A UPS freight plane crashed on takeoff from UPS’s home base in Louisville, Kentucky on 4 November 2025. The plane was bound for Honolulu, Hawaii, and thus had a significant quantity of fuel on board. Eyewitness videos showed a large fireball as the aircraft hit the ground in an industrial zone around 5:15 local time (2215 UTC) and nearby residents were told to shelter in place and turn off external air handling systems in order to protect themselves from the smoke plume.

Several GOES products are designed to monitor fires, and thus can be applied to this situation as well. the following animation shows Band 7 (3.9 microns) from the GOES-19 Advanced Baseline Imager. This band is a key component of fire detection as it is very sensitive to high temperatures. Even if a small portion of a pixel is actively burning, that will create enough emission at 3.9 microns to be detected by Band 7. The following animation shows Band 7 from 2200-2300 UTC (5:00-6:00 PM local time). The airport is located in the central part of the image due south of the central city. Note the dark spot that rapidly appears starting at 22:20 UTC. This is the signature of the initial fire, and while the hot spot fades quickly, its presence is still detectable from space for some time after.

This is a tricky time of day to use many RGB products, as several of them rely on channels that are only available during the day or use channels whose interpretation change between day and night. Sunset took place just 24 minutes after the crash, and by this time the visible channels had lost their ability to show the surface. However, the fire temperature RGB can be used to detect fires both day and night. This RGB is a combination of the 1.6, 2.2, and 3.9 micron channels. Since these are shortwave channels, they are normally only used during the day as the earth’s natural emission at these wavelengths is negligible. However, fires are hot enough to produce radiant emission at these wavelengths at values that far exceed solar reflectance, and so this product can be used for fire detection 24 hours a day. The fire temperature RGB clearly shows the presence of a fire at the time the plane crashed. However, this signal does not last very long as the initial burst cools and clouds move in to obscure the surface.

_____________________________________________

5-minute GOES-19 Microphysics RGB images with overlays of NGFS Fire Detection polygons and surface reports, from 2201 UTC on 04 November to 0041 UTC on 05 November (courtesy Scott Bachmeier, CIMSS) [click to play MP4 animation]

5-minute CONUS Sector GOES-19 (GOES-East) Microphysics RGB images and Next Generation Fire System (NGFS) Fire Detection polygons (above) displayed a pronounced thermal signature of the UPS Flight 2976 crash-related fire beginning at 2216 UTC (cursor sample) — which became less distinct as it lingered until 0036 UTC. Plots of surface observations showed that the winds were light south-southeasterly at Louisville International Airport.

The south-southeast winds from the crash site fire transported some smoke over Louisville International Airport (KSDF), but this smoke remained aloft (with its base at altitudes of 2600-3200 ft) and the surface visibility remained at 10 statute miles (below).

METAR reports at Louisville International Airport (KSDF) with remarks of smoke (FU) sky coverage and altitude highlighted in red, from 2210 UTC on 04 November to 0105 UTC on 05 November  [click to enlarge]

A longer time series of KSDF surface report data (below) showed that the base of the low-altitude smoke descended from 3200 ft at 2300 UTC to 800 ft at 0800 UTC.

Time series of surface observation data at Louisville International Airport (KSDF), from 2200 UTC on 04 November to 1000 UTC on 05 November [click to enlarge]

In spite of a much larger viewing angle from the GOES-18 (GOES-West) satellite — 67 degrees, vs 43 degrees from GOES-19 — a thermal signature was also apparent in 10-minute Full Disk scan Microphysics RGB imagery, beginning at 2220 UTC (below). However, the GOES-18 thermal signature dissipated more quickly.

10-minute GOES-18 Microphysics RGB images with overlays of surface reports, from 2200 UTC on 04 November to 0040 UTC on 05 November (courtesy Scott Bachmeier, CIMSS) [click to play MP4 animation]

View only this post Read Less

5-minute CONUS Sector GOES-19 (GOES-East) Visible images (above) showed a rapidly intensifying Hurricane Force Low (surface analyses) that was approaching Atlantic Canada on 04 November 2025.

Two examples of Metop ASCAT surface scatterometer winds having speed values greater than Category 1 hurricane intensity are shown below.

5-minute GOES-19 Water Vapor images (below) depicted a mid-tropospheric filament of dry air (shades of yellow to orange) that wrapped into the circulation of the storm. Peak wind gusts at METAR sites along the southern coast of Newfoundland included 83 kts at Sagona Island.

A NUCAPS profile within the dry filament showed the depth of very dry air throughout the troposphere (below).

GOES-19 Water Vapor image at 1631 UTC on 04 November, with the location of a NUCAPS profile within the dry filament circled in black [click to enlarge]

NUCAPS profile of temperature (red) and dew point (green) at the black circled location within the dry filament in GOES-19 Water Vapor imagery [click to enlarge]

10-minute GOES-19 True Color RGB images (below) revealed an area of water that exhibited a hazy appearance — this was due to enhanced scattering of light off of highly-agitated seas with large waves and abundant sea spray (caused by a localized area of very strong winds).

At 1501 UTC (Visible image | True Color RGB image) this area of hazy-appearing water was located just SSE of Buoy 44139 — where the peak wind gust was 68 kts, the maximum wave height was 31 ft and upwelling (caused by significant wind/wave action) cooled the water temperature by 6.5 degrees F in 6 hours (below).

Plots of wind speed (blue), wind gusts (red) and air pressure (green) at Buoy 44139 on 04 November

Plot of wave height at Buoy 44139 on 04 November

Plot of water temperature at Buoy 44139 on 04 November

A Synthetic Aperture Radar (SAR) image (source) from RCM-1 (below) showed the strong winds (shades of red) around the periphery of the storm center as it was located just south of the Newfoundland coast at 2132 UTC. Note the light wind speeds (shades of cyan to blue) in the vicinity of the storm center.

RCM-1 Synthetic Aperture Radar image at 2132 UTC on 04 November [click to enlarge]

View only this post Read Less

Long-time residents of the Great Lakes region know the significant impact that the lakes can have on the local weather. One of the most familiar effects is the lake breeze. Because it takes a long time for the lakes to warm up after winter while the land heats up comparatively quickly, during the spring and early summer the lakes are much cooler than the surrounding land. This creates a density gradient which forces a baroclinic circulation, resulting in a cool breeze blowing from the lake to the land.

However, in the fall the opposite can happen. Just as it takes the lakes a long time to warm up in the spring, it takes a long time for them to cool down in the fall. At the same time, the land cools off quickly, especially at night.  This creates a similar gradient as the springtime lake breeze, but this time the gradient is reversed. As a result, an offshore flow of cool air flows toward the lake.

An example of this can be seen in the GOES-19 satellite imagery over Lake Michigan on the morning of October 30 2025. Overnight, the outer edges of a midlatitude cyclone centered over the southeast continental United States pushed moisture and cloudiness westward across Lake Michigan. Earlier in the evening, the eastward motion was able to propagate unabated cross the lake and into Wisconsin. However, as the night grew longer, the land breeze developed and served as a counterbalancing force, creating an edge to the cloudiness that persisted until sunrise.

The following animation shows this process in detail. It begins at 2130 UTC (3:30 PM local time) on the 29th and depicts both the GOES-19 Band 13 (10.3 micron) brightness temperature and the surface weather observations. Given the fact that this event transitions from day to night and back to day again, looking at a single-channel infrared loop allows us to use the same product for the entire period. A visible product is not usable at night, and many RGB products rely on shortwave channels and thus cannot be easily interpreted for both day and night scenes.

Note that at the beginning of the loop, the air temperature recorded by the buoy in the middle of the lake is not that different from the air temperatures from sites in western Wisconsin. At 0230 UTC (9:30 PM local time) the central lake buoy records an air temperature of 54 F with a northeasterly wind. Racine, Wisconsin, shows an air temperature of 52 F at the same time, with the same northeasterly wind. However, as the sun sets, the temperature difference grows as the night continues. By 0530 UTC (12:30 AM local time) the situation has significantly changed. The buoy is in a location with very little nocturnal cooling: it has drooped one degree to 51 F. However, Racine has cooled substantially and is now reporting a temperature of 45 F. Furthermore, the buoy wind is still northeasterly while the wind at Racine is now northwesterly. This flip in the wind direction indicates that the land breeze has formed.

At 10 UTC the temperature gradient has gotten even stronger: 50 F for the lake air and 40 F for the terrestrial air, and by 1100 UTC (6:00 AM local time) the lake breeze is visible on the satellite as an edge to the cloudiness over the lake running roughly parallel to the lake’s western shore. Cloud were forming over the lake as the relatively warm and moist lake surface was able to create convection in the cooler air lying higher up. However, the advancing land breeze stabilized the air and inhibited convection where it advanced. This boundary is clearly visible on the night microphysics RGB. (Note also the large regions of fog throughout many parts of Wisconsin. These are identifiable as the areas of light green. The valley fogs forming in the river valleys of western Wisconsin are particularly notable).

As the sun rises, however, the temperature gradient breaks down. The land near the shore is able to heat up much more quickly than the lake is. By 1400 UTC (9:00 AM local time) the lake is only 6 degrees warmer than the land stations. The larger-scale synoptic winds are able to overcome the land breeze and push the clouds ashore. This animation from CSPP Geosphere shows the night to day transition.

Note that the land breeze front is better maintained in the northern part of the lake. There are a couple of reasons for this. First, it is colder up there. Overnight, terrestrial air temperatures dropped to the mid 30s F while the lake air temperatures remained around 50 F. Second, the presence of fog near shore means that less solar energy is available to change the temperature of the land due to reflection and latent heating. Thus, the gradient persists longer and the lake breeze is still able to inhibit convection over the western part of Lake Michigan.

View only this post Read Less

1-minute GOES-19 Visible images (left) and Infrared images (right) with plots of 1-minute GOES-19 GLM Flash Points, from 1131-1800 UTC on 28 October [click to play MP4 animation]

1-minute Mesoscale Domain Sector GOES-19 (GOES-East) Visible and Infrared images (above) showed Category 5 Hurricane Melissa as it made landfall along the far southwest coast of Jamaica around 1700 UTC on 28 October 2025. Low-altitude mesovortices were evident within the eye — and GLM Flash Points revealed abundant lightning activity within the inner eyewall region.

1-minute GOES-19 Infrared images displayed using RealEarth (below) indicated that the center of the eye of Melissa passed over White House, Jamaica during landfall — with the right-front quadrant of the inner eyewall (where wind damage and storm surge is often highest) passing over Black River.

1-minute GOES-19 Infrared images displayed with Google Maps labels of cities and roads, from 1500-2100 UTC on 28 October [click to play MP4 animation]

Shortly after Melissa made landfall, the eye quickly became cloud-filled as the hurricane moved inland and interacted with topography across the western part of Jamaica (below).

1-minute GOES-19 Visible images (left) and Topography+Visible images (right) with plots of GLM Flash Points, from 1451-1900 UTC on 28 October [click to play MP4 animation]

A larger-scale view of GOES-19 CIMSS True Color RGB images created using Geo2Grid (below) indicated that after landfall, the eye of Melissa remained cloud-filled as the hurricane finished crossing the island and eventually emerged over water north of Jamaica around 2100 UTC. 

1-minute CIMSS True Color RGB images, from 1131-2159 UTC on 28 October (courtesy Dave Stettner, CIMSS) [click to play MP4 animation]

A larger-scale view of 1-minute GOES-19 Infrared images during the 5 hours leading up to landfall is shown below — which revealed a continual series of cloud-top gravity waves propagating radially outward from the eye of Melissa. The radius of hurricane-force winds appeared to be rather small, since surface wind speeds at the Kingston METAR site in eastern Jamaica were only in the 20-40 kt range during that time period.

1-minute GOES-19 Infrared images with plots of 1-minute GOES-19 GLM Flash Points, from 1401-1900 UTC on 28 October [click to play MP4 animation]

A GMI Microwave image (below) displayed the compact eye and eyewall of Melissa about 3 hours prior to landfall.

GMI Microwave image at 1406 UTC on 28 October

Just prior to landfall, Melissa was moving through an environment characterized by fairly low values of deep-layer wind shear (below).

GOES-19 Infrared images, with an overlay of contours and streamlines of deep-layer wind shear at 1600 UTC on 28 October

A Synthetic Aperture Radar (SAR) image (source) at 1110 UTC (below) retrieved a maximum wind velocity value of 143 kts in the NW quadrant of Melissa.

RADARSAT-2 Synthetic Aperture Radar (SAR) image at 1110 UTC on 28 October [click to enlarge]

It is noteworthy that the Advanced Dvorak Technique (ADT) satellite-derived intensity estimate peaked at 185 kts for Melissa (below) — which according to the developer at CIMSS is the highest value on record for any tropical cyclone since the ADT was implemented.

Plot of Advanced Dvorak Technique (ADT) values for Melissa [click to enlarge]

Dropsonde released into Hurricane Melissa at 1350 UTC on 28 October [click to enlarge]

View only this post Read Less