This website works best with a newer web browser such as Chrome, Firefox, Safari or Microsoft Edge. Internet Explorer is not supported by this website.

LightningCast version 2

The ProbSevere LightningCast version 1 (v1) model uses machine learning and GOES-R Advanced Baseline Imager (ABI) data to predict the probability of lightning in the next 60 minutes. This version will be operational at NOAA later in 2025.As research and development continues, a new version of the model (LightningCast v2) adds... Read More

The ProbSevere LightningCast version 1 (v1) model uses machine learning and GOES-R Advanced Baseline Imager (ABI) data to predict the probability of lightning in the next 60 minutes. This version will be operational at NOAA later in 2025.

As research and development continues, a new version of the model (LightningCast v2) adds Multi-Radar Multi-Sensor (MRMS) Reflectivity at -10oC as a predictor. Reflectivity at -10oC is well correlated with imminent lightning activity due to its ability to depict hydrometeors in the mixed-phase region of convection (generally 0oC to -20oC). Other radar-derived parameters are being investigated as well.

Preliminary results demonstrate that the radar-derived predictor adds value to the problem of short-term lightning prediction, without diminishing the power of the satellite predictors—that is, the model appears to have learned and uses the strengths of each data source for making better predictions.

Here are two recent examples. First, in Ohio and Pennsylvania, warm air advection at 850 and 700 mb forced some elevated but shallow thunderstorms.

Figure 1: 850 mb (left) and 700 mb (right) analyses from the Storm Prediction Center’s mesoscale analysis page (from 16-17 UTC on 01/31/2025). The filled red areas are regions with warm air advection.

A short-term model profile from the NAM 3-km model also shows the elevated but abbreviated extent of CAPE.

Figure 2: A skew-T log-P chart showing the thermal and moisture profile from southwest Pennsylvania at 18Z on 01/31/2025. The profile is from www.pivotalweather.com and powered by SHARPpy.

The first animation below is from LightningCast v1. The shallow nature of the storms and the thick layer of ice clouds above the convection obscures key signatures for the satellite-only model, resulting in poor probabilistic guidance.

Figure 3: Animation of LightningCast v1 (contours), GOES-16 day-cloud-phase-distinction RGB (background), and GLM flash-extent density (foreground).

The animation below uses the LightningCast v2 model. While not a cure-all, the reflectivity at -10oC clearly helps the model provide better guidance to lightning, albeit with little lead time to lightning initiation in this case.

Figure 4: Animation of LightningCast 2 (contours), GOES-16 day-cloud-phase-distinction RGB (background), and GLM flash-extent density (foreground).

The day prior, weak 850 mb warm air advection forced convective development in northeast Colorado and southwest Nebraska. The thermal profile appeared to be too cold to generate much CAPE or lightning.

Figure 5: 850 mb analysis from the Storm Prediction Center’s mesoscale analysis page

However, from a satellite perspective, the convection certainly looks like it could produce lightning. Thus, the LightningCast v1 output shows high probabilities of lightning (animation below).

Figure 6: Animation of LightningCast v1 (contours), GOES-16 day-cloud-phase-distinction RGB (background), and GLM flash-extent density (foreground).

But the reflectivity at -10oC predictor generally has convective cores of only 25-30 dBZ. Typically 35-40 dBZ is needed for lightning at this isotherm.

Figure 7: MRMS reflectivity at -10oC, from the MRMS operational product viewer.

The radar predictor helped reduce the false alarm predictions of lightning markedly. See the animation of LightningCast v2 predictions below.

Figure 8: Animation of LightningCast 2 (contours), GOES-16 day-cloud-phase-distinction RGB (background), and GLM flash-extent density (foreground).

Overall, we’ve found that LightningCast v2 improves the critical success index over the contiguous U.S. (CONUS), while not harming predictions outside of the CONUS. In the image below, red regions show improvement of v2 over v1, whereas blue regions show degradation of performance in v2, with respect to v1 (note that this is a limited sample).

Figure 9: The difference in performance between LightningCast v2 and LightningCast v1 (red is where v2 is better; blue is where v1 is better). Note that these results are preliminary.

An assessment of lead time to lightning initiation (LI) has shown that rather than diminishing lead time to LI, LightningCast v2 actually appears to increase lead time to LI over the CONUS by a small amount. Work is on-going to quantify how much ABI predictors alone increase lead time to LI ahead of a radar-only lightning nowcasting model. We hope to have forecasters evaluate LightningCast v2 at the 2025 Hazardous Weather Testbed.

View only this post Read Less

High Resolution Views of a Lake Michigan Winter

High spatial resolution satellites like the Landsat series operated by NASA and the US Geological Survey, or Sentinel operated by the European Space Agency, are often used for land process investigations. Their very high spatial resolution comes at the cost of less frequent overpasses over specific locations, and so they... Read More

High spatial resolution satellites like the Landsat series operated by NASA and the US Geological Survey, or Sentinel operated by the European Space Agency, are often used for land process investigations. Their very high spatial resolution comes at the cost of less frequent overpasses over specific locations, and so they aren’t used as regularly for meteorological purposes as the coarser-but-more-frequent observations from low-earth orbiting instruments like VIIRS or the multiple observations per hour that geostationary satellites produce. However, phenomena of interest to meteorologists can still be found in these satellites if they happen to fly over at the right time.

A month’s worth of observations from the Sentinel-2 satellite shows this very well. Depending on the channel, spatial resolution from Sentinel-2 can be as fine as 10 m which results in very highly detailed imagery of Earth’s surface and the clouds above it, but together the two Sentinel-2 satellites only pass over a location once every five days. Here, several images from the month of January 2025 are shown depicting Chicago and the southern tip of Lake Michigan. Early on in the month, the lake was ice-free as the water temperatures remained well above freezing. By mid-month, however, temperatures plunged into the single digits Fahrenheit overnight and ice began to form in the shallower regions near the lakeshore (for example, see this 15 January blog post).

Consistent offshore winds helped push newly-formed ice into the middle of the lake. At times, these winds can be measured from space using the Advanced Scatterometer (ASCAT) instrument deployed aboard EUMETSAT’s Metop series of polar orbiting satellites which derives wind speed and direction from the radar reflective properties of surface water waves. The Great Lakes are large enough that useful wind measurements can be obtained from them via satellite. For example, ASCAT winds from 20 January show strong westerly flow from the land and over the lake.

As the wind pushed newly-formed ice away from the shore, new ice could form in the same shallow regions. The very low temperatures in mid-to-late January(the high temperature in Chicago was only 2 F on 21 January) caused a significant amount of western Lake Michigan to freeze.

The Sentinel overflights also uncovered some very interesting industrial impacts on cloud formation. The below image is from 24 January 2024. Very faint plumes are visible in the true-color imagery stretching to the northeast, originating from downtown Chicago and the steel mills and oil refineries of northwest Indiana. These plumes track across Lake Michigan where they serve as cloud nucleation that enhances the development of the elevated convection seen off of the western edge of the state of Michigan. In essence, the output of these significant areas of heavy industry is seeding clouds that form dozens of kilometers downwind. The very-small scale of these plumes means they are difficult to identify when using almost any other satellite system.

A zoomed-out view, showing the size and scope of these enhanced cloud regions, comes from the more coarsely-resolved Sentinel-3 Ocean and Land Color Instrument (OLCI). In this case, it appears there is additional seeding from industry in southwestern Wisconsin.

While these instruments that are focused on land studies won’t have applications for every day in operational meteorology, these examples show that they are still useful for gaining a larger-picture view of the environment and uncovering interesting processes occurring at scales that might be too small to otherwise be detected.

Imagery source: Copernicus Browser, https://browser.dataspace.copernicus.eu/

View only this post Read Less

Satellite estimates of Heavy Rain over Hawaii

The potent cyclone (surface analysis, an airmass RGB animation from the College of Dupage website is below) that caused severe weather over Hawaii (blog post) is also drenching the islands with considerable rains. The animation of MIMIC Total Precipitable Water, above, shows the storm drawing tropical moisture northward from the Intertropical Convergence Zone. The airmass RGB,... Read More

MIMIC estimates of Total Precipitable Water, hourly from 0000 UTC 28 January through 1500 UTC 31 January 2025 (Click to enlarge)

The potent cyclone (surface analysis, an airmass RGB animation from the College of Dupage website is below) that caused severe weather over Hawaii (blog post) is also drenching the islands with considerable rains. The animation of MIMIC Total Precipitable Water, above, shows the storm drawing tropical moisture northward from the Intertropical Convergence Zone. The airmass RGB, below, also shows that connection to the tropics, with the green area in the RGB that is moving northward over Hawaii.

GOES-18 airmass RGB, 2310 UTC 30 January 2025 – 1530 UTC 31 January 2025 (Click to enlarge)

How have various estimates of precipitation quantified the rains that have fallen? Hourly CMORPH-2 Rain Estimates captured at this RealEarth site (enter ‘CMORPH’ into the Search box; this website has data through 31 December 2024), below, capture the progress of the heavy rain band across the islands, mostly west to east, but with a couple of seemingly backward steps.

CMORPH-2 esimates of hourly rainfall, 0400-1300 UTC on 31 January 2025 (Click to enlarge)

In contrast, GOES-18 GREMLIN estimates of MRMS radar reflectivity, shown below with GOES-18 mid-level water vapor infrared imagery, shows steady eastward progress to the heavy rainband. A band of lighter precipitation persists over Kauai however. The animation below suggests the backward steps in the CMORPH hourly precipitation shown above are artifacts that bear investigation. A discussion on how GREMLIN rain estimates are used over Pago Pago in the south Pacific is available here.

GOES-18 mid-level water vapor imagery (band 9, 6.95, left) and GOES-18 GREMLIN estimates of MRMS radar (right), every 10 minutes from 2230 UTC 30 January through 1500 UTC 31 January 2025 (Click to enlarge)

There is a direct broadcast antenna at Honolulu Community College (link) and microwave data from that site can be used to derive instantaneous rain rates (via MIRS algorithms). The animation below shows three snapshots between 0715 and 1215 UTC.

Rain rate estimates from MetopB AMSU (0715 UTC), NOAA-21 ATMS (1149 UTC) and Suomi NPP ATMS (1215 UTC) (Click to enlarge)

How do the instantaneous rain rates compare with GREMLIN MRMS estimates? Three comparisons are shown below. The figures are qualitatively similar.

GREMLIN estimates of MRMS rainfall, 0720 UTC, left and MIRS Rain Rate from MetopB AMSU/MHS, 0715 UTC on 31 January 2025 (Click to enlarge)
GREMLIN estimates of MRMS rainfall, 1150 UTC, left and MIRS Rain Rate from NOAA-21 ATMS, 1149 UTC on 31 January 2025 (Click to enlarge)
GREMLIN estimates of MRMS rainfall, 1220 UTC, left and MIRS Rain Rate from Suomi-NPP ATMS, 1215 UTC on 31 January 2025 (Click to enlarge)

When evaluating heavy rains, there are many products available to a forecaster. Use them all to increase confidence in the diagnostics.

View only this post Read Less

Severe thunderstorms across Hawai’i

With the increasing threat for strong to severe convection across the western portion of Hawai’i, a Severe Thunderstorm Watch was issued for an area centered on the island of Kauai at 1522 UTC on 30th January 2025 (above).5-minute CONUS Sector GOES-18 Clean Infrared Window (10.3 µm) images (above) showed elongated... Read More

GOES-18 Clean Window Infrared (10.3 µm) image at 1528 UTC, with a cursor sample of the Severe Thunderstorm Watch [click to enlarge]

With the increasing threat for strong to severe convection across the western portion of Hawai’i, a Severe Thunderstorm Watch was issued for an area centered on the island of Kauai at 1522 UTC on 30th January 2025 (above).

5-minute GOES-18 Clean Infrared Window (10.3 µm) images, from 1106 UTC on 30th January to 0901 UTC on 31st January [click to play MP4 animation]

5-minute CONUS Sector GOES-18 Clean Infrared Window (10.3 µm) images (above) showed elongated clusters of convection moving eastward and northeastward across Kauai and Oahu — with embedded thunderstorms that produced heavy rain and strong winds on those two islands (Local Storm Reports).

A corresponding animation of GOES-18 Infrared images with an overlay of GLM Flash Extent Density (below) displayed the abundant lightning activity associated with these thunderstorms.

5-minute GOES-18 Clean Infrared Window (10.3 µm) images with an overlay of GLM Flash Extent Density, from 1106 UTC on 30th January to 0901 UTC on 31st January [click to play MP4 animation]

This convection was occurring within the warm sector of a deepening Gale Force to Storm Force low pressure system (bottom left on these surface analyses) that was located NW of the island chain — and the strong SW flow in advance of this storm system (and its cold front) interacting with island terrain produced wind gusts as high as 120 mph on Maui at 0850 UTC.

The air mass within the warm sector was also unusually moist — in fact, a new daily record rainfall of 3.57 inches occurred at Honolulu. The Total Precipitable Water (PW) value of 1.98″ derived from Lihue, Kauai rawinsonde data at 1200 UTC on 30th January was a record high value for that date/time, according to the SPC Sounding Climatology site (below).

Plot of Lihue rawinsonde data at 1200 UTC on 30th January [click to enlarge]


Climatology of Total Precipitable Water for Lihue at 1200 UTC on 30th January [click to enlarge]

The PW value derived from Lihue rawinsonde data at 1800 UTC on 30th January (about 2.5 hours after the Severe Thunderstorm Watch was issued) was even higher, at 2.06″ (below).

Plot of Lihue rawinsonde data at 1800 UTC on 30th January [click to enlarge]

View only this post Read Less