LightningCast in Tampa, Florida

April 8th, 2022 |

Thunderstorms were slowly but surely edging their way dangerously close to Raymond James Stadium in Tampa, FL, on April 2nd. The New York Yankees and Atlanta Braves had just completed a spring training game at the stadium, when two people were struck by lightning in one of the parking lots surrounding the stadium (they were hospitalized but reported to be in stable condition).

ProbSevere LightningCast is an experimental deep-learning model that is running in near-real time at CIMSS. It uses images of GOES-R Advanced Baseline Imager (ABI) visible, near-IR, and longwave-IR channels to predict the probability of lightning (as observed by the GOES-R Geostationary Lightning Mapper [GLM]) in the next 60 minutes.

Below is a time series of the LightningCast probability and GLM-observed lightning at and near Raymond James Stadium (left panel), along with an animation of LightningCast probability contours, GOES-16 0.64-µm reflectance (from a 1-minute mesoscale sector), and GLM flash-extent density (right panel) near the stadium (red circle). In this way, users can see how the model’s probabilities evolved over time at a specific location and within the vicinity.

Figure 1: Time series of LightingCast probability and GLM-observed lightning at Raymond James Stadium (left). Animation of LightningCast contours, GOES-16 visible reflectance, and GLM flash-extent density (right) near the stadium (red circle).

Police officers responded to the two individuals struck by lightning at 3:45 PM local time (19:45 UTC). Based on the footprint of the GLM flash-extent density, they were struck at approximately 3:32 PM. The LightningCast probability of lightning was 75% 30 minutes before the lightning strike (remaining mostly above 50% between 3:00 and 3:32 PM). The probability of lightning first reached 50% about 1 hour before the lightning strike, and lightning started occurring within the vicinity (within 25 km) about 45 minutes before the strike.

Output from LightningCast, which leverages the high spatial, temporal, and spectral information found in GOES-R ABI, can help objectively quantify the short-term threat of convective hazards such as lightning. The model could perhaps be used by forecasters to advise outdoor venues such as stadiums to take mitigating actions sooner, or by individuals to help make safe decisions.

Figure 2: Annotated time series of LightningCast probability of lightning and GLM observations near Raymond James Stadium.

Forecasting lightning

July 15th, 2021 |

Lightning safety is important for aircraft, mariners, and many outdoor activities. CIMSS is working to evaluate a model that nowcasts lightning. This model was trained using GOES-16 ABI visible, near-infrared, and long-wave infrared channels, as well as GOES-16 Geostationary Lightning Mapper (GLM) observations. It predicts the probability of lightning (IC or CG, as observed by GLM) in the next 60 minutes at any given point. The model routinely provides lead-time to lightning initiation of 20 minutes or more. We’re hopeful that one day such a model will help forecasters provide guidance for aviators, mariners, and decision support services (DSS) for things like sporting events, festivals, and theme parks. Near-real-time model output can be viewed using SSEC’s RealEarth.

Below are a few examples, with the forecast lightning probability contoured over the daytime cloud phase RGB and GOES-16 GLM flash-extent density.

So this summer, whether you’re going to the South Carolina beach,

or sailing in the Gulf of Maine,

or hiking in the Rocky Mountains,

or catching the first MLB game in Iowa,

be on the lookout for lightning!

Using GOES ABI and deep learning to nowcast lightning

September 2nd, 2020 |

NOAA and CIMSS are developing a product that uses a deep-learning model to recognize complex patterns in weather satellite imagery to predict the probability of lightning in the short term. Deep learning is a branch of machine learning based on artificial neural networks, which have the ability to automatically learn targeted features in the data by approximating how humans learn.

A convolutional neural network (CNN) was trained on over 23,000 images of GOES-16 Advanced Baseline Imager (ABI) data to predict the probability of lightning, within any given ABI pixel, in the next 60 minutes. The CNN was trained using 118 days of data collected between May and August of 2018. Images of GOES Geostationary Lightning Mapper (GLM) flash-extent density (created with glmtools) were used as the source of lightning observations. Note that GLM is an optical sensor that observes both in-cloud and cloud-to-ground lightning.

The CNN currently uses four ABI channels: band 2 (0.64-µm), band 5 (1.6-µm), band 13 (10.3-µm), and band 15 (12.3-µm). Bands 2 and 5 are only utilized under sunlit conditions. Utilization of additional channels and time sequences of images is under investigation. The model uses a semantic image segmentation architecture to assign the probability of lighting in the next 60 minutes to each pixel in the image. The model is very computationally efficient, only needing 30 seconds to process the ABI CONUS domain and 3 seconds to process an ABI mesoscale domain using multithreading on a 40-CPU linux server.

Currently, the model only utilizes satellite radiances. Thus, it can be applied to nearly any spatial domain covered by the ABI or an ABI-like sensor (e.g. AHI). Based on near-real-time testing, the model routinely nowcasts lightning initiation with 10-30 minutes of lead-time. We expect the skill and lead-time will increase as new predictors (e.g. more ABI fields, NWP, radar where available) are added to the model.

Below are a sampling of recent examples. The base images are 0.64-µm reflectance, with GLM-derived flash-extent density overlaid as filled semi-transparent polygons. The flash-extent density is the accumulated number of flashes within the previous 5 minutes. The CNN-derived probabilities are displayed as contours at select levels (near-real-time output is available through RealEarth).

The overall objective is to improve lightning nowcasts in support of aviation, mariners, and outdoor events/activities. Beyond improving the CNN, our work will focus on packaging the output into actionable information for forecasters and other decision makers.

A cold front in Iowa

 

Thunderstorm development on sea-breeze boundaries in Florida and the Bahamas

 

Diurnally and orographically forced storms in the Southwest U.S. and Rocky Mountains

 

Storms in central Oklahoma, on the edge of Hurricane Laura’s cloud shield

 

A couple of examples over the Northeast U.S.

 

A boundary of convection on the southern bank of Lake Ontario

 

Storms in a warm sector in IL/IN/OH, perhaps along an outflow boundary

 

Southeast U.S. offshore region

The background image in this example transitions from 10.3-µm brightness temperature to 0.64-µm reflectance, while the flash-extent density enhancement also changes, in an attempt to enhance contrast.