Hurricane Lightning from Space

Plus tracking fire tornadoes via radar

Like an angry Zeus who found out the Fates had a party and he wasn’t invited, tropical cyclones can generate a lot of lightning.

Lightning rates in hurricanes can be associated with their strength. As cloud tops rise in a strengthening hurricane, they become colder. This leads to the formation of graupel, hail, and other ice crystals. Their collision in the clouds can create regions of opposite polarity, which leads to lightning. Thus, meteorologists monitor hurricane lightning to gather information about changes in the storm's strength and structure—two crucial elements in forecasting.

However, hurricanes are large, intense, dangerous, and unpredictable. Despite the allure, it's not feasible to casually sit on a porch chair with a notepad to tally lightning strikes. Moreover, most ground-based lightning tracking systems require stations fixed to the… well, ground - a rarity in open oceans.

One vantage point we can use to monitor tropical cyclone lightning is from outer space. The NOAA GOES satellites have a tool for doing just that - the Geostationary Lightning Mapper (GLM). You may remember the GLM from such WWATs as the longest-lightning-flash-ever-recorded.

The GLM captures optical light from Earth and detects changes in its brightness. Naturally, it operates more effectively at night, achieving about 70% efficiency due to the darker background, compared to about 40% during the day. Unlike many ground-based lightning detectors, the GLM can detect both cloud-to-ground and intra-cloud lightning. The GLM generates a tremendous amount of lightning data. For this research, they studied groups of lightning that lasted less than two milliseconds and were detected within a contiguous area of the pixel array.

The GLM dataset also includes numerous artifacts that may be mistakenly identified as lightning groups. These fakes typically stem from one of five reasons:

• The ‘Bahama Bar’ Effect: It is caused by differences in sensitivity among areas of the GLM’s sensor. The artifact appears over the Bahamas and is primarily associated with reflections and scattering of sunlight, particularly around sunrise and sunset.

• Blooming Events: When a flash is bright enough, it fills up a pixel, and the “extra” energy spills over into neighboring pixels.

• Spurious False Lighting: When the GLM hallucinates and observes lightning without clouds¹.

• Sun Glint: When sunlight reflects off the water and is misinterpreted as lightning.

Ground-Truthing the Bahama Bar

The GLM has a built-in algorithm, the Lightning Cluster Filter Algorithm (LCFA), to flag some of these non-lightning events in real time. The LCFA has continually improved over time, but some false lightning reports still make it through.

A research team led by Dr. Bejamin C. Trabing recently published a paper testing two mechanisms for supplementing the quality control of the LCFA. Using GLM archival data from 2017-2021, they developed and tested two approaches:

• Threshold approach: This looked at the amount of energy received by the pixels and their distribution. If the pixel pattern was too bright and too similar to each other, it could possibly be caused by blooming or the Bahama Bar.

• CNN approach: The other method used convolutional neural networks (CNN). This AI-based method was trained on GLM data, and then its predictions were compared with ground-based data from the Worldwide Lightning Location Network.

  Results

  The Threshold approach worked very well. When combining this method with the LCFA, they reduced false lightning reports by about 98% in the 2021 data.

  

  Left is a GLM lightning map of Hurricane Florence (2018). The Bahama Bar artifact is the purple horizontal line across the middle. The right is the same map after the new threshold algorithm has been applied. Poof! It’s gone. The outline is the hurricane as seen from satellite cloud observations. It was weakening at the time and thus had lost much of its symmetry, like one’s hair at the end of a long day. (Habring, et al. 2024; https://doi.org/10.1175/JTECH-D-23-0138.1)

  

  The annual increase in the effectiveness of filtering out false lightning data within 100 km of Atlantic tropical cyclones using the LCFA+ Threshold approach. (Habring, et al 2024; https://doi.org/10.1175/JTECH-D-23-0138.1)

  The CNN approach also worked (especially when the false lightning reports were extensive). However, there was no improvement over the Threshold approach. And since the latter is simpler and easier to interpret, it’s the preferred method for now (Occam’s Razor).

Occam says, “Maintain thine affairs with simplicity, thou fool.” (self-created (Moscarlop), CC BY-SA 3.0 via Wikimedia Commons)

The new techniques will be useful to operational meteorologists monitoring the development of tropical cyclones. However, they do not work as well for lightning over land, which can show up as more potent flashes to the GLM (and thus may be erroneously removed). However, when there are no other options, such as when a cyclone is over the open ocean, this new analysis method is a nice improvement.

And Now for Something Completely Different

Firestorms can cause flames to rotate in a way that resembles a vortex or a tornado. Often called fire whirls, they are not tornadoes in the strictest definition, but they do look exciting. On July 25, 2024, one was large enough to be picked up on weather radar during the Park Fire in California.

Above is a video of the firestorm I put together from RadarScope. The top frame is the default reflectivity image, which is likely showing smoke and debris. The bottom image is the Doppler radar velocity, showing the speed and direction of wind. In tight circulations like tornadoes, red and green colors will be side-by-side, reflecting the two sides of the circulation - one side coming towards the radar (green) and another going away from the radar (red).

In both frames, the firestorm is moving north by northeast. You can see the movement of the green patch in the bottom frame synchronized with the movement of the dark red/orange patch in the top frame. While it was not a tornado in the strictest definition, it was strong enough to play one on TV.

We acknowledge Dr. Milind Sharma for additional contributions.

ICYMI: If you like stories about lightning, check out our roundup of lightning-related presentations at this year’s meeting of the American Meteorological Society.

1. It is rumored this is caused by fantastic first-date kisses. More studies are needed.