In the evening of 28 July 2025, residents of the Samoan island of Upolu experienced an intense thunderstorm which produced hail, a very rare event for maritime tropical Pacific environments. Social media videos show 1 – 2 cm hailstones falling around residents excited to see such novel conditions. The forecasters at the National Weather Service office in neighboring Pago Pago, American Samoa (approximately 50 mi or 80 km away), requested and received a mesoscale sector from GOES-18, which easily covered both the US territory of Amercian Samoa as well as the independent nation of Samoa.

Since American Samoa doesn’t have a radar, the CIMSS ProbSevere algorithm is not well-suited for deployment in that location. However, the CIMSS LightningCast product was available, and it is seen superimposed on the 1 minute mesoscale Band 13 (10 micron data). LightningCast uses AI and machine learning techniques to quantify the probably of a particular storm producing lighting based on its satellite signatures, and here it is clear that this particular storm is capable of producing substantial lightning.

Zooming in on a single frame of the above animation, at 0653 UTC, shows the presence of a very cold brightness temperature at the center of the storm. This is indicative of an overshooting top and is a hallmark of deep, moist convection. Given the storm’s location in the tropics (around 14 degrees south) the troposphere is naturally quite deep and thus the overshooting top indicates the tip of an updraft that penetrates through a very thick layer of the atmosphere.

In fact, the 0000 UTC sounding from Pago Pago, AS, shows a tropopause height well over 12 kilometers above the ground, which is a reasonable estimate of the height of the anvil of the cumulonimbus cloud. The overshooting top is soaring even higher above the surface, and thus it is clear that this storm has substantial vertical lift. A classic rule of thumb in the midlatitudes for enhanced hail growth is steep lapse rates in the -10 C to -30 C layer of the atmosphere as this is where hail growth is at its greatest. (Warmer than -10 means water is less likely to freeze on contact, colder than -30 means supercooled droplets are rare). This event exhibits strong instability in the hail growth zone with lapse rates that are much steeper than the moist adiabatic lapse rate (blue lines) between those temperatures.

With no radar at Pago Pago, satellite-based estiamtes of precipitation intensity are required. This animation shows a loop of the CIMSS GREMLIN product, which aims to simulate radar reflectivity from satellite observations. GREMLIN was trained using GOES ABI and GLM inputs over the continental United States, and while its accuracy may be highest there, it can still provide useful information in other parts of the GOES domains about the intensity of a storm and its evolution.

The Night Microphysics RGB also can provide some insight. Here, the presence of large areas of red cloud help verify that we are looking at thick ice anvils associated with deep convection, and really helps separate the deep convective clouds from the low level ones. This also makes the vertical shear of the wind profile easy to recognize, with the low level clouds moving to the west with eastward moving anvils superimposed on top of them.

Hail is rare in the tropics for a variety of reasons. The first is that it is quite warm. In the above skew-T, the freezing level is approximately 580 mb. While deep, vigorous convection may be producing hail storms more frequently, they melt by the time they reach the surface. A second factor is that humidity and precipitable water values are high. This inhibits evaporative cooling by the rain (helping to lower the temperature below the cloud and giving the hailstones a greater opportunity to reach the surface). This storm appears to buck that trend with a strong updraft and particularly strong buoyancy in the layer of the atmosphere where hail is most likely to form. Whatever the reason, the people of Samoa will remember this event for a long time to come.

View only this post Read Less

5-minute CONUS Sector GOES-19 (GOES-East) Nighttime Microphysics RGB images from the CSPP GeoSphere site (above) depicted a nocturnal arc of low-level cloud bands (pale shades of white) associated with an undular bore produced by the outflow boundary from a decaying Mesoscale Convective System (MCS) over northern Iowa on 28 July 2025 — however, the boundary layer cloud bands of this undular bore began to dissipate around or shortly after sunrise, as seen in the subsequent daytime True Color images.

GOES-19 Mid-level Water Vapor images (below) showed that the MCS outflow boundary produced a few wind gusts >50 kts (red) in Nebraska and Iowa, including a gust to 62 kts at Mason City IA — and this outflow boundary / undular bore also acted as the forcing mechanism for a vertically-propagating gravity wave that continued to move south and southwest for about 12 hours. The notable shift of surface wind direction as the outflow boundary passed began to diminish after about 1401-1501 UTC.

The vertically-propagating nature of the gravity wave was evident due to its signature in both GOES-19 Mid-level Water Vapor (6.9 µm) as well as Upper-level Water Vapor (6.2 µm) images (below). This feature was somewhat analogous to leeside cold frontal gravity waves that have been previously documented on this blog.

The train of vertically-propagating gravity waves was not as apparent farther to the south, in more moisture-rich areas like much of Kansas and Missouri — but due to this moisture, an arc of mid-level clouds was forming along and just behind its leading edge (below). A Pilot Report noted Light to Moderate Turbulence (TB LGT-MOD) at altitudes of 11-15 kft during its eastward climb from the St. Louis airport at 1322 UTC; about 30 minutes earlier, the 1251 UTC METAR surface report from St. Louis KSTL included the mention of Altocumulus castellanus (ACC) from the northwest to the northeast (Altocumulus castellanus is a cloud type that indicates the presence of mid-level instability, which can cause light to moderate turbulence).

View only this post Read Less

As discussed on the CIMSS Satellite Blog on July 23rd, active tropical weather continues in the western Pacific Ocean. Specifically, the storm previously known as Invest 98W has organized into Tropical Storm Krosa, now located northwest of Guam and directly west of the Northern Mariana Islands, broadly moving north. Low-Earth Orbiting (LEO) satellite observations collected via direct broadcast by SSEC’s antenna system in Guam provides a valuable perspective on this intensifying storm in near-real time.

Above on the left is a sequence of JPSS VIIRS I-05 imagery centered over Guam between 15:30-16:30 UTC on Friday, July 25th. This is the “Legacy IR Window” band at 11.45 microns, very similar to the traditional IR imagery routinely available from geostationary satellites, albeit with higher spatial resolution. While the broad cloud shield of Krosa is easily visible, it is hard to diagnose what is going on in the lower levels of the storm. Since the time of these observations is during the local nighttime, traditional visible imagery is unavailable from both LEO and GEO platforms. However, VIIRS Adaptive Day Night band imagery (above, on the right) from the same passes, does a great job highlighting the low-level circulation of Krosa, displaced just to the north of the strongest convection.

Another valuable tool for monitoring tropical systems is the AMSR2 microwave imaging instrument aboard GCOM-W1. Above is the Level 1 36.5 GHz Horizontal Polarization brightness temperature observations over the region, on the left at 3:40 UTC on July 25th, and on the right at 15:53 UTC July 25th. This channel is particularly sensitive to the structure of precipitation in the lower portion of the atmosphere. As such, the tightening of the concentric features near the low level circulation is easy to spot over this approximately 12 hour period.

AMSR2 microwave data can also be processed to a number of Level 2 products utilizing the CSPP GAASP software package. Again, the image on the left is from 3:40 UTC July 25th, and on the right from 15:53 UTC July 25th. Comparing the two, there appears to be greater organization of the precipitation bands on the eastern and southern side of the storm in the later pass, with longer, more circular structure evident.

Another GAASP product, Ocean Wind Speed, provides an estimate of wind speeds over water. The time sequence matches the other products shown above. This product makes it clear that both the maximum strength and overall size of Tropical Storm Krosa’s wind field grew substantially in the roughly 12 hours between overpasses at 3:40 and 15:53 UTC. Interests in the region should continue to follow updates on the storm from the National Weather Service. All of the products featured in this post were collected and processed by SSEC’s direct broadcast antenna system in Guam (utilizing software such as CSPP Polar2Grid) and provided to the National Weather Service with low latency.

View only this post Read Less

1-minute Mesoscale Domain Sector GOES-19 (GOES-East) Visible images (above) showed the rapid development of thunderstorms that produced strong wind gusts across parts of northeast Illinois and northwest Indiana on 24 July 2025 — forcing ground stops for all scheduled inbound flights to Chicago’s Midway and O’Hare airports (beginning at 1945 UTC). The peak wind gust at Midway airport (KMDW) was 56 kts or 64 mph at 2002 UTC. In addition, GLM Flash Points indicated that abundant lightning activity was associated with these thunderstorms.

The corresponding GOES-19 Infrared imagery (below) displayed cloud-top infrared brightness temperatures as cold as -75ºC with some of the more robust thunderstorm overshooting tops.

According to a plot of rawinsonde data from Lincoln, Illinois (below) the coldest GOES-19 cloud-top infrared brightness temperatures of -75ºC represented a ~1 km overshoot of the Most Unstable (MU) air parcel’s Equilibrium Level (EL).

1-minute GOES-19 Visible and Infrared images (below) included plots of SPC Storm Reports — which showed a wind gust to 80 mph (W80) at 1946 UTC (caused by a thunderstorm microburst) in northeast Illinois, and a wind gust to 63 mph (W63) at 2034 UTC in far northwest Indiana.

View only this post Read Less