J. F. Wen, C. Zhao, and L. Tang, 2008: Observation 40 SUPERCOOLED FOG AND ICING IN LUSHAN VOLUME 36

study on properties of cloud and fog in ice accretion ar- the southern China in January 2008. Acta Meteorologica eas. Journal of Applied Meteorological Science, 19(1), 91– Sinica, 66(3), 405–422, https://doi.org/10.3321/j.issn:0577- 95, https://doi.org/10.3969/j.issn.1001-7313.2008.01.012. (in 6619.2008.03.011. (in Chinese with English abstract) Chinese with English abstract) Wang, Y. F., Y. Li, P. Y. Li, and Y. Liu, 2008b: The large scale cir- Makkonen, L., 1989: Estimation of wet snow accretion on struc- culation of the snow disaster in South China in the beginning tures. Cold Regions Science and Technology, 17, 83–88, of 2008. Acta Meteorologica Sinica, 66(5), 826–835, https:// https://doi.org/10.1016/S0165-232X(89)80018-7. doi.org/10.3321/j.issn:0577-6619.2008.05.015. (in Chinese Makkonen, L., 2000: Models for the growth of rime, glaze, with English abstract) icicles and wet snow on structures. Philosophical Transac- Wang, Z. Y., 2011: Climatic characters and changes of ice- tions: Mathematical, Physical and Engineering Sciences, freezing days in China. Chinese Journal of Atmospheric Sci- 358, 2913–2939, https://doi.org/10.1098/rsta.2000.0690. ences, 35(3), 411–421, https://doi.org/10.3878/j.issn.1006- Makkonen, L., and B. Wichura, 2010: Simulating wet snow loads 9895.2011.03.03. (in Chinese with English abstract) on power line cables by a simple model. Cold Regions Sci- Wu, D., X. J. Deng, J. T. Mao, W. K. Mao, Y. X. Ye, X. Y. ence and Technology, 61, 73–81, https://doi.org/10.1016/ Bi, H. H. Tang, and Q. L. Wang, 2007: A study on macro- j.coldregions.2010.01.008. and micro-structures of heavy fog and visibility at freeway Musilek, P., D. Arnold, and E. P. Lozowski, 2009: An ice accre- in the Nanling Dayaoshan mountain. Acta Meteorologica tion forecasting system (IAFS) for power transmission lines Sinica, 65(3), 406–415, https://doi.org/10.3321/j.issn:0577- using numerical weather prediction. Scientific Online Letters 6619.2007.03.009. (in Chinese with English abstract) on the Atmosphere Sola, 5, 25–28, https://doi.org/10.2151/ Yang, G. M., Q. Kong, D. Y. Mao, F. H. Zhang, Z. M. sola.2009-007. Kang, and Z. P. Zong, 2008: Analysis of the long-lasting Niu, S. J., Y. Zhou, R. Jia, J. Yang, J. J. Lu,¨ Y. M. Ke, and Z. cryogenic freezing rain andsnow weather in the begin- B. Yang, 2012: The microphysics of ice accretion on wires: ning of 2008. Acta Meteorologica Sinica, 66(5), 836–849, Observations and simulations. Science China Earth Sciences, https://doi.org/10.11676/qxxb2008.076. (in Chinese) 55(3), 428–437, https://doi.org/10.1007/s11430-011-4325-8. Yang, J., B. J. Chen, and Y. Yin, 2011: Physics of Clouds and Nygaard, B. E. K., H. Ag´ ustsson,´ and K. Somfalvi-Toth,´ 2013: Precipitation. Meteorological Press, 364 pp. (in Chinese) Modeling wet snow accretion on power lines: Improvements Zeng, M. J., W. S. Lu, X. Z. Liang, H. Y. Wu, M. J. Pu, and D. to previous methods using 50 years of observations. Jour- P. Yin, 2008: Analysis of temperature structure for persis- nal of Applied Meteorology and Climatology, 52, 2189–2203, tent disasterous freezing rain and snow over southern China https://doi.org/10.1175/JAMC-D-12-0332.1. in early 2008. Acta Meteorologica Sinica, 66(6), 1043–1052, Pytlak, P., P. Musilek, E. Lozowski, and D. Arnold, 2010: Evo- https://doi.org/10.11676/qxxb2008.093. (in Chinese with En- lutionary optimization of an ice accretion forecasting system. glish abstract) Mon. Wea. Rev., 138(7), 2913–2929, https://doi.org/10.1175/ Zhou, Y., S. J. Niu, and J. J. Lu,¨ 2013: The influence of freezing 2010MWR3130.1. drizzle on wire icing during freezing fog events. Adv. Atmos. Tan, G. R., 1982: On the microclimatological characteristics Sci., 30(4), 1053–1069, https://doi.org/10.1007/s00376-012- of icing on wires. Acta Meteorologica Sinica, 40, 13–23, 2030-y. https://doi.org/10.11676/qxxb1982.002. (in Chinese with En- Zhou, Y., S .J. Niu, J. J. Lu,¨ and L. J. Zhao, 2012: Meteorologi- glish abstract) cal conditions of ice accretion based on real-time observation Tao, Y., Y. Q. Shi, and W. G. Liu, 2012: Characteristics of strati- of high voltage transmission line. Chinese Science Bulletin, fication structure and cloud physics of the freezing rain over 57(7), 812–818, https://doi.org/10.1007/s11434-011-4868-2. southern China in January 2008. Chinese Journal of Atmo- Zhou, Y., S. J. Niu, J. J. Lu,¨ and Y. H. Zhou, 2016: The effect of spheric Sciences, 36(3), 507–522, https://doi.org/10.3878/ freezing drizzle, sleet and snow on microphysical characteris- j.issn.1006-9895.2011.11082. (in Chinese with English ab- tics of supercooled fog during the icing process in a mountain- stract) ous area. Atmosphere, 7, 143, https://doi.org/10.3390/atmos Wang, D. H., and Coauthors, 2008a: A preliminary analy- 7110143. sis of features and causes of the snow storm event over