1Steven Goodman, 2Richard Blakeslee, 2William Koshak, and 3Douglas Machwith contributions from the GOES-R GLM AWG and Science Team

1GOES-R Program Senior Scientist, NOAA/NESDIS GPOhttp://www.goes-r.gov

Direct Readout Conference

Miami, FLApril 4-8, 2011

GOES-R Geostationary Lightning Mapper (GLM): Algorithm and Instrument Report

GOES-R Geostationary Lightning Mapper (GLM)

GLM Characteristics• Staring CCD imager (1372x1300 pixels)

- Single band 777.4 nm- 2 ms frame rate - 7.7 Mbps downlink data rate• Near uniform spatial resolution to 52

deg N lat- 8 km nadir to 14 km at edge

- 70-90% flash detection• Product Availability (event, group, flash)

- < 20 sec total latency- satellite rebroadcast and internet

GLM Combined R/S Coverage Sensor Unit Design at CDR

EU

GLM

3

Natural Hazards and Lightning

• Tornadoes• Hailstorms• Wind• Thunderstorms• Floods• Hurricanes• Volcanoes• Forest Fires• Air Quality/NOx

Volcanic Lightning: Eruption of Redoubt Volcano

NM Tech

Volcanic Lightning: Eyjafjallajökull Eruption, Iceland 2010

Reykjavik

NM Tech

May 3, 1999 Tornado Outbreak

1-minute of observations from TRMM/LISLightning superimposed on TRMM/VIRS IR image

7

Total Lightning Increases with Storm Growth and Updraft Intensification

GOES-R Risk Reduction ScienceLightning Jump Algorithm

Algorithm POD FAR CSI HSS

Gatlin and Goodman*

90% 66% 33% 0.49

Gatlin and Goodman 45

97% 64% 35% 0.52

2σ 87% 33% 61% 0.75

3σ 56% 29% 45% 0.65

Threshold 10 72% 40% 49% 0.66

Threshold 8 83% 42% 50% 0.67

• Six separate lightning jump configurations tested

• Case study expansion:– 107 T-storms analyzed

• 38 severe• 69 non-severe

• The “2σ” configuration yielded best results – POD beats NWS performance

statistics (80-90%); – FAR even better i.e.,15% lower

(Barnes et al. 2007)• Caveat: Large difference in

sample sizes, more cases are needed to finalize result.

• M.S. Thesis completed and study accepted to JAMC (Schultz, Petersen, Carey 2009); forms the conceptual basis of the lightning jump algorithm

Thunderstorm breakdown:

North Alabama – 83 storms

Washington D.C. – 2 storms

Houston TX – 13 storms

Dallas – 9 storms

(*Gatlin and Goodman, JTECH, Jan. 2010)

Proving the Utility of Total Lightning• Examined total and CG rates in the same set of 711 thunderstorms

– 784 SVR reports-hail, wind, tornado– 133 tornado reports (F0=55, F1=46, F2=16, F3=11, F4=5, F5=0)– Total lightning trends outperform CG lightning trends

• Schultz et al., WAF, accepted (in revision)

GOES-R Performance Measure Target7 min increased lead time for tornadoes, National average= 13

min

GLM Product DefinitionsProducts: Events- Level 1B, Groups and Flashes-Level 2:

» Events = A single pixel* illumination in one CCD frame (2 ms) time.» Groups = Multiple adjacent events that all occur at the same time.» Flash = A cluster of groups; the groups must be within prescribed distance and

time limits of each other (i.e., a connected sequence of groups occurring over one or more frames- initial criteria based on LIS uses within 16 km, 300 ms).

– *(GLM FPA 1372x1300 pixels, 8 km at nadir to 12 km at edge of fov)

The GLM Lightning Cluster-Filter Algorithm initially will follow the same definitions applied with the heritage EOS instruments (LIS, OTD):

» Flash Time = Time of the first detected lightning event in a flash.              » Flash Location = The optical radiance-weighted centroid of the flash. That is, one

would use the radiance-weighted latitude (RWLAT) and the radiance weighted longitude (RWLONG) to characterize the “location” of a flash.  

» Flash Footprint = Unique areal extent of the flash (measured in square kilometers). Note: some flashes, called “spider lightning”  can extend hundreds of kilometers in the horizontal.

GRB Format: CCSDS binary filesLevel 2 Product Formats: netCDF4, McIDAS Areas

11

A Time-Resolved Ground Flash(flash consists of event and group components)

time

Group a

Group b... etc.

.

.

.etc.

Altit

ude

Plan View(CCD Array)

Groups Help Us Track the Strokes and other components of the lightning flash!

(Methodology based on 12 years successful on-orbit experience with TRMM LIS)

• Dots (red, green, blue) are LMA* data• Gray squares are (simulated) GLM data• Time is indicated by color

– Red first– Green next– Blue last

• GLM radiance is indicated by greyscale (darker = greater amplitude)

• Shown is a single flash with 2 groups and 20 events

– Amplitude weighted centroid is indicated by the large X

– Time of flash is time of first event– The two groups (red & blue) are close

enough in time/space to be clustered into a single flash (16.5 km & 330 ms)

– In this example, the green LMA pulses did not create an optical pulse large enough to be detected by the (simulated) GLM (below threshold) 12* LMA: VHF Lightning Mapping

Array (GLM Proxy Data source)

Testing and Validation(data used in generating GLM proxy)

• Key:• LIS (squares) • LMA (dots)• NLDN (Xs)

• Rule maps VHF to optical.• Optical represents flash

seen by GLM.• Repeat for many VHF

flashes in a storm.• Repeat storm in various

regions of GLM FOV.• See how many of these

repeated storms the algorithm can handle before one “breaks” the code.

• Check accuracy of flash retrievals.

13

Testing Using Suite of Proxy Data

Ground Processing Algorithm Block Diagram

courtesy, LMATC

• Field Campaign– Leverage observing assets associated with CHUVA with U.S. supplied portable

LMA network (and European supplied LINET) to generate proxy data sets for GLM and ABI that include total lightning (LIS and ground-based) and SEVIRI.

– Allow GLM and Combined AWG/Science teams to better address and assess several areas of on-going research

15

• Science Objectives:o Algorithm and Proxy Data Validationo Validation Systems Performance

Assessmento Storm Electrification/Physicso Applications for GLM+ABI+…

• Key scientific measurements: VHF 3-D Lightning Mapping Array (LMA), LINET, TRMM/LIS, MSG SEVERI (ABI proxy data), high speed digital video, VLF lightning networks, dual-pol radar, electric field change, airplane in-situ microphysics, and ancillary meteorological data

CHUVA Ground Validation IOPSao Paulo, Brazil 2011-2012

TRMM/LIS MSG/SEVERI

Partners and CollaboratorsGLM Science Team

Richard Blakeslee, NASA (LMA lead)Larry Carey, Jeff Bailey, UAH (NOAA funded to deploy LMA)John Hall, UAH (Web support for real time network operations)Monte Bateman, USRA (proxy data, other analyses)Many others (and other GOES-R teams) (algorithm and proxy val)

InPE (CPTEC, ELAT) and USPLuiz Machado, InPE/CPTEC (overall CHUVA lead)Rachel Albrecht, InPE/CPTECCarlos Morales, USP (Electrification processes lead)Osmar Pinto Jr., InPE/ELAT Research Group

EUMETSAT MTG Lighting Imager Science TeamHartmut Hoeller, DLR (LINET lead) (Collaborator)

Commercial Data ProvidersVaisala LS8000, WeatherBug WTLN (Collaborator)

16

HWT: Forecaster Feedback from 2010

• “We saw several instances where the total lightning was picking up on storms before the AWIPS lightning [NLDN] program picked up on them. One could see the utility of this in the future, bringing with it a potential for lighting statements and potentially lightning based warnings.”

-Pat Spoden (SOO, NWSFO Paducah, KY)

http://ewp.nssl.noaa.gov/

http://goesrhwt.blogspot.com/

K. Kuhlman

• “[GLM] will also prove very beneficial as we get more into decision support services, especially to support the safety of responders to incidents who are exposed to lightning hazards.”

-- Frank Alsheimer (SOO, NWSFO Charleston, SC)

“lightning data provide a reassurance that I can see leading me to make a warning decision a little earlier due to having more confidence in imminent severe weather.”

“would be of great benefit to aviation forecasting in those situations where there is a developing shower or embedded thunderstorms in stratiform rain.”

K. Kuhlman

HWT: Forecaster Feedback from 2010

A Vision for Operations

2 sigma threshold reached

Schematic courtesy of Kristin Kuhlman (OU/NSSL): Trending Display for Total Lightning

Summary

• The GOES-R GLM offers information on high impact weather phenomena-storm growth, decay and intensity

• The GOES-R Proving Ground provides mechanism to:– Involve CIs, AWG, National Centers, NOAA Testbeds and WFOs in

user readiness– Get prototype GOES-R products in hands of forecasters– Keep lines of communication open between developers and

forecasters– Allow end user to have say in final product, how it is displayed and

integrated into operations• Opportunities for forecast demonstrations of experimental GLM

total lightning products