development of a lightning model and implementation into a
TRANSCRIPT
Development of a lightning model and implementation into a
meteorological model
developed in Japan ~ Validation through the comparison with the ground
base measurement ~
Akihiro Hashimoto (Meteorological Research Institute, Japan)
EGU21-6984, https://doi.org/10.5194/egusphere-egu21-6984
What we have done in this study?
1. Development of a lightning model and implement into Community Model of Japan (SCALE: https://scale.riken.jp/)
2. Validation of the lightning model through the comparison with ground base measurement
3. Examine the reason of the difference of lightning frequency between two heavy rain events in Japan
Introduction
Relationship between lightning frequency and heavy rain (Hamada et al. 2015, Hamada and Takayabu 2018)
The intensity of instantaneous rainfall rate has convention- ally been linked to the intensity of convective storms. There is an extensive literature on the characteristics of
severe storms on both global and regional scales, in which these storms are usually characterized as having extremely strong convection. On the basis of spaceborne radar and radiometer measurements, the global distribution of extreme convective events has been relatively well described1–4. There are also many detailed regional studies on the rainfall characteristics of extreme convective events5–7.
On the other hand, there are observational facts that certain types of extreme rainfall events, such as flash flooding storms, do not necessarily accompany extremely strong convection and intense lightning activity, even in regions where severe convective storms are representative extreme weather events8–12. In some regions, heavy surface rainfall is typically associated with
relatively low-echo-top heights13–15. Moreover, heavy oro- graphic rainfall can inherently exhibit lower echo-top heights16. There are also observations which show that convective intensity is not necessarily related to the near-surface rainfall intensity in some specific regions17–19.
Extreme rainfall over short durations could have devastating socioeconomic effects, as one of the key factors of flash flooding. Therefore, such a decorrelation between the top heights of strong echoes and heavy/extreme surface rainfall rates highlights the importance of a detailed study that focuses on the inherent characteristics of extreme rainfall events and contrasts those with extreme convective events.
Despite many detailed regional studies, a global picture of the linkage between extreme convection and extreme surface rainfall rate remains unclear. Here we use a unique long-term record from a spaceborne precipitation radar (PR) to investigate the
GLB, Land, ANN, v7f, p999
R_ONLY
3,000
2,000
1,000
500
200
100
50
10
20
15
10
5
R_ONLY
3,000
2,000
1,000
500
200
100
50
10
20
15
10
5
Reflectivity (dBZ)
H_ONLY RH
H_ONLY RH
Figure 1 | Echo profiles for three types of extreme events. Composite structures of radar reflectivity at extreme pixels within the TRMM observation domain (37.5!S–37.5!N). Colours show joint histograms of effective radar reflectivity and height, superimposed by solid and dashed lines that indicate the mean and s.d. for each height bin, respectively. Solid and dashed lines along the right-hand axis of each panel show the histograms of echo-top heights and 0 !C levels, respectively. The number of samples for the corresponding extreme type is indicated. (a–c) R-only, H-only and RH-extreme events over land, respectively. (d–f) R-only, H-only and RH-extreme events over the ocean, respectively.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7213
2 NATURE COMMUNICATIONS | 6:6213 | DOI: 10.1038/ncomms7213 | www.nature.com/naturecommunications
& 2015 Macmillan Publishers Limited. All rights reserved.
(Hamada et al. 2015 Nat. Comm.)
Hamada et al. (2015) suggested that Heavy rain can be divided three groups based on satellite measurement, R-only, H-only, RH extreme. • RH-extreme: Heavy rain from high convective cloud • R-only extreme: Heavy rain from low convective cloud • H-only extreme: Weak rain from high convective cloud
Why?
Introduction: Relationship between lightning frequency and heavy rain (Hamada et al. 2015, Hamada and Takayabu 2018)
It should be mentioned that in this study, we only focus on the large-scale moisture transport in the free troposphere that is vertically integrated over a wide range of the troposphere. Further investigation is re- quired to specify the height range, which is key to the occurrence of the extreme rainfall events and extreme convective events, to make dynamical and physical interpretations.
Anomaly composite maps for the vertically integrated moisture and its horizontal fluxes are shown in Fig. 7. For the R-extreme events, a positive (i.e., moist) anomaly dominates over the extreme event region both in the free troposphere and the boundary layer (Figs. 7a,c). However, the spatial extents of the positive moisture anomaly region are substantially different between the two layers. In the free troposphere, the positive anomaly extends zonally from southern China to the east of Ja- pan. This feature indicates that the excessive moisture in the free troposphere of the composite region can be attributed partially to transportation from the west by a large-scale flow rather than a direct injection by the extreme precipitation event within the region. There is also a negative anomaly region extending zonally, paired with the positive anomaly. A positive anomaly is also observed in the boundary layer and is attributed to southwesterly moisture flux; however, the spatial extent is quite limited compared with that in the free troposphere. For the H-extreme events, in contrast, negative (i.e.,
dry) anomalies dominate over the composite region in both the free troposphere and boundary layer (Figs. 7b,d). Moisture fluxes are northwesterly in both layers. Posi- tive anomalies are observed instead on the western and eastern sides of the composite region. The negative anomalous moisture fluxes both in the boundary layer and free troposphere over the composite region might seem against the general understanding that moisture inflow at the boundary layer is essential to the initiation and formation of heavy rainfall events over Japan. Further detailed study is needed, but this is partly be- cause about 65% of the R- and H-extreme events occurred over the sea, where forced lifting of moisture inflow at the lowest level by the topography is not es- sential to the initiation of convection, or is simply because smaller-scale flows are not resolved in the re- analysis dataset used in this study. An interesting feature that contrasted with R-extreme events is seen in the free troposphere west of 1108E, where a significant positive moisture anomaly dominates north of 308N and a sig- nificant negative anomaly dominates south of 308N. The patterns of anomalous circulation and temperature in the middle–upper troposphere in this region associated with the H-extreme events (not shown) are quite similar to those associated with the ‘‘warm heavy rainfall days’’ that are determined by Sun et al. (2015, their Figs. 3a,c) over central north China. This possibly indicates a dynamical link between the H-extreme events over southern Japan and warm-type heavy rainfall events over central north China. The differences between the vertically integrated
moisture anomalies associated with R- and H-extreme
FIG. 5. Box-and-whisker plots of (a) size (km2), (b) stratiform- area ratio (%), (c) maximum 40-dBZ echo-top height (km), (d) maximum precipitation-top height (km), (e) flash rate (per convective pixel per minute), and (f) minimum85GHz PCT (K), for R-, H-, and RH-extreme events. The boxes show the interquartile ranges and medians, the whiskers indicate the ranges from 10th to 90th percentiles, and filled dots indicate the averages. Note that the statistics on flash rate are calculated from extreme events with LIS viewing time longer than 80 s (the number of events are 44, 35, and 6 for R-, H-, RH-extreme events, respectively).
1 SEPTEMBER 2018 HAMADA AND TAKAYABU 6939
D ow
(Hamada and Takayabu 2018, JC)
• In R-only extreme, lightning frequency was low even if heavy rainfall occurred.
characteristics of rainfall events that produce the greatest instantaneous rainfall rates in the tropics and subtropics. We further demonstrate statistically a weak linkage between the extreme rainfall events and extreme convective events. Furthermore, the differences in rainfall characteristics between extreme rainfall events and extreme convective events and their regionality are investigated.
Results Robust differences between two extremes. Figure 1 shows composite structures of effective radar reflectivity (Ze) at extreme
pixels where the maximum near-surface rainfall rate or maximum 40-dBZ echo-top height is observed for each type of extreme event for land and the ocean. There are clear and robust differ- ences in the vertical structures between R-only and H-only extreme events. Note that the rain type, that is, convective or stratiform, is not considered, because more than 95% of the extreme pixels were determined as convective for all extreme types. Clear differences are observed in echo profiles between R-only and H-only extreme events, both over land and over the oceans, since the differences are more significant over land than over the ocean. The most striking characteristic is the echo structure of R-only extreme events both over land and over the
R_ONLY
n=1,071 n=1,002 n=363 n=378 n=641 n=622
n=2,078 n=1,345 n=1,513 n=1,177 n=1,698 n=1,281 256
128
64
32
16
8
4
2
1
256
128
64
32
16
8
4
2
1
20
15
10
20 40 60 20 40 60 Reflectivity (dBZ)
EWPAC, Ocean, ANN, v7f, p999 JP, Ocean, ANN, v7f, p999 SWPAC, Ocean, ANN, v7f, p999
20 40 60 20 40 60
Reflectivity (dBZ) 20 40 60 20 40 60
Reflectivity (dBZ)
20 40 60 20 40 60
Reflectivity (dBZ)
AMZ, Land, ANN, v7f, p999 EWAF, Land, ANN, v7f, p999
H_ONLY
R_ONLY H_ONLY R_ONLY H_ONLY
R_ONLY H_ONLY R_ONLY H_ONLY R_ONLY H_ONLY
Figure 2 | Regional variation of extreme event characteristics. The same as in Fig. 1, but for several specific geographical regions. Only the panels for R-only and H-only extreme events are shown for each region. (a) Amazon (70!W–40!W, 15!S–0!, over land). (b) Equatorial western Africa (15!W–15!E, 0!–15!N, over land). (c) Southern North America (110!W–80!W, 22!N–37!N, over land). (d) Equatorial western Pacific (140!E–170!E, 0!–15!N, over ocean). (e) Northwestern Pacific around Japan (120!E–150!E, 22!N–37!N, over ocean). (f) Southwestern Pacific (180!–150!W, 37!S–22!S, over ocean).
RH extreme event fraction (RH/total) 200109–201208 40N
20N
EQ
20S
(%)
70
50
30
10
30W
Figure 3 | Weak linkage between extreme rainfall and convective events. Fraction of the number of RH-extreme events to that of total extreme events in each region.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7213 ARTICLE
NATURE COMMUNICATIONS | 6:6213 | DOI: 10.1038/ncomms7213 | www.nature.com/naturecommunications 3
& 2015 Macmillan Publishers Limited. All rights reserved.
(Hamada et al. 2015 Nat. Comm.)
R-only is main contributors to the heavy rain ↓
Convective clouds with frequent lightning does not always results in heavy rainfall
Lightning frequency used for rainfall estimation
• Lightning frequency have been used for the estimation of the rainfall from the measurement (e.g., Tapia et al. 1998, Xu et al. 2013)
• However, the report of Hamada et al. (2015) indicated that the convective clouds without frequent lightning (R-only extreme) are main contributor to the heavy rainfall.
• The deeper understanding about the difference between R- only extreme and other extreme (RH-extreme) is required
Target cases of this study “Heavy rain events in 2017 and 2018 in Japan”
Heavy rain events on July, 2017Case 2017 • Heavy rainfall triggered by Baiu front • Cumulative precipitation exceeded 800 mm/48 hours • The lightning frequency was high
Heavy rain events on July, 2018 Case 2018 • Heavy rainfall triggered by Baiu front • Cumulative precipitation exceeded 800 mm/48 hours • The lightning frequency was much smaller than that in case 2017.
(Produced by Japan Meteorological Agency)
Precipitation and measured lightning
Weather map (2017/7/5 00UTC)
(Produced by Japan Meteorological Agency)
Precipitation and measured lightning
Weather map (2018/7/6 00UTC)
51
831
Fig. 15. Sche a ic f ai fa e e f (a) he 2018 ca e a d (b) f he 2017 ca e. 832
833
(Tsuji et al. 2020 JMSJ)
Tsuji et al. (2020) reported that: The difference between Case 2017 and Case 2018 was corresponding to the difference between the R-only extreme and RH-extreme →The two cases are good example to understand the difference in the lightning frequency between R-only extreme and RH-extreme (Purpose of this study)
51
831
Fig. 15. Sche a ic f ai fa e e f (a) he 2018 ca e a d (b) f he 2017 ca e. 832
833
Previous studies for the two cases Case 2017 • Kato et al. (2018), Takemi (2018), Kawano and Kawamura (2020), Hashimoto and Hayashi (2020)
Case 2018 • Kotsuki et al. (2019), Moteki et al. (2019), Sekizawa et al. (2019), Shimpo et al. (2019), Hashimoto and Hayashi (2020)
Most of them did not discuss the difference in the lightning frequency between these two cases
Purposes of this study • To understand the reason of the difference in the lightning frequency between R-only extreme and RH- extreme
• To do so, we used the meteorological model coupled with a lightning model, which was developed by our groups.
Model and experimental setup
in enhancing the warm rain process [Takahashi, 1976; Takahashi and Lee, 1978]. Above the freezing level a strong updraft creates high supersaturation and small nuclei are activated. Some of them will act as new ice forming nuclei due to high supersaturation [Hussain and Saunders, 1984]. Small drops collide with small frozen drops and ice crystals carried from a lower level, and graupel are formed. Because the drops are small, low-density graupel grows, with brittle surface branches [Macklin, 1977]. Collision with a large frozen drop or large graupel leads to ejection of ice frag- ments [Takahashi et al., 1995b]. Additional release of latent heat by the nucleation of small nuclei also enhances the updraft.
6. Conclusions
[32] During the MCTEX project, 14 videosondes were launched into clouds over Tiwi Island near Darwin, Aus- tralia. Seven entered a ‘‘Hector’’ squall line, returning data related to the distribution of hydrometeor species and their electrical charges. Knowledge of their positions in relation to the storm allowed the data to be combined in the construction of a composite map of the mass, number, and space charge distributions. [33] The composite images thus obtained showed a two-
step hydrometeor growth process: one at low levels at the front in warm rain freezing and the other at upper levels within main convection by graupel and ice crystal growth. The space charge evolution has been successfully explained by riming electrification. [34] The space charge distribution exhibited a basic tri-
pole structure, but with a positive space charge extended to the anvil. The high ice crystal concentration in the ‘‘Hector’’ clouds was explained by a higher updraft and higher concentrations of small nuclei over the maritime continent than has been observed over the ocean. The small droplets may thus help in forming brittle graupel as ice fragment generators.
[35] The basic concept of growth modes and particle charge evolution obtained in a Hector squall line may be applied to other squall lines, although different microphys- ics in different areas may modify the distribution. It is highly desirable to conduct similar measurements in other areas.
[36] Acknowledgments. This research was conducted as a part of the MCTEX project. The authors would like to thank D. Jasper for his tremendous help during this expedition. Many students, K. Suzuki, T. Kawano, T. Tajiri, M. Sugiyama, and Y. Kushiyama, from Kyusyu University participated. Research was supported by the Japan Aerospace Exploration Agency, the Ministry of Education, Science, Sports, and Culture of Japan, and the Climate Center of Tokyo University.
References Braham, R. B., Jr. (1990), Snow particle size spectra in lake effect snows, J. Appl. Meteorol., 29, 200–207.
Carey, L. D., and S. A. Rutledge (2000), The relationship between precip- itation and lightning in tropical island convection: A C-band polarimetric radar study, Mon. Weather Rev., 128, 2687–2710.
Chauzy, S., M. Chong, A. Delannoy, and S. Despiau (1985), The June 22 tropical squall line observed during COPT 81 experiment: Electrical signature associated with dynamical structure and precipitation, J. Geo- phys. Res., 90, 6091–6098.
Houze, R. A., Jr., and P. V. Hobbs (1982), Organization and structure of precipitating cloud systems, Adv. Geophys., 24, 225–315.
Hussain, K., and C. P. R. Saunders (1984), Ice nucleus measurement with a continuous flow chamber, Q. J. R. Meteorol. Soc., 110, 75–84.
Jorgenson, D. P., and M. A. LeMone (1989), Vertical velocity character- istics of oceanic convection, J. Atmos. Sci., 46, 621–640.
Keenan, T., B. R. Morton, X. S. Zhang, and K. Nuguen (1990), Some characteristics of thunderstorms over Bathurst and Melville Island near Darwin, Australia, Q. J .R. Meteorol. Soc., 116, 1153–1172.
Keenan, T., et al. (2000), The Maritime Continent Thunderstorm Experi- ment (MCTEX): Overview and some results, Bull. Am. Meteorol. Soc., 81, 2433–2455.
Leary, C. A., and R. A. Houze Jr. (1979), Melting and evaporation of hydrometeors in precipitation from the anvil clouds of deep tropical convection, J. Atmos. Sci., 36, 669–679.
Lyons, W. A., T. E. Nelson, E. R. Williams, S. A. Cummer, and M. A. Stanley (2003), Characteristics of Sprite-producing positive cloud-to- ground lightning during the 19 July STEPS mesoscale convective systems, Mon. Weather Rev., 131, 2417–2427.
Macklin, W. C. (1977), The characteristics of natural hailstones and their interpretation, Meteorol. Monogr., 16(38), 65–88.
Marsh, S., and T. C. Marshall (1993), Charged precipitation measurements before the first lightning flash in a thunderstorm, J. Geophys. Res., 98, 16,605–16,611.
Marshall, T. C., and W. D. Rust (1993), Two types of vertical electrical structures in stratiform precipitation regions of mesoscale convective systems, Bull. Am. Meteorol. Soc., 74, 2159–2170.
May, P. T., A. R. Jameson, T. D. Keenan, P. E. Johnson, and C. Lucas (2002), Combined wind profiler/polarimetric radar studies of the vertical motion and microphysical characteristics of tropical sea-breeze thunder- storms, Mon. Weather Rev., 130, 2228–2239.
Miura, K., S. Nakae, T. Sekikawa, and T. Kumakura (1993), Global distribution of Aitken particles over the oceans, J. Atmos. Electr., 13, 133–144.
Podzimek, J. (1980), Advances in maritime aerosol research, J. Rech. Atmos., 14, 35–61.
Simpson, G. C., and F. J. Scrase (1937), The distribution of electricity in thunderclouds, Proc. R. Soc. London, Ser. A, 161, 309–352.
Simpson, J., T. D. Keenan, B. Ferrier, R. H. Simpson, and G. J. Holland (1993), Cumulus mergers in the maritime continent region, Meteorol. Atmos. Phys., 51, 73–99.
Takahashi, T. (1973), Measurement of electric charge of cloud droplets, drizzle, and raindrops, Rev. Geophys., 11, 903–924.
Takahashi, T. (1976), Warm rain, giant nuclei, and chemical balance—A numerical model, J. Atmos. Sci., 33, 269–286.
Takahashi, T. (1978a), Riming electrification as a charge generation mechanism in thunderstorms, J. Atmos. Sci., 35, 1536–1548.
Takahashi, T. (1978b), Electrical properties of oceanic tropical clouds at Ponape, Micronesia, Mon. Weather Rev., 106, 1598–1612.
Takahashi, T. (1984), Thunderstorm electrification—A numerical study, J. Atmos. Sci., 41, 2541–2558.
Takahashi,T. (1990),Near absenceof lightning in torrential rainfall producing Micronesian thunderstorms,Geophys. Res. Lett., 17, 2381–2384.
Figure 7. Conceptual model of hydrometeor growth and space charge. Red circles are for raindrops originating on frozen drops. Blue circles are for raindrops, originating on graupel. Green circles are frozen drops, blue triangles are graupel, and red crosses are ice crystals. A two-step growth process is suggested by warm rain-frozen in front and graupel growth in the upper level.
D16208 TAKAHASHI AND KEENAN: SQUALL PRECIPITATION AND ELECTRICITY
10 of 11
Lightning Model developed by this study (Sato et al. 2019)
The lightning component consists of 1. Charge separation via collision between graupel and ice/snow (Takahashi 1978) 2. Movement of charge density (tracer advection, sedimentation, diffusion and so on) 3. Calculation of electric field (Poisson equation) 4. Neutralization (lighting)Ferrio et al. 2013: Calculate Total lightning
Physical variables required for the lightning component • Charge density of hydrometeor (prognostic) • Electrical potential (diagnose) • Electric field (diagnose)
Takahashi and Keenan (2004)
Basic assumption Lightning in the atmosphere occurs to neutralize the charge density of “cloud hydrometeors”
Implement
Japanese community model
Model and Sensor simulator Model SCALE (Nishizawa et al. 2015, Sato et al. 2015) • Dynamics HE-VI • Microphysics 2-moment bulk (Seik and Nakajima 2014) • Radiation MSTRNXSekiguchi et al. 2008) • Turbulence MYNNNakanishi and Niino 2006) • Lightning Bulk lightning modelSato et al. 2019) • Surface flux Beljaars-type bulk modelBeljaars and Holtslag 1994 • Land model Bucket type model • Urban model Single layer urban canpy model (Kusaka et al. 2001) • Ocean Given from parent model
Sensor simulatorJoint Simulator (Hashino et al. 2013) • Sensor Radar (Masunaga and Kumerrow, 2005) • Wavelength 5.345 GHz CBand radar operated by JMA
Experimental setup Case Case 2017 Case 2018
Calculation period 00 UTC of 2017/7/5 ~ 00 UTC of 2017/7/6
21 UTC of 2018/7/6 ~ 00 UTC of 2018/7/8
Initial and Boundary condition MANL (Δx=5km 50 layer3hours interval) Horizontal grid spacing 1km Number of vertical layer 60 (Δz = 40 ~ 638.5 m) 57 (Δz = 40 ~ 638.5 m)
Calculation domain Case 2017 Case 2018
GreyCalculation domain RedDomain for analyses
(Sato et al. submitted to ASL)
Measurement data • Radar/Raingauge-analyzed precipitation productby Japan Meteorological Agency: JMA • Radar Reflectivity by C-band radar operated by JMA (CAPPI, 1km vertical/horizontal grid resolution)
• Ground base measurement of lighting named LIghtning DEtection Network (LIDEN) operated by JMA (Ishii et al. 2014)
Location of LIDEN
ResultsValidation
Ca se 2 01 7
(7 /5 0 0U TC ~ 7 /6 0 0 U TC )
Ca se 2 01 8
(0 7/ 06 2 1U TC ~ 7 /7 0 0 U TC )
Geographical distribution of the heavy precipitation was reasonably simulated for both case except for the area close to lateral boundary
(Sato et al. submitted to ASL) An artifact by lateral boundary
Vertical distribution (CFAD)
The model overestimated reflectivity, but the difference in the characteristics between two cases was well reproduced
CF AD (C as e 20 17 )
(7 /5 0 0U TC ~ 7 /6 0 0 U TC )
CF AD (C as e 20 18 )
(0 7/ 06 2 1U TC ~ 7 /7 0 0 U TC )
Ca se 2 01 7
(7 /5 0 0U TC ~ 7 /6 0 0 U TC )
Ca se 2 01 8
(0 7/ 06 2 1U TC ~ 7 /7 0 0 U TC )
Cumulative lightning frequency
Geographical distribution of the lightning was reasonably simulated for both case except for the area close to lateral boundary
An artifact by lateral boundary
Temporal evolution Domain accumulated lightning number
0
20
40
60
80
100
120
6 8 10 12 14 16 18 20 22 24
Fl as
h nu
m be
r/F O
Time from 2017 July 5 00 UTC (09 JST) [h]
LIDEN Model
Time from 2018 July 6 21 UTC (06 JST) [h]
Model Model(limit domain)
The model successfully simulated temporal evolution of lightning.
Red Measurement (LIDEN) Blue Model Green Modelwithout area close to boundary
Case 2017 Case 2018
Summary of the Validation
• The model successfully simulated heavy rain, lightning frequency, and contrast of the lightning frequency between Case 2017 and Case 2018.
Results Difference of the lightning frequency between Case 2017 and 2018
Reason of the tri-pole structure
~10km
z
~7km
Upward motion
charge separation, we considered non-inductive charge separation, which is assumed to occur with the rebound of snow/ice particles after the collision with graupel par- ticles, as in previous modeling studies (Takahashi 1978; Mansell et al. 2005). In addition to non-inductive charge separation, inductive charge separation (Ziegler et al. 1991) is often considered by other models. However, inductive charge separation was ignored in this study, because the contribution of inductive charge separation to the total charge density is much smaller than that of non-inductive charge separation (Mansell et al. 2005; Mansell et al. 2010). The charge density of graupel (ρe,g) and ice/snow (ρe,i,
ρe,s) obtained by the rebound after the collision in each time step was calculated based on the stochastic coales- cence equation (cf. Pruppacher and Klett 2010) in the three microphysical schemes as:
dρe;g dt
0 ng r1ð Þni;s r2ð Þπ r1 þ r2ð Þ
% Vg r1ð Þ−V i;s r1ð Þ !! !!Ecoll 1−Ecoalð Þdr1dr2; ð1Þ
dρe;i=s dt
¼ − dρe;g dt
; ð2Þ
where r1, r2, ng, ni,s, Vg, Vi,s, Ecoll, and Ecoal are the radius of graupel, radius of ice/snow, number concentration of graupel, number concentration of ice/snow, terminal vel- ocity of graupel, terminal velocity of ice/snow, collision efficiency, and coalescence efficiency, respectively. δρ’, defined as:
δρ’ ¼ αδρ; ð3Þ
is the charge density separated by one collision and re- bound, and α is given as
α ¼ 5:0 r2=r0ð Þ2Vg=V 0; ð4Þ
based on Takahashi (1984), where Vg is graupel terminal velocity, and V0 and r0 are 8 m s−1 and 50 μm, respect- ively. δρ is obtained from a look-up table (LUT) from Takahashi (1978), as shown in Fig. 1. β is 1, 1 − [(T + 30)/13]2, and 0 for temperature (T) warmer than − 30 °C, between − 43 and − 30 °C, and colder than − 43 °C, re- spectively, based on Mansell et al. (2005). To avoid an unrealistically large charge density, α is not permitted to be larger than 10, based on Takahashi (1984). The electric field (E) is calculated by solving the Pois-
son equation as:
ε ; ð5Þ
E ¼ −∇; ð6Þ where and ε are electrical potential and atmospheric permittivity, respectively, and ρe(x,y,z) is the charge
density of each grid, which is the sum of the charge density of each hydrometeor in each grid. The calcula- tion of Eq. (5) requires a large computational cost, as shown in Table 1, and a fast Poisson solver is necessary. In this study, the bi-conjugate gradient stabilized (Bi- CGSTAB) method (van der Vorst 1992), which was ori- ginally implemented in SCALE, was used to solve Eq. (5). To calculate the charge neutralization, two schemes
(MacGorman et al. 2001; Fierro et al. 2013) were imple- mented into SCALE. Both schemes neutralize the charge density when the electric field exceeds a threshold value (Eint). The initial point of electric discharge is deter- mined randomly among the grids in which |E| exceeds the Eint. By the scheme of MacGorman et al. (2001), the path of the lightning is explicitly calculated, and the discharge occurs along the path. By contrast, in the scheme of Fierro et al. (2013), the discharge occurs within vertical cylinders with a radius of rcylinder around the initial point. The details of both schemes are de- scribed in the literature.
Experimental setup The experimental setup of this study targeted on an ide- alized TC, and it followed almost the same setup as a previous study targeting an idealized TC (Miyamoto and Takemi 2013), which was based on Rotunno and Emanuel (1987). The differences in this setup are the grid reso- lution, the lateral boundary, and the domain size. The calculation domain covered 3040 × 2960 km2 with 5 km horizontal grid spacing and a doubly periodic lateral boundary. The number of vertical layers was 40, and the layer thickness was gradually stretched from 200 to 1040
Fig. 1 Look-up table for the charge density of graupel with one collision, used for the charge separation process in SCALE. The data originated from Figure 8 of Takahashi (1978 )
Sato et al. Progress in Earth and Planetary Science (2019) 6:62 Page 3 of 13
Negatively charged ice/snow at z ~ 10 km
Negatively charged graupel at z ~ 10km Negatively charged ice/snow at z ~ 7 km
Positively charged graupel at z ~ 7km
GraupelRedPositiveBlueNegative
Ice/snowRedPositiveBlueNegative
Area where graupels are charged negatively
Area where graupels are charged positively
Charge density that graupel obtained by one collision
Temperature [oC]
Li qu id w at er c on te nt [k g m -3 ]
Profile of Charge density
Large Electric field → Lightning
E
E
Domain averaged vertical profile of simulated Water mass, charge density and charge separation rate
(Sato et al. submitted to ASL)
In case 2017, graupel distributed more upper-layer. While, graupel mainly distributed lower layer (z~6~7 km) in case 2018
Due to the large amount of graupel at z = 10 km in case 2017, charge separation rate and charge density between z = 6 km to z = 14 km were larger than those in 2018
The large charge density in case 2017 resulted in large electric field, and therefore high lightning frequency
Mixing ratio Charge density (ρ) Charge separation rate (Δρ)
GreenTotal hydrometeor RedGraupel BlueLiquid
51
831
Fig. 15. Sche a ic f ai fa e e f (a) he 2018 ca e a d (b) f he 2017 ca e. 832
833
51
831
Fig. 15. Sche a ic f ai fa e e f (a) he 2018 ca e a d (b) f he 2017 ca e. 832
833
~10km
zz
~7km
~4km (0)
Graupel distributed
upper layer
Schematic illustration of the difference between case 2017 and case 2018
Large Positive charge above 10 km by ice and snow which obtained large positive charge around z = 10 km
Large Negative charge around z ~ 10 km by graupel which obtained large negative
charge around z = 10 km
Small ρ below z~7km
E
E
E
E
Area where graupels are charged negatively
Area where graupels are charged positively
Graupel distributed
lower layer
Amount of positively charged Ice and snow above z = 10 km and negatively charged graupel were small due to Small Δρ
around z = 10 km
Case 2017
Case 2018
Summary 1. We developed a lightning model (Sato et al. 2019), and implemented it into community model of Japan, SCALE (Nishizawa et al. 2015, Sato et al. 2015)
2. We checked the validity of our model through the comparison between ground base measurement (LIDEN, Ishii et al. 2014)
3. Our model successfully reproduced the difference in the lightning frequency between Case 2017 and Case 2018.
4. The reason of the difference in the lightning frequency is the vertical distribution of the graupel • In case 2017, the large amount of graupel existed upper layer (z ~ 10 km), where the graupel (snow/ice) obtained negative (positive) charge by collision.
• Due to the existence of the large amount of the graupel, the large charge separation occurred around z ~ 10 km in Case 2017.
• The large charge separation rate at z ~ 10 km resulted in the large charge density in Case 2017. • The large charge density in Case 2017 resulted in the large electric field in Case 2017, and high frequency of the lightning
References References about our lightning model • Sato et al. 2019, PEPS, doi: 10.1186/s40645-019-0309-7 • Sato et al. 2021, MWR, doi: 10.1175/mwr-d-20-0110.1
developed in Japan ~ Validation through the comparison with the ground
base measurement ~
Akihiro Hashimoto (Meteorological Research Institute, Japan)
EGU21-6984, https://doi.org/10.5194/egusphere-egu21-6984
What we have done in this study?
1. Development of a lightning model and implement into Community Model of Japan (SCALE: https://scale.riken.jp/)
2. Validation of the lightning model through the comparison with ground base measurement
3. Examine the reason of the difference of lightning frequency between two heavy rain events in Japan
Introduction
Relationship between lightning frequency and heavy rain (Hamada et al. 2015, Hamada and Takayabu 2018)
The intensity of instantaneous rainfall rate has convention- ally been linked to the intensity of convective storms. There is an extensive literature on the characteristics of
severe storms on both global and regional scales, in which these storms are usually characterized as having extremely strong convection. On the basis of spaceborne radar and radiometer measurements, the global distribution of extreme convective events has been relatively well described1–4. There are also many detailed regional studies on the rainfall characteristics of extreme convective events5–7.
On the other hand, there are observational facts that certain types of extreme rainfall events, such as flash flooding storms, do not necessarily accompany extremely strong convection and intense lightning activity, even in regions where severe convective storms are representative extreme weather events8–12. In some regions, heavy surface rainfall is typically associated with
relatively low-echo-top heights13–15. Moreover, heavy oro- graphic rainfall can inherently exhibit lower echo-top heights16. There are also observations which show that convective intensity is not necessarily related to the near-surface rainfall intensity in some specific regions17–19.
Extreme rainfall over short durations could have devastating socioeconomic effects, as one of the key factors of flash flooding. Therefore, such a decorrelation between the top heights of strong echoes and heavy/extreme surface rainfall rates highlights the importance of a detailed study that focuses on the inherent characteristics of extreme rainfall events and contrasts those with extreme convective events.
Despite many detailed regional studies, a global picture of the linkage between extreme convection and extreme surface rainfall rate remains unclear. Here we use a unique long-term record from a spaceborne precipitation radar (PR) to investigate the
GLB, Land, ANN, v7f, p999
R_ONLY
3,000
2,000
1,000
500
200
100
50
10
20
15
10
5
R_ONLY
3,000
2,000
1,000
500
200
100
50
10
20
15
10
5
Reflectivity (dBZ)
H_ONLY RH
H_ONLY RH
Figure 1 | Echo profiles for three types of extreme events. Composite structures of radar reflectivity at extreme pixels within the TRMM observation domain (37.5!S–37.5!N). Colours show joint histograms of effective radar reflectivity and height, superimposed by solid and dashed lines that indicate the mean and s.d. for each height bin, respectively. Solid and dashed lines along the right-hand axis of each panel show the histograms of echo-top heights and 0 !C levels, respectively. The number of samples for the corresponding extreme type is indicated. (a–c) R-only, H-only and RH-extreme events over land, respectively. (d–f) R-only, H-only and RH-extreme events over the ocean, respectively.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7213
2 NATURE COMMUNICATIONS | 6:6213 | DOI: 10.1038/ncomms7213 | www.nature.com/naturecommunications
& 2015 Macmillan Publishers Limited. All rights reserved.
(Hamada et al. 2015 Nat. Comm.)
Hamada et al. (2015) suggested that Heavy rain can be divided three groups based on satellite measurement, R-only, H-only, RH extreme. • RH-extreme: Heavy rain from high convective cloud • R-only extreme: Heavy rain from low convective cloud • H-only extreme: Weak rain from high convective cloud
Why?
Introduction: Relationship between lightning frequency and heavy rain (Hamada et al. 2015, Hamada and Takayabu 2018)
It should be mentioned that in this study, we only focus on the large-scale moisture transport in the free troposphere that is vertically integrated over a wide range of the troposphere. Further investigation is re- quired to specify the height range, which is key to the occurrence of the extreme rainfall events and extreme convective events, to make dynamical and physical interpretations.
Anomaly composite maps for the vertically integrated moisture and its horizontal fluxes are shown in Fig. 7. For the R-extreme events, a positive (i.e., moist) anomaly dominates over the extreme event region both in the free troposphere and the boundary layer (Figs. 7a,c). However, the spatial extents of the positive moisture anomaly region are substantially different between the two layers. In the free troposphere, the positive anomaly extends zonally from southern China to the east of Ja- pan. This feature indicates that the excessive moisture in the free troposphere of the composite region can be attributed partially to transportation from the west by a large-scale flow rather than a direct injection by the extreme precipitation event within the region. There is also a negative anomaly region extending zonally, paired with the positive anomaly. A positive anomaly is also observed in the boundary layer and is attributed to southwesterly moisture flux; however, the spatial extent is quite limited compared with that in the free troposphere. For the H-extreme events, in contrast, negative (i.e.,
dry) anomalies dominate over the composite region in both the free troposphere and boundary layer (Figs. 7b,d). Moisture fluxes are northwesterly in both layers. Posi- tive anomalies are observed instead on the western and eastern sides of the composite region. The negative anomalous moisture fluxes both in the boundary layer and free troposphere over the composite region might seem against the general understanding that moisture inflow at the boundary layer is essential to the initiation and formation of heavy rainfall events over Japan. Further detailed study is needed, but this is partly be- cause about 65% of the R- and H-extreme events occurred over the sea, where forced lifting of moisture inflow at the lowest level by the topography is not es- sential to the initiation of convection, or is simply because smaller-scale flows are not resolved in the re- analysis dataset used in this study. An interesting feature that contrasted with R-extreme events is seen in the free troposphere west of 1108E, where a significant positive moisture anomaly dominates north of 308N and a sig- nificant negative anomaly dominates south of 308N. The patterns of anomalous circulation and temperature in the middle–upper troposphere in this region associated with the H-extreme events (not shown) are quite similar to those associated with the ‘‘warm heavy rainfall days’’ that are determined by Sun et al. (2015, their Figs. 3a,c) over central north China. This possibly indicates a dynamical link between the H-extreme events over southern Japan and warm-type heavy rainfall events over central north China. The differences between the vertically integrated
moisture anomalies associated with R- and H-extreme
FIG. 5. Box-and-whisker plots of (a) size (km2), (b) stratiform- area ratio (%), (c) maximum 40-dBZ echo-top height (km), (d) maximum precipitation-top height (km), (e) flash rate (per convective pixel per minute), and (f) minimum85GHz PCT (K), for R-, H-, and RH-extreme events. The boxes show the interquartile ranges and medians, the whiskers indicate the ranges from 10th to 90th percentiles, and filled dots indicate the averages. Note that the statistics on flash rate are calculated from extreme events with LIS viewing time longer than 80 s (the number of events are 44, 35, and 6 for R-, H-, RH-extreme events, respectively).
1 SEPTEMBER 2018 HAMADA AND TAKAYABU 6939
D ow
(Hamada and Takayabu 2018, JC)
• In R-only extreme, lightning frequency was low even if heavy rainfall occurred.
characteristics of rainfall events that produce the greatest instantaneous rainfall rates in the tropics and subtropics. We further demonstrate statistically a weak linkage between the extreme rainfall events and extreme convective events. Furthermore, the differences in rainfall characteristics between extreme rainfall events and extreme convective events and their regionality are investigated.
Results Robust differences between two extremes. Figure 1 shows composite structures of effective radar reflectivity (Ze) at extreme
pixels where the maximum near-surface rainfall rate or maximum 40-dBZ echo-top height is observed for each type of extreme event for land and the ocean. There are clear and robust differ- ences in the vertical structures between R-only and H-only extreme events. Note that the rain type, that is, convective or stratiform, is not considered, because more than 95% of the extreme pixels were determined as convective for all extreme types. Clear differences are observed in echo profiles between R-only and H-only extreme events, both over land and over the oceans, since the differences are more significant over land than over the ocean. The most striking characteristic is the echo structure of R-only extreme events both over land and over the
R_ONLY
n=1,071 n=1,002 n=363 n=378 n=641 n=622
n=2,078 n=1,345 n=1,513 n=1,177 n=1,698 n=1,281 256
128
64
32
16
8
4
2
1
256
128
64
32
16
8
4
2
1
20
15
10
20 40 60 20 40 60 Reflectivity (dBZ)
EWPAC, Ocean, ANN, v7f, p999 JP, Ocean, ANN, v7f, p999 SWPAC, Ocean, ANN, v7f, p999
20 40 60 20 40 60
Reflectivity (dBZ) 20 40 60 20 40 60
Reflectivity (dBZ)
20 40 60 20 40 60
Reflectivity (dBZ)
AMZ, Land, ANN, v7f, p999 EWAF, Land, ANN, v7f, p999
H_ONLY
R_ONLY H_ONLY R_ONLY H_ONLY
R_ONLY H_ONLY R_ONLY H_ONLY R_ONLY H_ONLY
Figure 2 | Regional variation of extreme event characteristics. The same as in Fig. 1, but for several specific geographical regions. Only the panels for R-only and H-only extreme events are shown for each region. (a) Amazon (70!W–40!W, 15!S–0!, over land). (b) Equatorial western Africa (15!W–15!E, 0!–15!N, over land). (c) Southern North America (110!W–80!W, 22!N–37!N, over land). (d) Equatorial western Pacific (140!E–170!E, 0!–15!N, over ocean). (e) Northwestern Pacific around Japan (120!E–150!E, 22!N–37!N, over ocean). (f) Southwestern Pacific (180!–150!W, 37!S–22!S, over ocean).
RH extreme event fraction (RH/total) 200109–201208 40N
20N
EQ
20S
(%)
70
50
30
10
30W
Figure 3 | Weak linkage between extreme rainfall and convective events. Fraction of the number of RH-extreme events to that of total extreme events in each region.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7213 ARTICLE
NATURE COMMUNICATIONS | 6:6213 | DOI: 10.1038/ncomms7213 | www.nature.com/naturecommunications 3
& 2015 Macmillan Publishers Limited. All rights reserved.
(Hamada et al. 2015 Nat. Comm.)
R-only is main contributors to the heavy rain ↓
Convective clouds with frequent lightning does not always results in heavy rainfall
Lightning frequency used for rainfall estimation
• Lightning frequency have been used for the estimation of the rainfall from the measurement (e.g., Tapia et al. 1998, Xu et al. 2013)
• However, the report of Hamada et al. (2015) indicated that the convective clouds without frequent lightning (R-only extreme) are main contributor to the heavy rainfall.
• The deeper understanding about the difference between R- only extreme and other extreme (RH-extreme) is required
Target cases of this study “Heavy rain events in 2017 and 2018 in Japan”
Heavy rain events on July, 2017Case 2017 • Heavy rainfall triggered by Baiu front • Cumulative precipitation exceeded 800 mm/48 hours • The lightning frequency was high
Heavy rain events on July, 2018 Case 2018 • Heavy rainfall triggered by Baiu front • Cumulative precipitation exceeded 800 mm/48 hours • The lightning frequency was much smaller than that in case 2017.
(Produced by Japan Meteorological Agency)
Precipitation and measured lightning
Weather map (2017/7/5 00UTC)
(Produced by Japan Meteorological Agency)
Precipitation and measured lightning
Weather map (2018/7/6 00UTC)
51
831
Fig. 15. Sche a ic f ai fa e e f (a) he 2018 ca e a d (b) f he 2017 ca e. 832
833
(Tsuji et al. 2020 JMSJ)
Tsuji et al. (2020) reported that: The difference between Case 2017 and Case 2018 was corresponding to the difference between the R-only extreme and RH-extreme →The two cases are good example to understand the difference in the lightning frequency between R-only extreme and RH-extreme (Purpose of this study)
51
831
Fig. 15. Sche a ic f ai fa e e f (a) he 2018 ca e a d (b) f he 2017 ca e. 832
833
Previous studies for the two cases Case 2017 • Kato et al. (2018), Takemi (2018), Kawano and Kawamura (2020), Hashimoto and Hayashi (2020)
Case 2018 • Kotsuki et al. (2019), Moteki et al. (2019), Sekizawa et al. (2019), Shimpo et al. (2019), Hashimoto and Hayashi (2020)
Most of them did not discuss the difference in the lightning frequency between these two cases
Purposes of this study • To understand the reason of the difference in the lightning frequency between R-only extreme and RH- extreme
• To do so, we used the meteorological model coupled with a lightning model, which was developed by our groups.
Model and experimental setup
in enhancing the warm rain process [Takahashi, 1976; Takahashi and Lee, 1978]. Above the freezing level a strong updraft creates high supersaturation and small nuclei are activated. Some of them will act as new ice forming nuclei due to high supersaturation [Hussain and Saunders, 1984]. Small drops collide with small frozen drops and ice crystals carried from a lower level, and graupel are formed. Because the drops are small, low-density graupel grows, with brittle surface branches [Macklin, 1977]. Collision with a large frozen drop or large graupel leads to ejection of ice frag- ments [Takahashi et al., 1995b]. Additional release of latent heat by the nucleation of small nuclei also enhances the updraft.
6. Conclusions
[32] During the MCTEX project, 14 videosondes were launched into clouds over Tiwi Island near Darwin, Aus- tralia. Seven entered a ‘‘Hector’’ squall line, returning data related to the distribution of hydrometeor species and their electrical charges. Knowledge of their positions in relation to the storm allowed the data to be combined in the construction of a composite map of the mass, number, and space charge distributions. [33] The composite images thus obtained showed a two-
step hydrometeor growth process: one at low levels at the front in warm rain freezing and the other at upper levels within main convection by graupel and ice crystal growth. The space charge evolution has been successfully explained by riming electrification. [34] The space charge distribution exhibited a basic tri-
pole structure, but with a positive space charge extended to the anvil. The high ice crystal concentration in the ‘‘Hector’’ clouds was explained by a higher updraft and higher concentrations of small nuclei over the maritime continent than has been observed over the ocean. The small droplets may thus help in forming brittle graupel as ice fragment generators.
[35] The basic concept of growth modes and particle charge evolution obtained in a Hector squall line may be applied to other squall lines, although different microphys- ics in different areas may modify the distribution. It is highly desirable to conduct similar measurements in other areas.
[36] Acknowledgments. This research was conducted as a part of the MCTEX project. The authors would like to thank D. Jasper for his tremendous help during this expedition. Many students, K. Suzuki, T. Kawano, T. Tajiri, M. Sugiyama, and Y. Kushiyama, from Kyusyu University participated. Research was supported by the Japan Aerospace Exploration Agency, the Ministry of Education, Science, Sports, and Culture of Japan, and the Climate Center of Tokyo University.
References Braham, R. B., Jr. (1990), Snow particle size spectra in lake effect snows, J. Appl. Meteorol., 29, 200–207.
Carey, L. D., and S. A. Rutledge (2000), The relationship between precip- itation and lightning in tropical island convection: A C-band polarimetric radar study, Mon. Weather Rev., 128, 2687–2710.
Chauzy, S., M. Chong, A. Delannoy, and S. Despiau (1985), The June 22 tropical squall line observed during COPT 81 experiment: Electrical signature associated with dynamical structure and precipitation, J. Geo- phys. Res., 90, 6091–6098.
Houze, R. A., Jr., and P. V. Hobbs (1982), Organization and structure of precipitating cloud systems, Adv. Geophys., 24, 225–315.
Hussain, K., and C. P. R. Saunders (1984), Ice nucleus measurement with a continuous flow chamber, Q. J. R. Meteorol. Soc., 110, 75–84.
Jorgenson, D. P., and M. A. LeMone (1989), Vertical velocity character- istics of oceanic convection, J. Atmos. Sci., 46, 621–640.
Keenan, T., B. R. Morton, X. S. Zhang, and K. Nuguen (1990), Some characteristics of thunderstorms over Bathurst and Melville Island near Darwin, Australia, Q. J .R. Meteorol. Soc., 116, 1153–1172.
Keenan, T., et al. (2000), The Maritime Continent Thunderstorm Experi- ment (MCTEX): Overview and some results, Bull. Am. Meteorol. Soc., 81, 2433–2455.
Leary, C. A., and R. A. Houze Jr. (1979), Melting and evaporation of hydrometeors in precipitation from the anvil clouds of deep tropical convection, J. Atmos. Sci., 36, 669–679.
Lyons, W. A., T. E. Nelson, E. R. Williams, S. A. Cummer, and M. A. Stanley (2003), Characteristics of Sprite-producing positive cloud-to- ground lightning during the 19 July STEPS mesoscale convective systems, Mon. Weather Rev., 131, 2417–2427.
Macklin, W. C. (1977), The characteristics of natural hailstones and their interpretation, Meteorol. Monogr., 16(38), 65–88.
Marsh, S., and T. C. Marshall (1993), Charged precipitation measurements before the first lightning flash in a thunderstorm, J. Geophys. Res., 98, 16,605–16,611.
Marshall, T. C., and W. D. Rust (1993), Two types of vertical electrical structures in stratiform precipitation regions of mesoscale convective systems, Bull. Am. Meteorol. Soc., 74, 2159–2170.
May, P. T., A. R. Jameson, T. D. Keenan, P. E. Johnson, and C. Lucas (2002), Combined wind profiler/polarimetric radar studies of the vertical motion and microphysical characteristics of tropical sea-breeze thunder- storms, Mon. Weather Rev., 130, 2228–2239.
Miura, K., S. Nakae, T. Sekikawa, and T. Kumakura (1993), Global distribution of Aitken particles over the oceans, J. Atmos. Electr., 13, 133–144.
Podzimek, J. (1980), Advances in maritime aerosol research, J. Rech. Atmos., 14, 35–61.
Simpson, G. C., and F. J. Scrase (1937), The distribution of electricity in thunderclouds, Proc. R. Soc. London, Ser. A, 161, 309–352.
Simpson, J., T. D. Keenan, B. Ferrier, R. H. Simpson, and G. J. Holland (1993), Cumulus mergers in the maritime continent region, Meteorol. Atmos. Phys., 51, 73–99.
Takahashi, T. (1973), Measurement of electric charge of cloud droplets, drizzle, and raindrops, Rev. Geophys., 11, 903–924.
Takahashi, T. (1976), Warm rain, giant nuclei, and chemical balance—A numerical model, J. Atmos. Sci., 33, 269–286.
Takahashi, T. (1978a), Riming electrification as a charge generation mechanism in thunderstorms, J. Atmos. Sci., 35, 1536–1548.
Takahashi, T. (1978b), Electrical properties of oceanic tropical clouds at Ponape, Micronesia, Mon. Weather Rev., 106, 1598–1612.
Takahashi, T. (1984), Thunderstorm electrification—A numerical study, J. Atmos. Sci., 41, 2541–2558.
Takahashi,T. (1990),Near absenceof lightning in torrential rainfall producing Micronesian thunderstorms,Geophys. Res. Lett., 17, 2381–2384.
Figure 7. Conceptual model of hydrometeor growth and space charge. Red circles are for raindrops originating on frozen drops. Blue circles are for raindrops, originating on graupel. Green circles are frozen drops, blue triangles are graupel, and red crosses are ice crystals. A two-step growth process is suggested by warm rain-frozen in front and graupel growth in the upper level.
D16208 TAKAHASHI AND KEENAN: SQUALL PRECIPITATION AND ELECTRICITY
10 of 11
Lightning Model developed by this study (Sato et al. 2019)
The lightning component consists of 1. Charge separation via collision between graupel and ice/snow (Takahashi 1978) 2. Movement of charge density (tracer advection, sedimentation, diffusion and so on) 3. Calculation of electric field (Poisson equation) 4. Neutralization (lighting)Ferrio et al. 2013: Calculate Total lightning
Physical variables required for the lightning component • Charge density of hydrometeor (prognostic) • Electrical potential (diagnose) • Electric field (diagnose)
Takahashi and Keenan (2004)
Basic assumption Lightning in the atmosphere occurs to neutralize the charge density of “cloud hydrometeors”
Implement
Japanese community model
Model and Sensor simulator Model SCALE (Nishizawa et al. 2015, Sato et al. 2015) • Dynamics HE-VI • Microphysics 2-moment bulk (Seik and Nakajima 2014) • Radiation MSTRNXSekiguchi et al. 2008) • Turbulence MYNNNakanishi and Niino 2006) • Lightning Bulk lightning modelSato et al. 2019) • Surface flux Beljaars-type bulk modelBeljaars and Holtslag 1994 • Land model Bucket type model • Urban model Single layer urban canpy model (Kusaka et al. 2001) • Ocean Given from parent model
Sensor simulatorJoint Simulator (Hashino et al. 2013) • Sensor Radar (Masunaga and Kumerrow, 2005) • Wavelength 5.345 GHz CBand radar operated by JMA
Experimental setup Case Case 2017 Case 2018
Calculation period 00 UTC of 2017/7/5 ~ 00 UTC of 2017/7/6
21 UTC of 2018/7/6 ~ 00 UTC of 2018/7/8
Initial and Boundary condition MANL (Δx=5km 50 layer3hours interval) Horizontal grid spacing 1km Number of vertical layer 60 (Δz = 40 ~ 638.5 m) 57 (Δz = 40 ~ 638.5 m)
Calculation domain Case 2017 Case 2018
GreyCalculation domain RedDomain for analyses
(Sato et al. submitted to ASL)
Measurement data • Radar/Raingauge-analyzed precipitation productby Japan Meteorological Agency: JMA • Radar Reflectivity by C-band radar operated by JMA (CAPPI, 1km vertical/horizontal grid resolution)
• Ground base measurement of lighting named LIghtning DEtection Network (LIDEN) operated by JMA (Ishii et al. 2014)
Location of LIDEN
ResultsValidation
Ca se 2 01 7
(7 /5 0 0U TC ~ 7 /6 0 0 U TC )
Ca se 2 01 8
(0 7/ 06 2 1U TC ~ 7 /7 0 0 U TC )
Geographical distribution of the heavy precipitation was reasonably simulated for both case except for the area close to lateral boundary
(Sato et al. submitted to ASL) An artifact by lateral boundary
Vertical distribution (CFAD)
The model overestimated reflectivity, but the difference in the characteristics between two cases was well reproduced
CF AD (C as e 20 17 )
(7 /5 0 0U TC ~ 7 /6 0 0 U TC )
CF AD (C as e 20 18 )
(0 7/ 06 2 1U TC ~ 7 /7 0 0 U TC )
Ca se 2 01 7
(7 /5 0 0U TC ~ 7 /6 0 0 U TC )
Ca se 2 01 8
(0 7/ 06 2 1U TC ~ 7 /7 0 0 U TC )
Cumulative lightning frequency
Geographical distribution of the lightning was reasonably simulated for both case except for the area close to lateral boundary
An artifact by lateral boundary
Temporal evolution Domain accumulated lightning number
0
20
40
60
80
100
120
6 8 10 12 14 16 18 20 22 24
Fl as
h nu
m be
r/F O
Time from 2017 July 5 00 UTC (09 JST) [h]
LIDEN Model
Time from 2018 July 6 21 UTC (06 JST) [h]
Model Model(limit domain)
The model successfully simulated temporal evolution of lightning.
Red Measurement (LIDEN) Blue Model Green Modelwithout area close to boundary
Case 2017 Case 2018
Summary of the Validation
• The model successfully simulated heavy rain, lightning frequency, and contrast of the lightning frequency between Case 2017 and Case 2018.
Results Difference of the lightning frequency between Case 2017 and 2018
Reason of the tri-pole structure
~10km
z
~7km
Upward motion
charge separation, we considered non-inductive charge separation, which is assumed to occur with the rebound of snow/ice particles after the collision with graupel par- ticles, as in previous modeling studies (Takahashi 1978; Mansell et al. 2005). In addition to non-inductive charge separation, inductive charge separation (Ziegler et al. 1991) is often considered by other models. However, inductive charge separation was ignored in this study, because the contribution of inductive charge separation to the total charge density is much smaller than that of non-inductive charge separation (Mansell et al. 2005; Mansell et al. 2010). The charge density of graupel (ρe,g) and ice/snow (ρe,i,
ρe,s) obtained by the rebound after the collision in each time step was calculated based on the stochastic coales- cence equation (cf. Pruppacher and Klett 2010) in the three microphysical schemes as:
dρe;g dt
0 ng r1ð Þni;s r2ð Þπ r1 þ r2ð Þ
% Vg r1ð Þ−V i;s r1ð Þ !! !!Ecoll 1−Ecoalð Þdr1dr2; ð1Þ
dρe;i=s dt
¼ − dρe;g dt
; ð2Þ
where r1, r2, ng, ni,s, Vg, Vi,s, Ecoll, and Ecoal are the radius of graupel, radius of ice/snow, number concentration of graupel, number concentration of ice/snow, terminal vel- ocity of graupel, terminal velocity of ice/snow, collision efficiency, and coalescence efficiency, respectively. δρ’, defined as:
δρ’ ¼ αδρ; ð3Þ
is the charge density separated by one collision and re- bound, and α is given as
α ¼ 5:0 r2=r0ð Þ2Vg=V 0; ð4Þ
based on Takahashi (1984), where Vg is graupel terminal velocity, and V0 and r0 are 8 m s−1 and 50 μm, respect- ively. δρ is obtained from a look-up table (LUT) from Takahashi (1978), as shown in Fig. 1. β is 1, 1 − [(T + 30)/13]2, and 0 for temperature (T) warmer than − 30 °C, between − 43 and − 30 °C, and colder than − 43 °C, re- spectively, based on Mansell et al. (2005). To avoid an unrealistically large charge density, α is not permitted to be larger than 10, based on Takahashi (1984). The electric field (E) is calculated by solving the Pois-
son equation as:
ε ; ð5Þ
E ¼ −∇; ð6Þ where and ε are electrical potential and atmospheric permittivity, respectively, and ρe(x,y,z) is the charge
density of each grid, which is the sum of the charge density of each hydrometeor in each grid. The calcula- tion of Eq. (5) requires a large computational cost, as shown in Table 1, and a fast Poisson solver is necessary. In this study, the bi-conjugate gradient stabilized (Bi- CGSTAB) method (van der Vorst 1992), which was ori- ginally implemented in SCALE, was used to solve Eq. (5). To calculate the charge neutralization, two schemes
(MacGorman et al. 2001; Fierro et al. 2013) were imple- mented into SCALE. Both schemes neutralize the charge density when the electric field exceeds a threshold value (Eint). The initial point of electric discharge is deter- mined randomly among the grids in which |E| exceeds the Eint. By the scheme of MacGorman et al. (2001), the path of the lightning is explicitly calculated, and the discharge occurs along the path. By contrast, in the scheme of Fierro et al. (2013), the discharge occurs within vertical cylinders with a radius of rcylinder around the initial point. The details of both schemes are de- scribed in the literature.
Experimental setup The experimental setup of this study targeted on an ide- alized TC, and it followed almost the same setup as a previous study targeting an idealized TC (Miyamoto and Takemi 2013), which was based on Rotunno and Emanuel (1987). The differences in this setup are the grid reso- lution, the lateral boundary, and the domain size. The calculation domain covered 3040 × 2960 km2 with 5 km horizontal grid spacing and a doubly periodic lateral boundary. The number of vertical layers was 40, and the layer thickness was gradually stretched from 200 to 1040
Fig. 1 Look-up table for the charge density of graupel with one collision, used for the charge separation process in SCALE. The data originated from Figure 8 of Takahashi (1978 )
Sato et al. Progress in Earth and Planetary Science (2019) 6:62 Page 3 of 13
Negatively charged ice/snow at z ~ 10 km
Negatively charged graupel at z ~ 10km Negatively charged ice/snow at z ~ 7 km
Positively charged graupel at z ~ 7km
GraupelRedPositiveBlueNegative
Ice/snowRedPositiveBlueNegative
Area where graupels are charged negatively
Area where graupels are charged positively
Charge density that graupel obtained by one collision
Temperature [oC]
Li qu id w at er c on te nt [k g m -3 ]
Profile of Charge density
Large Electric field → Lightning
E
E
Domain averaged vertical profile of simulated Water mass, charge density and charge separation rate
(Sato et al. submitted to ASL)
In case 2017, graupel distributed more upper-layer. While, graupel mainly distributed lower layer (z~6~7 km) in case 2018
Due to the large amount of graupel at z = 10 km in case 2017, charge separation rate and charge density between z = 6 km to z = 14 km were larger than those in 2018
The large charge density in case 2017 resulted in large electric field, and therefore high lightning frequency
Mixing ratio Charge density (ρ) Charge separation rate (Δρ)
GreenTotal hydrometeor RedGraupel BlueLiquid
51
831
Fig. 15. Sche a ic f ai fa e e f (a) he 2018 ca e a d (b) f he 2017 ca e. 832
833
51
831
Fig. 15. Sche a ic f ai fa e e f (a) he 2018 ca e a d (b) f he 2017 ca e. 832
833
~10km
zz
~7km
~4km (0)
Graupel distributed
upper layer
Schematic illustration of the difference between case 2017 and case 2018
Large Positive charge above 10 km by ice and snow which obtained large positive charge around z = 10 km
Large Negative charge around z ~ 10 km by graupel which obtained large negative
charge around z = 10 km
Small ρ below z~7km
E
E
E
E
Area where graupels are charged negatively
Area where graupels are charged positively
Graupel distributed
lower layer
Amount of positively charged Ice and snow above z = 10 km and negatively charged graupel were small due to Small Δρ
around z = 10 km
Case 2017
Case 2018
Summary 1. We developed a lightning model (Sato et al. 2019), and implemented it into community model of Japan, SCALE (Nishizawa et al. 2015, Sato et al. 2015)
2. We checked the validity of our model through the comparison between ground base measurement (LIDEN, Ishii et al. 2014)
3. Our model successfully reproduced the difference in the lightning frequency between Case 2017 and Case 2018.
4. The reason of the difference in the lightning frequency is the vertical distribution of the graupel • In case 2017, the large amount of graupel existed upper layer (z ~ 10 km), where the graupel (snow/ice) obtained negative (positive) charge by collision.
• Due to the existence of the large amount of the graupel, the large charge separation occurred around z ~ 10 km in Case 2017.
• The large charge separation rate at z ~ 10 km resulted in the large charge density in Case 2017. • The large charge density in Case 2017 resulted in the large electric field in Case 2017, and high frequency of the lightning
References References about our lightning model • Sato et al. 2019, PEPS, doi: 10.1186/s40645-019-0309-7 • Sato et al. 2021, MWR, doi: 10.1175/mwr-d-20-0110.1