Lightning Observations from Earth
and Space: Some Applications
Jim Weinman
University of Washington, Oct 27, 2005
Example of “TRMM” Merged Microwave-IR
3 Hour Rain Product
•
This display on http://trmm.gsfc.nasa.gov runs ~ 10 hrs
behind real time.
• Files are available via ftp://aeolus.nascom.nasa.gov/pub/merged .
within ~ 5 hrs.
• We need more frequent observations!
Global Convection Diagnostic (GCD) :
A Possible Thunderstorm Interpolator
(% Fred Mosher)
Tir = Twv
Tir >Twv
• For deep convective clouds, updraft
brings cloud particles and water vapor
to top of cloud. (IR and Water Vapor
temperature are the same)
• Clouds drifting away from lift will
allow cloud particles to fall. Water
Vapor will remain at original level. (IR
and WV temperatures will be different)
Comparison of GCD and Lightning Distributions
GCD
Lightning
• Variable consistency, but both are continuous measurements.
• GCD responds to vertical motion at the very top of clouds.
• Lightning depends on vertical motion deeper within clouds.
GPM Reference Concept
OBJECTIVES
• Understand horizontal & vertical Core
structure of rainfall, its macro- &
micro-physical nature, & its
associated latent heating
• Train & calibrate retrieval
algorithms for constellation
radiometers
Core Satellite
•Non-sun-synchronous orbit
~ 65° inclination
~400 km altitude
•Dual frequency radar
(NASDA)
Ku-Ka Bands (13.6-35 GHz)
~ 4 km horizontal resolution
~250 m vertical resolution
•Multifrequency radiometer
(NASA)
10.7, 19, 22, 37, 85, (150/183 )
GHz V&H
•TRMM-like spacecraft (NASA)
Constellation
OBJECTIVES
• Provide sufficient global
sampling to significantly
reduce uncertainties in shortterm rainfall accumulations
• Extend scientific and societal
applications
Constellation Satellites
3-hour goal at ~90% of
time
•Revisit time
•Sun-synch & non-sunsynch orbits
•Pre-existing operationalexperimental &
dedicated satellites with
PMW radiometers
600-900 km altitudes
Mean Diurnal Rainfall and Lightning Cycles
CaPE Days 189-231
Mean Rainfall (mm)
Mean C-G Flash Rate
Hour (UTC)
•Current observation frequency may miss significant
convective events.
•The situation has improved with addition of AQUA and 3
NOAA satellite with microwave sounders.
Motivations for Lightning Mapping
•
Better determination of thunderstorm location and
intensity in regions of poor radar coverage.
• Complements the storm information from radar.
• Only technically feasible technology for oceanic and
global coverage.
• Permits continuous observations.
• Aviation (safety and routing)
• Initialization of Numerical Models
(places storms in right places;
redistributes energy )
• Lightning detection is cheap.
Graupel mass
From Ziegler et al.
(2003)
Some
Relationships
in Model &
Observations
Lightning
Lightning rates are proportional to
graupel mass, to graupel volume, to
cloud ice, and to updraft mass flux
through the –10 C level
Compendium of LIS Flash Density vs IWC between 7-9 km
(1998-2000)
IWC (g/m2)
Schematic view of electrified cloud charge distribution
Evolution of storm growth and electrification.
•Total lightning (top panel)
follows the precipitation (middle
panel) and updraft (bottom
panel) as storm grows and
decays.
•C-C lightning starts a bit sooner
and is more abundant than C-G
[After Goodman et al., 1988;
Kingsmill and Wakimoto, 1991].
Lightning
Correlations with
Radar-Inferred
Storm Properties
Consistently highest
correlations with
 Hail Probability or
Severe Hail Probability
or Hail Diameter
 Area with 45 dBZ
( from MacGorman)
Also for maritime lightning?
3 June 2000: Bird City
Parameter
RL
Confidence Level
VIL
0.78
99
30dBZ Height 0.51
95
Hail diameter NA
NA
Hail prob.
0.64
99
Svr hail prob.
0.72
99
Rotation
0.44
95
cells (45dBZ)
0.64
99
45 dBZ Area
0.75
99
cells (10 km)
0.47
95
11 June 2000: Idalia 1
Parameter
RL
Confidence Level
VIL
0.70
95
30dBZ Height 0.84
95
Hail diameter
0.67
99
Hail prob.
0.79
95
Svr hail prob.
0.72
99
Rotation
0.12
-cells (45dBZ)
0.76
95
45 dBZ Area
0.85
95
cells (10 km)
-0.07
--
22 June 2000: Idalia 2
Parameter
RL
Confidence Level
VIL
0.45
95
30dBZ Height 0.35
90
Hail diameter
0.77
99
Hail prob.
0.62
99
Svr hail prob.
0.73
99
Rotation
0.27
-cells (45dBZ)
0.69
99
45 dBZ Area
0.89
99
cells (10 km)
0.70
99
22 June 2000: Burlington
Parameter
RL
Confidence Level
VIL
0.84
95
30dBZ Height 0.93
99
Hail diameter
0.85
95
Hail prob.
0.82
99
Svr hail prob.
0.85
95
Rotation
0.78
95
cells (45dBZ)
0.76
95
45 dBZ Area
0.89
95
cells (10 km)
0.78
95
24 June 2000: Haigler
Parameter
RL
Confidence Level
VIL
0.41
90
30dBZ Height 0.46
95
Hail diameter
0.31
-Hail prob.
0.57
99
Svr hail prob.
0.29
-Rotation
0.46
95
cells (45dBZ)
0.41
95
45 dBZ Area
0.50
95
cells (10 km)
0.53
95
29 June 2000: Wheeler
Parameter
RL
Confidence Level
VIL
0.73
99
30dBZ Height 0.67
95
Hail diameter
0.83
90
Hail prob.
0.58
99
Svr hail prob.
0.74
99
Rotation
0.72
95
cells (45dBZ)
0.60
99
45 dBZ Area
0.86
90
cells (10 km)
0.43
95
Empirical Rain Rate as a Function of Sferics Rate
r
m
r
Probability matched rain
rate, Ri(Fi) (mm/h) ,from
microwave retrieval vs
sferics rate / 15 min .
Ri
∫
r
xel exact pixel- by(
Fi
0
 (R) dR = ∫
0
 (F) dF
Technique does not
require exact pixel- bypixel co-registration.
(Zawadski & Claheiros)
Lightning vs Convective Rain Rate from TRMM Microwave Imager
Rainfall Rate (mm/h)
(mm/h)
Rate(mm/h)
RainfallRate
Rainfall
+ Sferics and from Solomon & Baker Model (1998)
Flash Rate (min-1)
TMI convective rain over 0.6ox0.6o
box vs sferics rate within 15 min of
overpass.
Smooth curve was obtained from
histogram matching
Flash Rate (min-1)
Rain rate vs Flash Rate from Solomon
& Baker, (1998).
The smooth curve is an extrapolation
of fitted curve from measurements on
left.
Electromagnetic Spectrum Terminology
Designation
ELF
SLF
ULF
VLF
LF
MF
HF
VHF
UHF
MW
MmW
extremely low frequency
superlow frequency
ultralow frequency
very low frequency
low frequency
medium frequency
high frequency
very high frequency
ultrahigh frequency
microwaves
millimneter wave
Frequency
Wavelength
3Hz to 30Hz
100'000km to 10'000 km
30Hz to 300Hz
10'000km to 1'000km
300Hz to 3000Hz 1'000km to 100km
3kHz to 30kHz
100km to 10km
30kHz to 300kHz 10km to 1km
300kHz to 3000kHz 1km to 100m
3MHz to 30MHz
100m to 10m
30MHz to 300MHz 10m to 1m
300MHz to 3000MHz 1m to 10cm
3GHz to 100GHz
10cm to 3 mm
100GHz to 500GHz 3 mm to 0.6mm
VHF Lightning Mapping Station
3-dimensional lightning structure in MCSs
VHF Signal
Antenna
Communication
Antennas
Electronics Building
Site north of Chickasha, Oklahoma
• VHF lightning mappers detect pulses of radiation
produced by the electrical breakdown processes of
lightning in a 5 MHz band within a subset of the VHF (50120 MHz) band
• VHF pulses of radiation are then used to reconstruct the
path (map) of CG and Cloud lightning discharges in 2D or
3-Dimensional Mapping within Network Perimeter
• 100-200 meter location accuracy
• Greater than 95% expected flash detection efficiency
• Reduces to 2-Dimensional Mapping well outside of the
Network (~150 km)
• 2 km or better location accuracy
• Greater than 90% expected flash detection efficiency out to 120 km
Schematic view of LDAR
3-D lightning imaging from 7 stations + central processor
Area covered by WSR-88D Reflectivity, LDAR II
Total Lightning and CG Lightning
Fort Worth WSR-88D Radar
Base Reflectivity Image
13 October 2001 at 0105 UTC
DFW LDAR II Sources (red) and
NLDN CG Flashes (black)
detected between
0103-08 UTC 13 October 2001
from N. W. S. Demetriades, Ronald L. Holle and Martin J. Murphy (2004)
Example of LDAR data
• Plan position indicator
comparable to radar (PPI).
• Side views show c-g and
c-c lightning.
• Probability distributions
are shown in the upper
right.
(from P. Krehbiel et al. 2002)
Electromagnetic Spectrum Terminology
Designation
ELF
SLF
ULF
VLF
LF
MF
HF
VHF
UHF
MW
MmW
extremely low frequency
superlow frequency
ultralow frequency
very low frequency
low frequency
medium frequency
high frequency
very high frequency
ultrahigh frequency
microwaves
millimneter wave
Frequency
Wavelength
3Hz to 30Hz
100'000km to 10'000 km
30Hz to 300Hz
10'000km to 1'000km
300Hz to 3000Hz 1'000km to 100km
3kHz to 30kHz
100km to 10km
30kHz to 300kHz 10km to 1km
300kHz to 3000kHz 1km to 100m
3MHz to 30MHz
100m to 10m
30MHz to 300MHz 10m to 1m
300MHz to 3000MHz 1m to 10cm
3GHz to 100GHz
10cm to 3 mm
100GHz to 500GHz 3 mm to 0.6mm
VLF Sferics Detection Networks
•UKMO ( Europe etc., Tony Lee et al.)
•STARNET ( Eastern US and Atlantic, NASA GSFC)
•WWLLN ( Global, NZ & UW)
•Zeus ( Europe and Africa, Athens Observatory,
STARNET- II)
•PACNET ( Pacific, Hawaii, Steve Businger)
•Los Alamos Nat’l. Lab. ( US )
Components of the STARNET Digital Receiver
Cheap accurate timing
Fast signal
processing
Hi speed
Internet
Determination of Arrival Time Difference (ATD)
We know neither the location of the event nor when it occurred.
We can only measure the ATD at pairs of receivers. That is an art form.
Pulse shape at a range of 6,000 Km
Pulse shape at a range of 15,000 Km
Δt
Correlation
Arrival Time Differences Measured by Various Pairs of Five Receivers
Actual Flash
Note that slight errors in measuring ATD can produce streaks
WWLLN: World-Wide Long-range Lightning Network
P.I.s Dick Dowden & Bob Holzworth
<http://webflash.ess.washington.edu>
• Red circles identify operating receivers. (25 and more to
come) Data collected over 40 min. Sizes of dots diminish in
10 min intervals
An Example of WWLLN Sferics Observations
Superimposed on IR Imagery from Geostationary Satellites.
Lightning Production
• To produce lightning, a storm’s updraft
speed usually must be large enough to
loft graupel in the mixed phase region.
(> 6-7 m s-1)
• Graupel also scatters microwave
radiation that is measured by passive
radiometers on operational satellites.
Two-Hourly Distribution of Lightning in
1998 Ground-hog day Super-storm
(1400 UTC, Feb. 2 - 1200 UTC Feb. 3, 1998)
Loop current trigger
Post-frontal convection,
Cloud Dynamics, Houze (1993)
p .472
Power outage in Miami
SSM-I Microwave Image of Rainfall
Sferics Distribution on Dec. 24 1999
After
Before
NCEP Reanalysis of Xmas eve 1999 Storm
Note 960 mb minimum surface pressure
Electromagnetic Spectrum Terminology
Designation
ELF
SLF
ULF
VLF
LF
MF
HF
VHF
UHF
MW
MmW
extremely low frequency
superlow frequency
ultralow frequency
very low frequency
low frequency
medium frequency
high frequency
very high frequency
ultrahigh frequency
microwaves
millimneter wave
Frequency
Wavelength
3Hz to 30Hz
100'000km to 10'000 km
30Hz to 300Hz
10'000km to 1'000km
300Hz to 3000Hz 1'000km to 100km
3kHz to 30kHz
100km to 10km
30kHz to 300kHz 10km to 1km
300kHz to 3000kHz 1km to 100m
3MHz to 30MHz
100m to 10m
30MHz to 300MHz 10m to 1m
300MHz to 3000MHz 1m to 10cm
3GHz to 100GHz
10cm to 3 mm
100GHz to 500GHz 3 mm to 0.6mm
Coincident Observations at 2000UTC on 2/2/1998
85 GHz PCT
brightness
temperature
from TRMM
TRMM LIS
optical
flashes
•
•
•
VLF
sferics
NLDN C-G
strokes
Low PCTs correspond to intense microwave scattering by ice.
LIS observes lightning for 90s. (accuracy 4 km)
NLDN has limited range, ~ 400 km, for CG strokes,
Convective Rainfall Averaged over 0.5o x 0.5o from TMI and Sferics
Feb. 2, 1998
From Chang et al.
Electromagnetic Spectrum Terminology
Designation
ELF
SLF
ULF
VLF
LF
MF
HF
VHF
UHF
MW
MmW
extremely low frequency
superlow frequency
ultralow frequency
very low frequency
low frequency
medium frequency
high frequency
very high frequency
ultrahigh frequency
microwaves
millimneter wave
Frequency
Wavelength
3Hz to 30Hz
100'000km to 10'000 km
30Hz to 300Hz
10'000km to 1'000km
300Hz to 3000Hz 1'000km to 100km
3kHz to 30kHz
100km to 10km
30kHz to 300kHz 10km to 1km
300kHz to 3000kHz 1km to 100m
3MHz to 30MHz
100m to 10m
30MHz to 300MHz 10m to 1m
300MHz to 3000MHz 1m to 10cm
3GHz to 100GHz
10cm to 3 mm
100GHz to 500GHz 3 mm to 0.6mm
Comparison of Microwave Observations of Himalayan Storm
topography
Snow covered
Ice clouds
mountains
.
.
85 GHz
183 + 1 GHz
Comparison between PR radar and lightning in Himalaya Region, 6/8/03
PR RHI
(solid line)
Note 40 dBZ
PR RHI
(dashed line)
)
Red curve outlines topography
@ 17 km!
LIS
lightning
location
Location
of
transects
Lightning Data
Assimilation into Weather
Forecast Models
-12 Hours
0
Assimilation Period
+12 Hours
Forecast Period
Assimilation by
• Estimating latent heat release
• Influencing the convective trigger function
― Simply turn convection on or off
― Influence character of convection
•
Nudging latent heating and IWV distribution
From MacGorman
Rainfall Forecast from Assimilation using Convective Trigger Function
19 July, 2000
3-Hour Accumulation Valid at 0600 UT
12
9
6
3
0
mm
COAMPS Model
82 X 70 X 30 grid, 22 km spacing
(from Mac Gorman)
Model vs Accumulated Rainfall, 7/19/2000.
03-0600 UTC Model Rainfall
12
0600 UTC Composite
NWS Radar
9
6
3
0
mm
From MacGorman
Continuous Modification of IWV Distributions
Verification
Dry region
Bogus IWV field
EFFECT ON MM5 OF INCLUDING CONVECTIVE ACTIVITY
a) Only 0.5 of latent heating is
utilized. Little improvement.
b) 0.7 of latent heating is
utilized. As good as modeled.
c) 2 x latent heating used in model.
Augmented rain band in correct
location.
d) Latent heating perturbed by
+ 50% random noise. Rain band
is similar to original retrieval.
Conclusion: Amount of latent
heat is not critical, but location is.
Effect of including 6 hr lightning data into a 9 hr forecast
TRMM radar
reflectivity
@ 5 km (dBZ)
TRMM radar
RHI display
Control forecast
cross section
lines: rain rate
(mm/h)
colors: vertical
motion (μb/s)
Lightning assimilated forecast
Global Flash Rate Density:
LIS and OTD
-70
Courtesy of Hugh Christian, NASA/MSFC
NASA Lightning Mapper May Fly on
GOES-R (2012) and possibly on TGM
ZINGER! ECMWF IS GETTING INTO THE
LIGHTNING
Indirect Validation
of GAME
Convective Activity (ECMWF)
Flashes /km2 /month
Mean of two 1-year
ECMWF model runs
T95 L60
5-year LIS/OTD
climatology
(Christian et al.,
2003)
(Lopez and Bauer,2005)
The End
and Thank You
Cliff: This was an attempt to present hourly lightning activity
in a cockpit
THE EXPERIMENT (AWC product Atlantic sector)
Floyd (9/99) and Lightning
Some Applications of CG Mapping
• Warning of the lightning hazard
itself
• Thunderstorm detection,
particularly with poor or absent
radar coverage. (agriculture,
disaster forecasting, aviation
safety and routing))
• Storm system configuration,
growth, and reformation
• Initialization of Numerical Models
(storms in right places;
redistribution of energy)