Forecasting storms with AI

March 22nd, 2022 |

Severe weather season is underway across the southern U.S. NOAA and CIMSS are using satellite, radar, and lightning observations of thunderstorms to develop and evaluate artificial intelligence (AI) tools that forecast and diagnose convection.The NOAA/CIMSS ProbSevere portfolio has several such tools to help forecasters keep tabs on storms.

LightningCast


ProbSevere LightningCast is a model that uses images of geostationary ABI data from GOES-16 or GOES-17 to predict where lightning will strike (as observed by the GLM) out to 60 minutes. We’ve found that the product frequently provides 15-30 minutes of lead-time to lightning initiation, measured from the 30-40% probability range (the most skillful range). LightningCast will be evaluated at the 2022 NOAA Hazardous Weather Testbed and at certain offices within the NWS. LightningCast may be able to one day aid forecasters in providing decision support services and general convective initiation situational awareness. Below is a movie of LightningCast output (contours) overlaid on the GOES-16 daytime cloud phase distinction RGB and GLM flash-extent density, for the developing severe storms in Texas on March 21st.


IntenseStormNet

Another image-based AI model within ProbSevere is called IntenseStormNet, which seeks to identify the most intense regions of thunderstorms. It uses images of ABI and GLM data as predictors to probabilistically diagnose storm intensity from a geostationary perspective. Its goal is to identify intense parts of storms as humans do: holistically; by picking up on the spatial and multispectral features that imagery captures, such as overshooting tops, cold-U/AACP, cloud-top texture patterns, and stark cloud edges. In a paper published in 2020, we found that high probabilities from IntenseStormNet are frequently correlated with severe weather reports. Below shows IntenseStormNet output (contours) for some of the storms over Texas on March 21st, most of which spawned hail, damaging winds, and several tornadoes. IntenseStormNet contours can also be tracked over time, providing a novel way to investigate convective properties of storms.


ProbSevere v3

IntenseStormNet output is also being used as a predictor within the experimental ProbSevere v3, which uses satellite, radar, lightning, and NWP data, and machine-learning (ML) models to forecast the probabilities of hail, severe winds, and tornadoes in the next 60 minutes. While ProbSevere v2 is operational at NOAA’s Centers for Environmental Prediction, ProbSevere v3 is being evaluated by NWS forecasters at the 2022 Hazardous Weather Testbed. An analysis of thousands of storms from 2021 showed that additional predictors and the more sophisticated ML models in v3 improve upon v2 performance. ProbSevere uses multi-sensor storm tracking and feature extraction to predict probabilities of severe weather across the U.S. Below are several animations of ProbSevere v3 output in Texas on March 21st.

ProbSevere v3 output (contours), MRMS composite reflectivity, and NWS severe weather warnings for storms in northern Texas on March 21st.
ProbSevere v3 output (contours), MRMS composite reflectivity, and NWS severe weather warnings for storms in central Texas on March 21st.

ProbSevere products over the Southern Plains

May 3rd, 2021 |

The NOAA/CIMSS ProbSevere portfolio contains AI models for nowcasting convective weather. I’ll use Monday’s severe weather over the Southern Plains to highlight several of them.

A strong cold front spawned numerous severe-hail, wind, and tornado producing storms over Texas and Oklahoma, aided by very large values of convective available potential energy (CAPE; > 4000 J/kg).  You can see numerous storm reports in Figure 1.

210503_rpts Reports Graphic

Storm Prediction Center’s preliminary severe storm reports for May 3rd, 2021.

Probsevere version 2 (PSv2) is an operational set of models at NOAA, which predict the probability of severe hail, severe wind, and tornadoes, in the next 60 minutes. The models are storm-centric, and the models’ domain is the entire contiguous United States (CONUS).  These models use MRMS (radar), GOES (satellite), short-term NWP, and terrestrial-based lightning observations to generate probabilistic guidance of severe hazards. Figure 2 shows output from an experimental version (PSv3), which includes additional MRMS, GOES, and NWP fields as predictors in a machine learning model.

Figure 2: ProbSevere v3 contours (colored, around storms), MRMS MergedReflectivity, and NWS severe weather warnings (yellow and red boxes) for storms over the Southern Plains. The second outer contour around some storms is colored by the probability of tornado.

 

Another ProbSevere product is a convolutional neural network that uses GOES-R ABI and GLM images to detect regions of intense convection, and is often correlated with strong overshooting tops, “bubbly-like” texture in visible imagery, strong lightning cores, and the cold-U/above-anvil cirrus plume signature. The intense convection probability (ICP) can be run on the 1-minute mesoscale scans as well as 5-minute CONUS sector scans aboard the GOES satellites. The ICP does not require radar data, and may also be able to operate on data from satellites with similar intruments (e.g., Meteosat Third Generation). ICP output is being used as a predictor in the experimental ProbSevere v3.

 

Predicting when and where lightning will occur is also important for many users, such as mariners, aviators, and outdoor event managers. The probability of lightning model (PLTG) is also a convolutional neural network, using images of visible, near-infrared, and longwave-infrared channels to nowcast lightning occurrence in the next 60 minutes. The purple-to-orange shaded regions in the video below show GLM flash-extent density (i.e., flashes passing through a location).

Monitoring severe weather as it happens

March 17th, 2021 |

NUCAPS/MADIS Lifted Index, GLM Group Density, GOES-16 Band 13 Infrared Imagery, and ProbSevere polygons, all at ~0939 UTC on 17 March 2021 (Click to enlarge) All imagery from RealEarth

When NOAA’s Storm Prediction Center issues a High Risk of severe weather (below), people sit up and take notice. Are there easily accessible tools to monitor the state of the atmosphere in/around a region of expected severe weather?

The toggle above shows products (early in the morning on 17 March — at 439 AM CDT) in RealEarth that can help. NOAA-Unique Combines Atmospheric Processing System (NUCAPS)/MADIS (Meteorological Assimilation Data Ingest System) Lifted Indices combine tropospheric information from NUCAPS profiles with lower-tropospheric/surface information from MADIS to create Lifted Index fields, twice daily. These fields are generated using HEAP (Hyper-spectral Enterprise Algorithm Package) software (incorporated into CSPP — the Community Software Processing Package) at the UW-CIMSS Direct Broadcast site. A Suomi-NPP (or NOAA-20) overpass will quickly yield stability information. Today’s afternoon Suomi-NPP overpasses occurs around 1730 UTC (east of the High Risk area) and 1915 UTC (Link, from this site.) The toggle above also includes GOES-16 Band 13 infrared (Clean Window, 10.3 µm) information, GLM Group Density, and NOAA/CIMSS ProbSevere (ProbSevere has a stand-alone RealEarth-based site here).  All of these products are useful in monitoring this evolving, dangerous event.   As is often the case, the strongest convection was occurring at 0939 UTC along the edges of the most unstable air, that is, in the instability gradient.

People within the region of elevated risk of Severe Weather on 17 March 2021, especially the region High Risk, should pay especial attention to the weather.

NOAA Storm Prediction Center Risk assessment for 17 March 2021, issued 1300 UTC on 17 March (Click to enlarge)


Added: the Geosphere site (link) gives rapid access to GOES-16 imagery (including mesoscale sectors) and can be used to monitor this evolving situation.


The afternoon image of stability is shown below.

NUCAPS/MADIS Lifted Index, GLM Group Density, and GOES-16 Band 13 Infrared Imagery, all at ~1830 UTC on 17 March 2021 (Click to enlarge) All imagery from RealEarth

Hail Storm in Daytona Beach

March 6th, 2021 |

GOES-16 Convection RGB over Florida, 1431 – 1701 UTC on 6 March 2021 (Click to animate)

Accumulating hail fell in Daytona Beach FL (Link) on 6 March 2021 in association with a front over the Florida peninsula.  Preliminary storm reports from SPC (link) show reports of 1″ to 1.75″ hail. (The region was under a general thunderstorm outlook from SPC: link). The animation above shows the Convection RGB from 1431 through 1701 UTC on 6 March, bracketing the hail event over Daytona Beach near 1600 UTC. A strong white/yellow signal develops in a cell over Volusia County (Dayton Beach is within Volusia County) around 1545 UTC. This is the cell that deposits the hail.

NOAA/CIMSS ProbSevere display, 1535 UTC on 6 March 2021 (click to enlarge)

ProbHail values for this event (from this website) were small, at less than 10%. The value of ProbSevere here could be in identifying the cell responsible for the Hail, and showing values for the radar object that exceed others nearby; that is, providing guidance as to which radar cell to interrogate most often. The image above shows ProbSevere at 1535, just before a Severe Thunderstorm Warning was issued. The image below shows ProbSevere at 1600 UTC, just after the Special Marine Warning was issued (and while the Severe Thunderstorm warning was still in effect).

NOAA/CIMSS ProbSevere readout, 1600 UTC on 6 March 2021 (click to enlarge)

The time series plot for the radar object that produced the hail is shown below. Note that ProbHail (and lightning) increased (marginally) before the hail events (reported between 1535 and 1615 UTC) before collapsing.

ProbSevere values associated with Storm Object 84638, which object produced hail over Daytona Beach, 1500-1800 UTC on 6 March 2021 (click to enlarge)

There are several features in the visible imagery, below, that might be affecting the thunderstorm producing the hail. An east-west boundary is moving down the Atlantic coastline, passing through Daytona Beach around 1551 UTC. A very strong reflective signal becomes apparent after 1541 UTC as well (link): the convective cell has penetrated through the cirrus shield in the region.

GOES-16 Band 2 Visible (0.64 µm) imagery, 1431 – 1701 UTC on 6 March 2021 (click to animate)

This was a challenging forecast in a marginal environment.