Storm/lightning as viewed from the National Mall, Saturday night, July 20 (Kevin Ambrose)

Can lightning predict the severity of a thunderstorm? The short answer:  Sometimes, yes;  sometimes, no.  Lightning arises from a specific combination of factors in a storm cell, not necessarily the same elements that breed severe weather.

Here we explore why some of the deadliest storm cells may or may not produce meaningful lightning activity.

Lightning detection and warning:  An imperfect science

Thunderstorms generate a variety of severe weather:   Damaging wind gusts, large hail, flash floods, intense lightning, and tornadoes.

The National Weather Service officially defines a severe thunderstorm if wind gusts exceed 58 mph and/or hail exceeds one inch diameter.   Flash floods are covered under a separate warning.  Ditto for tornadoes.  However, there is presently NO warning criteria for a cell that generates unusually frequent lightning strikes.  Such storms are inherently dangerous, causing utility outages, fatalities and house fires – even if no other form of severe weather results.

Several sensor networks monitor lightning activity in thunderstorms.  Two of most well-known, and widely distributed, are Earth Networks Total Lightning Network (TLN) and the U.S. National Lightning Detection Network (NLDN).

The Earth Networks TLN consists of nearly 700 sensors.  The majority of these are in the U.S., but the network has expanded into Canada, Mexico and 40 other countries.  Its lightning sensors detect both in-cloud (IC) and cloud-to-ground (CG) activity.

Here’s an animation of the TLN detecting a surge in lightning activity as the June 29, 2012 derecho moves into western Virginia.

The NLDN detects cloud-to-ground (CG) discharges, based on an array of 100 ground sensors that monitor electromagnetic radiation emitted by lightning strokes. The NLDN provides location, time, and polarity of individual strikes.  The data are processed by Vaisala, Inc. in Tucson, Arizona.  An example of lightning strike density for a Washington-area thunderstorm, based on NLDN, is shown below.

NLDN map of CG strikes for a squall line approaching D.C. on the evening of July 7, 2013. Colors indicate strike times. The NLDN detected hundreds of CGs associated with this line. (Screen capture from
NLDN map of CG strikes for a squall line approaching D.C. on the evening of July 7, 2013. Colors indicate strike times. The NLDN detected hundreds of CGs associated with this line. (Screen capture from

Recently, novel techniques using GPS technology and radio emissions (60 MHz) have enabled extremely precise, ultra-fast, and more complete lightning detection.

A type of experimental system called the Lightning Mapping Array (LMA) is being tested in select regions of the U.S., including the Washington-Baltimore corridor.   The LMAs are jointly funded by NASA and NOAA, with research conducted at several universities.   LMAs provide 3D lightning mapping and can resolve individual lightning channels fanning through the storm cloud.   LMAs measure total lightning, which includes both CG and intracloud (IC) discharges. An example of the LMA’s lightning mapping capabilities is shown below.

LMA map of total lightning for the same squall line depicted in Figure 1. Colors indicate lightning density (discharges per 2 km x 2 km pixel). The LMA measured nearly 200,000 discharges during this storm and accurately pinpointed dangerous lightning “hot spots”. (NASA)

Related: Next generation lightning and cloud data highlight tornado-producing storms in Md. | Can lightning help predict tornadoes? A D.C. area case may shed some light | Lightning gone wild during Washington D.C.’s derecho | Dissecting a derecho bolt: more to lightning than meets the eye (and camera lens) | Saturday night’s extreme lightning show north of Washington, D.C.

When lightning provides useful guidance on storm severity

Some types of severe convective storms produce CG discharges, while others do not.   When lightning does occur, a storm’s lightning flash rate (number of flashes per minute) is a useful indicator of a storm’s intensity.   As lightning production ramps up, the cloud updraft enters its most vigorous phase.   The most intense updrafts generate large hail.

Total lightning, as obtained by the LMA, is often a better indicator of a thunderstorm’s intensity level.  Studies have noted a several-fold increase in IC lightning, relative to CG lighting, in tornadic storm cells.  Curiously, a prominent lightning hole – a small, doughnut-shaped region free of lightning – pinpoints the location of a supercell’s mesocyclone.   Mesocyclones are a precursor to strong and violent tornadoes.

A number of studies have also correlated a sharp rise in lightning intensity with imminent onset of severe weather, including tornadoes and downburst winds.   There is usually a time lag between the abrupt rise in lightning frequency, called a lightning jump, and these high winds.   The lag is typically tens of minutes.

The lag may be tied to the sequential pulsing of intense updraft and downdraft phases in a storm cell.    First comes the updraft phase, when a large mass of precipitation forms aloft.  As explained below, this is tied to a fast rise in cloud electrification.  The downdraft phase follows on its heels, as the heavy core of precipitation descends.  When the cell collapses, downburst winds surge across the ground.   Additionally, strong tornadoes initiate near the surface, where the downdraft wraps around the mesocyclone.

When lightning is not a useful severe storm detector

Cloud electrification requires a mixed phase region in the storm cloud.   This is a cloud layer between 0 (32 °F) and -20 °C (-4 °F), containing high concentrations of supercooled water, small ice crystals and large, rimed ice particles.   Rimed ice is otherwise known as graupel, a type of mushy, opaque ice pellet.   The jostling of these particles in a strong updraft leads to charge induction and formation of an electrical field.  Radar studies using polarimetric doppler have shown that without a dense mix of ice and supercooled water, a convective cloud is electrically quiet.

But the absence of this layer does not preclude severe weather.   In fact, one of our nation’s leading weather killers, flash floods (#2 behind heat waves), are most commonly generated by summertime convective clouds.   A number of recent floods – including the June 27, 1995 Madison County, VA flash flood, and July 27, 1997 Ft. Collins, Colorado events – were produced by “thunderstorm” complexes largely devoid of lightning.

The Madison County storm was anchored to high terrain (the foothills of the Blue Ridge) and remained stationary over its lifetime.    It dumped an incredible 24 inches of rain in 8 hours over a small region near Ruckersville.   The ensuing mudslides and debris flows killed three, inundated 80 bridges and damaged or destroyed 2000 homes.  This flash flood defines the record discharge for a stream east of the Mississippi River!

Detailed radar studies indicate that the Madison County storm lacked a mixed-phase region.   Heavy rain formed principally by warm-cloud processes i.e. growth of rain drops by collision-coalescence, which occurs at temperatures warmer than freezing, near the base of the cloud.   This type of rain-making process is characteristic of tropical thunderstorms, and is highly efficient.  Tropical “thunderstorms”, including those in hurricanes, are well known for their deluges, but seldom produce lightning.

Without CG lightning on the NLDN display, the Madison County storm cell appeared quite benign.   Meteorologists had to rely on radar indicators, rain gauge and spotter reports to warn for this highly destructive flash flood.

The bottom line:  Decoding lightning, or lack thereof

A meteorologist must use ALL tools available – including Doppler radar, NLDN, LMA, and satellite – to accurately judge a thunderstorm’s severity.   Lightning can be a useful indicator of problematic storm cells, but its absence does not imply a non-lethal convective storm.    Additionally, storms sometimes yield dangerously high CG flash rates.  These may be separate cells, or embedded in lines of storms.  Locating them and issuing timely “lightning warnings” remain a unique challenge that the LMAs are only starting to address.

Jeffrey Halverson is a Professor in the Department of Geography and Environmental Systems at the University of Maryland, Baltimore County.