This document summarizes a study analyzing thunderstorm morphology in South America using TRMM satellite data from 1998-2011. Key findings include: 1) Thunderstorms peaked in January and October annually, with diurnal peaks between 13-17 hours local time. The most frequent thunderstorms occurred in central South America. 2) The highest lightning and thunderstorm densities were found along the northern coast of South America and in regions with high topography. 3) Severe thunderstorms were identified and had very high flash rates. The most severe storms occurred in areas with complex terrain in northern and central South America.

1. Morfologia das tempestades el´etricas na Am´erica do
Sul
Evandro M. Anselmo
Institute of Astronomy, Geophysics and Atmospheric Science of University of S˜ao Paulo
(IAG-USP), S˜ao Paulo-SP, Brazil.
IAG/USP – 2015

2. Introduction
South America is one of the 3 lightning chimes
(http://thunder.msfc.nasa.gov/data/)

3. Introduction
Infrastructure planning requires knowledge about thunderstorm position,
frequency and lightning efficient production

4. Objectives
Create a thunderstorm database based on TRMM measurements
Determine the annual cycle and diurnal cycle of thunderstorms in South
America
Determine the geographical distribution of lightning flashes and
thunderstorms
Evaluate the thunderstorms severity.

5. Tropical Rainfall Measuring Mission – TRMM
Launched in November 28th, 1997. Orbit with inclination of 35◦
and
altitude of 350 km (August, 2011, the altitude was change for 402.5 km)
Sensors: Precipitation Radar – PR, TRMM Microwave Imager – TMI,
Visible and Infrared Scanner – VIRS, Lightning Imaging Sensor – LIS,
Clouds and the Earth’s Radiant Energy System – CERES.
Goal: Estimate rainfall over the tropics and subsequently latent heat flux1
1KUMMEROW, C.; BARNES, W.; KOZU, T.; SHIUE, J.; SIMPSON, J. The tropical rainfall measuring mission
(TRMM) sensor package. J. Atmos. Oceanic Technol., v. 15, p. 809–817, 1998

6. Tropical Rainfall Measuring Mission – TRMM

7. Tropical Rainfall Measuring Mission – TRMM
TRMM overpasses

8. Tropical Rainfall Measuring Mission – TRMM
LIS processing

9. Data and Methodology
TRMM:
TRMM orbital v7 files: LIS, VIRS (1B01), and PR (2A25), between
1998–2011 ( 30TB of data).
During this period we had 79,932 TRMM orbits, but only 63,613 were
over South America (SA) 40◦
S–10◦
N and 90◦
–30◦
W.
NCEP RII:
NCEP RII reanalysis from 1998-2011: geopotential height and
temperature at 17 pressure levels. This data is used to convert TRMM PR
altitudes levels in temperature.
Defining thunderstorms:
Thunderstorms have been defined as clouds with brightness temperature
below 258 K at the 1B01 10.8 µm channel and had at least one LIS
lightning flash [Morales and Anagnostou, 2003]2
.
2Morales, C. A., and E. N. Anagnostou, Extending the capabilities of high-frequency rainfall estimation from
geostationary-based satellite infrared via a network of long-range lightning observations, J. Hydrometeor, 4,
141–159, 2003.

10. TRMM’s thunderstorms (flow chart)
Locate
thunderstorms
clusters (Tb ≤ 258
K and at least one
flash)
VIRS (1B01)
channel 4 (10.8 µm) LIS flash
PR (2A25): Ze, sur-
face rain, rain type
(2A23), geolocation
LIS: flashes, events,
groups, view
time, geolocation
VIRS (1B01): chan-
nel 4, geolocation
Write the
TRMM thun-
derstorm HDF file
read
read
read

11. TRMM’s thunderstorms
157,592 thunderstorms were found!
Only 94,733 had view time (V Tm) greater than 1 minute and with at least
one PR pixel with valid rain.

12. The onset of the South American thunderstorms
Where the thunderstorms are more often in SA?
Where are the places with high lightning densities that corresponds with
the highest density of thunderstorms?
How long does the thunderstorm season last in SA?

13. Onset: TRMM overpasses and LIS view time
(LIS view – VTlis) (VIRS overpasses – VTvirs)

14. Onset: LIS lightning flashes and number of
thunderstorms
(lightning – FLlis) (thunderstorms – Pte)

15. Onset: Defining densities ratios
lightning [km−2
year−1
]
DEfl =
FLlis
VTlisAg
31557600 (1)
thunderstorms per orbit [km−2
]
DEte =
Pte
VTvirsAg
(2)
lightning per thunderstorms [km−2
year−1
]
DErt =
FLlis
VTlisAgPte
31557600 (3)

16. Where the thunderstorms are more often in SA?

17. Onset: lightning × thunderstorms – Annual

18. Onset: lightning × thunderstorms – SON

19. Onset: lightning × thunderstorms – DJF

20. Onset: lightning × thunderstorms – MAM

21. Onset: lightning × thunderstorms – JJA

22. Where are the places with high lightning densities that
corresponds with the highest density of thunderstorms?

23. Onset: lightning per thunderstorms – Annual

24. How long does the thunderstorm season last in SA?

25. Onset: Diurnal and Annual Cycle

26. Annual cycle

27. Diurnal cycle

28. Thunderstorms Severity

29. Thunderstorms Severity: Defining severity index
Flash rate (FT) is defined as the ratio of number of lightning flashes (Nfl)
by the mean view time (V Tm) in the thunderstorm area extracted.
FT =
Nfl
V Tm
60 [min−1
] (4)
Flash rate normalized by thunderstorm area (FTA) is defined as the ratio
of number of lightning flashes (Nfl) by the mean view time (V Tm) and
thunderstorm area (At).
FTA =
Nfl
V TmAt
60 [min−1
km−2
] (5)

30. Thunderstorms Severity: FTA and FT pdf distribution

31. Thunderstorms Severity: defining the extreme FTA and
FT thunderstorms
Extreme: 90th percentile
FTA: 29.3 to 1,258.7 × 10−4
fl min−1
km−2
FT: 47.2 to 1,283.6 fl min−1

32. Thunderstorms Severity: Rain and CV/ST fraction
Thunderstorms with extreme FTA: rain areas
Thunderstorms with extreme FT: rain areas

33. Thunderstorms Severity: Top severe FTA index

34. Thunderstorms Severity: Top severe FT index

35. Thunderstorms Severity: CFAD – FTA (w/o lightning)

36. Thunderstorms Severity: CFAD – FT (w/o lightning)

37. Thunderstorms Severity: CFAD – FTA (with lightning)

38. Thunderstorms Severity: CFAD - FT (with lightning)

39. Thunderstorms Severity: CFADs
FTA pixels are taller than FT.
FTA pixels have higher Zc values than FT, especially above 5 km.

40. Thunderstorms Severity: temperature dependency
It is know that the mixed phase region regulates the electrification charge
process.

41. Thunderstorms Severity: temperature dependency
As a consequence, we converted the PR altitude levels into temperature
levels based on RII reanalysis.
Now we created the Contoured Frequency by Temperature Diagram
(CFTD) in addition to the cumulative CFTD (CCFTD).

42. Thunderstorms Severity: CFTD - FTA (with lightning)

43. Thunderstorms Severity: CFTD - FT (with lightning)

44. Thunderstorms Severity: CCFTD-FTA (with lightning)

45. Thunderstorms Severity: CCFTD-FT (with lightning)

46. Thunderstorms Severity: Rate of Zc change with
temperature
Above 0◦
C we might have ice and supercooled water droplets.
The rate of change of Zc with temperature might give an idea about of
extend of mixed phase layer.
We computed this rate for 30%, 50%, 70% and 95%.
We compare tropical versus sub-tropical regions (Amazon × La Plata
basin)

47. Thunderstorms Severity: Zc change – La Plata basin
(30S-40S and 60W-70W)

48. Thunderstorms Severity: Zc change – Amazon
(00N-10N and 60W-70W)

49. Thunderstorms Severity: identifying the efficient and
severe regions
For each grid box with 2.5◦
× 2.5◦
we obtained the FTA and FT pdf.
For each grid box, we extracted the 99th percentile value.
Based on 99th percentile FTA and FT values, we build a spatial
distribution map that show where are the most efficient (FTA) and severe
(FT).

50. Thunderstorms Severity: the most efficient thunderstorms

51. Thunderstorms Severity: the most severe thunderstorms

52. Thunderstorms Severity: where are the most efficient
and severe thunderstorms
Find the thunderstorms that have FTA and FT above 90th percentile.
Compute the geographical densities of these thunderstorms (normalized by
TRMM overpasses).

53. Thunderstorms Severity: the most severe and efficient
thunderstorms

54. Conclusions
This study characterized the severe thunderstorms over South America by
employing 14 years of TRMM measurements.
The thunderstorms were characterized by integrating TRMM LIS, PR and
VIRS measurements.
A total of 157,592 electrified clouds have been observed in South America
(40S-10N and 90W-30W).
For the severe thunderstorms database, only 94,733 thunderstorms were
used. Those thunderstorms had LIS view time greater and equal to 60
seconds and at least one valid PR raining pixel .

55. The onset of the thunderstorms
Diurnal cycle:
40% of thunderstorm occurred between 13h–17h
(LT)
In the ocean, we found one maximum at 20h (LT)
and another between 4–5 (LT)
The nocturnal peak of thunderstorms were found in
extreme North of Andes Mountains (Paramillo
National Natural Park, Maracaibo lake). 630
systems where observed in 14 years, only between 0h
e 00:59h
The thunderstorms were more frequent in center of
South America (10◦
–0◦
S and 70◦
–50◦
W and
20◦
–10◦
S and 60◦
–50◦
W), between 14h–16h (LT).

56. Onset:
Annual cycle:
Thunderstorms season in South America was
between October and March with two peaks:
January and October.
Northeast of South America (0◦
–10◦
N and 70◦
–50◦
W) the peak of thunderstorm was found in August.
In South of South America (40◦
–20◦
S and 70◦
–60◦
W, Plata basin) was found a shorter thunderstorm
season with 2 months.
The longer season was found in Colombia and West
of Venezuela with 9 months.

57. Onset: Numbers of thunderstorm associate with
seasonal geographical densities of lightning and
thunderstorms
1◦
Spring = 57,861
2◦
Summer = 46,077
3◦
Autumn = 36,804
4◦
Winter = 16,850

58. Onset: Total geographical densities
Thunderstorms:
North and Northwest of South America (Colombia, Panama, Northwest of
Amazon): 3.5–4.7 × 10−4
km−2
Regions with high topography: Northeast of Titicaca Lake in Peru,
mountain range in Santa Catarina, Emas National Park in Goias: 2.5 ×
10−4
km−2
Lightning:
Mouth of the Catatumbo River: 148.1 fl year−1
km−2
Cochabamba, Bolivia: 60 fl year−1
km−2
The Black Needles peak, Matiqueira mountain range, Neblina peak, La
Plata basin: 30–60 fl year−1
km−2
.
Lightning per thunderstorm:
11.73 × 10−2
ano−1
km−2
(Maracaibo lake). In this grid point (772 km2
)
each thunderstorm had in average 91 fl year−1
.

59. Severity of Thunderstorms in South America
Two groups of severe systems are found:
FTA 29.3 to 1,258.7 × 10−4
fl min−1
km−2
FT 47.2 to 1,283.6 fl min−1
(above category 3 of Cecil et al. [2005] and
Zipser et al. [2006])
Cloud top Tb: FTA thunderstorms are 10 K colder than FT
thunderstorms
Convective × Stratiform rain fraction: FTA thunderstorms have 72%
of convective rain fraction and 32% stratiform and FT systems 22%
convective and 65% stratiform.
Lightning × non Lightning pixels (CFAD): profiles with lightning
pixels have higher Zc values, and are taller. In the layer between 5 and 7
km height, the Zc difference can reach 5 to 10 dBZ. At lower levels
(2–3km) Zc is above 50 dBZ compared to 40 dBZ for non-lightning pixels;
Lightning pixels (CFAD): FTA thunderstorms show 1-3 dBZ higher Zc
values than FT systems, specially above 5 km height;

60. Top 1% FTA thunderstorms: More than 148.93 × 10−4
fl min−1
km−2
. Associated with orography in the Pantanal Matogrossense, Central
Plateau of Brazil, and the Xingu and Araguaia basin and Tocantins rivers
in the Amazon basin, the meridional Plateau of Brazil in Paran´a basin
and the Andes foothill – eastward of the Sierra of Cordoba in Argentina.
Top 1% severe FT thunderstorms: More than 272.88 fl min−1
and are
located in the elevated topography areas like Andes foothill and Sierra of
Cordoba in Argentina and also at less pronounced orography at Tocantins
central of Brazil, Sierras Gauchas and Catarinenses at South of Brazil to
more flat areas that have a water and vegetation contrast like south of
Amazon, Rio Branco in Acre, in the boundaries of the states of Acre,
Amazon with Peru, central Bolivia, Pantanal Matogrossense, Paran´a state,
Paraguay and a Plata basin.

61. Severity: Temperature dependence – CFTD
Zc distribution: FTA thunderstorms show broader Zc distribution than
FT systems for the same temperature level, specially above 0◦
C and they
have 1–3 dBZ higher Zc values;
Latitudinal variation: FT thunderstorm show an enhancement of the
collision-coalescence process towards the sub-tropics.
Zc change: the maximum Zc decrease with temperature for FTA(FT)
thunderstorms is at -2◦
C (0◦
C) and -12◦
C(-8◦
C) for 50% and 95%
levels profiles respectively in the Amazon and -6◦
C(-2◦
C) and -14◦
C (-6◦
C) for 50% and 95% respectively for the La Plata Basin.

62. ACKNOWLEDGEMENTS
This work is part of PhD project that is supported by CNPq grant
140842/2011-0.
This work is partially supported by CAPES PROEX program and
COELCE.
I would like to thank MSFCNASA for providing LIS data set and
GSFCNASA for providing TRMM dataset.
Finally I would like to thank Dra. Rachel Albrecht for the LIS lightning
and view time reprocessed dataset.