Proxy Data and Proxy Data and VHF/Optical VHF/Optical

ComparisonsComparisons

Monte BatemanMonte Bateman

GLM Proxy Data DesignerGLM Proxy Data Designer

Lightning Imaging Sensor (LIS)Lightning Imaging Sensor (LIS)on TRMMon TRMM

Satellite based optical detectorSatellite based optical detector Low-Earth OrbitLow-Earth Orbit Can detect daytime or nighttime Can detect daytime or nighttime

flashesflashes view time of about 80 sview time of about 80 s

North AlabamaNorth AlabamaLightning Mapping Array (LMA)Lightning Mapping Array (LMA)

10 sensor network; ground-based10 sensor network; ground-based VHF / TOA systemVHF / TOA system 24 x 724 x 7 can observe a storm throughout can observe a storm throughout

its lifetime, if it stays in rangeits lifetime, if it stays in range(about 200 km from network (about 200 km from network center)center)

Instrument ComparisonInstrument Comparison

LISLIS LMALMA

Satellite (LEO) basedSatellite (LEO) based ground basedground based

80s view time80s view time view time limited only by view time limited only by rangerange

Range = FOVRange = FOV Range = 150–200 kmRange = 150–200 kmfrom network centerfrom network center

optical detector (777.4 optical detector (777.4 nm)nm) VHF/TOA networkVHF/TOA network

2ms integration time2ms integration timeno dead time—500 no dead time—500 frames/sframes/s

Samples at 25 Ms/s; Samples at 25 Ms/s; reports one peak in 80µs reports one peak in 80µs windowwindow

Global coverage (± 35°)Global coverage (± 35°) North AlabamaNorth AlabamaDetection Efficiency ~ Detection Efficiency ~ 70%70%

Detection Efficiency ~ Detection Efficiency ~ 85%85%

Can we generate GLM proxy Can we generate GLM proxy from LMA?from LMA?

Extensive statistical comparison of Extensive statistical comparison of NALMA, LIS, & NLDN dataNALMA, LIS, & NLDN data

Most flash characteristics (size, Most flash characteristics (size, #events, duration, etc.) did not #events, duration, etc.) did not correlate between the systems.correlate between the systems.

Bottom line: LIS & LMA look at Bottom line: LIS & LMA look at completely different physics...completely different physics...

What we do know...What we do know...

LIS sees more pixels light up for CG LIS sees more pixels light up for CG flashes than for IC flashes (save for flashes than for IC flashes (save for later)later)

The more cloud above the lightning The more cloud above the lightning channel, the less light gets to LIS.channel, the less light gets to LIS.

That is, flashes that are higher in the That is, flashes that are higher in the cloud appear brighter to LIScloud appear brighter to LIS

Can we see this?Can we see this?

LMA Altitude vs. # LIS eventsLMA Altitude vs. # LIS events1 year of

N. AL data

Sidebar: LIS definitionsSidebar: LIS definitions

Event: pixelEvent: pixel

Group: pixels that light up at the Group: pixels that light up at the same time and touch each othersame time and touch each other

Flash: A collection of groups that Flash: A collection of groups that satisfy time and space criterion for satisfy time and space criterion for our best guess as to what is a our best guess as to what is a lightning flash.lightning flash.

GLM Proxy data, v1GLM Proxy data, v1

With the relation shown, and a With the relation shown, and a technique called “distribution technique called “distribution matching”matching”

Using storm data from the LMA, could Using storm data from the LMA, could look up “how many pixels” neededlook up “how many pixels” needed

We had no idea how to distribute We had no idea how to distribute those in space & time, sothose in space & time, so

For the v1 proxy, we had to simulate For the v1 proxy, we had to simulate the space & time distribution of pixels.the space & time distribution of pixels.

Science + simulationScience + simulation

LIS data as a ProxyLIS data as a Proxy LIS is the best proxy we have for GLMLIS is the best proxy we have for GLM Need to better quantify for proxy dataNeed to better quantify for proxy data Collected statistics on all flashes, groups Collected statistics on all flashes, groups

& events starting in Dec 2001& events starting in Dec 2001 Limited to GLM FOVLimited to GLM FOV Created a 5x5 km grid, fixed to the EarthCreated a 5x5 km grid, fixed to the Earth Binned each event into a grid boxBinned each event into a grid box Collected the size, shape, and temporal Collected the size, shape, and temporal

distribution of each flashdistribution of each flash

Annual

Seasonal

Monthly

2D Grid5x5 km bins

(sample month)

Seasonal5x5 km bins

(20 x 20)

GroupGaps

GroupGaps

GroupGaps

GroupGaps

GroupGaps

GroupGaps

GLM Proxy data, v2GLM Proxy data, v2

Now have the statistics of spatial and Now have the statistics of spatial and temporal distribution for LIS events, temporal distribution for LIS events, groups and flashesgroups and flashes

Need to quantify annual, seasonal, Need to quantify annual, seasonal, diurnal and geographical variationdiurnal and geographical variation

Will build this into the proxy generator Will build this into the proxy generator code to make it much more realisticcode to make it much more realistic

Should not affect the cluster-filter Should not affect the cluster-filter algorithm softwarealgorithm software

Future needsFuture needs

Bring in OKLMA (and others)Bring in OKLMA (and others) Create a “streaming” LMA flash Create a “streaming” LMA flash

algorithm, to allow real-time proxy algorithm, to allow real-time proxy creation for various testbedscreation for various testbeds

Add in CG flash coincidence — need Add in CG flash coincidence — need proxy data for areas with no LMAproxy data for areas with no LMA