NWS Pago Pago forecasters (and Friends of the Blog) Jane Allen and Brayden Barton sent along some video and pictures of a significant thunderstorm that rocked the American Samoan Island of Tutuila in the hours before sunrise on Wednesday, May 20th, 2026. First, here is a video showing the intensity of the rainfall. Keep an eye on the palm trees and note how the wind direction practically reverses itself about midway through at 23 seconds in.

Winds were significant enough to cause noticeable amounts of damage as can be seen by the pictures in the following photo gallery.

This storm was characterized by rapid upscaling as it moved southward. To gain a perspective on how this storm developed, let’s start by looking at the environment in those early morning hours. First up, here’s the seat surface temperature (SST) as captured by NESDIS’s blended geostationary/polar orbiter SST product as displayed in SSEC Real Earth. From this, we can see that the SSTs surrounding the Samoan islands are around 26 C, or close to 79 F. This is slightly warmer than the ambient air temperature at the time as recorded at the Pago Pago airport, which was at 78 F.

There was also a 1200 UTC radiosonde launched from the Pago Pago forecast office (as plotted by NOAA’s SPC). The sounding shows modest levels of instability for the tropics and a moderate amount of speed shear but little directional shear. The presence of some shear is critical for deep moist convection to develop in the tropics as otherwise a storm will rain in its updraft and kill it off shortly after initiating.

With that background in place, let’s check out the satellite perspective of this event. Since it’s night, the GOES-18 Band 13 (10.3 micron) infrared window channel is an excellent place to start. This first loop covers from 1300 to 1500 UTC (1:00 AM to 3:00 AM local time), covering the initiation and first growth of this storm. The brightness temperatures drop from -20 C to -70 C in a little over an hour, showing significant convective growth.

At the same time as this geostationary loop, the polar-orbiting NOAA-20 passed overhead which gives us the opportunity to see a NUCAPS sounding from Tutuila. There is remarkably good agreement in the mixed layer and most unstable CAPEs between this profile and the radiosonde an hour earlier, which helps us have some confidence in the values retrieved from the satellite. Of particular interest is the 1168 J/kg downward CAPE (DCAPE). This is a quite high value for this parameter, which quantifies the negative buoyancy of descending parcels in a downdraft. Higher values of DCAPE means downdraft parcels are bringing more kinetic energy to the surface, and when those parcels hit the ground it results in faster outflow winds.

The NUCAPS sounding shows a total precpitable water (TPW) value of 1.87 inches. Of course, there are other satellite-based measures of TPW. One of the most popular among forecasters is the CIMSS MIMIC-TPW2 product. Here’s a view of that product at 1300 UTC, where the Samoan islands appear to have a TPW around 2.0 to 2.2 inches. Regardless of the method you use, the TPW is quite high.

Given high SSTs, moderate amounts of instability, enough shear to prevent updraft contamination by rainfall, and a high tropopause, we can expect some deep convection. We can use the Community Satellite Processing Package (CSPP) Level 2 products as displayed on CSPP Geosphere to gain an understanding of just how tall these clouds are becoming. The following video shows the CSPP-derived cloud top heights from 1300 UTC to 1600 UTC (2:00 AM to 5:00 AM). As these clouds develop, they grow to a height exceeding 15000 m (that’s over 42,000 feet!). That’s consistent with the soundings which show a tropopause somewhere between 100 and 150 hPa.

One of the blessings of this storm is that it was pretty quick moving, which limited the total damage it could cause. Tutuila is only about 20 miles across at its maximum width, and from the direction the storms were coming from, it’s only seven miles from one side of the island to the other. Here’s a slowed down and zoomed in view of Band 13 from 1530 to 1730 UTC (4:30 AM to 6:30 AM local time)showing the storm as it traverses the island. Remember, American Samoa lies well outside of the CONUS scan for GOES-18, so in the absence of a mesoscale sector scan, the temporal resolution will only be 10 minutes. The deepest updrafts are over the island for only a few scans, so the worst of the storm only lasted 30 minutes or so. There may be some evidence of cell splitting given the separation of the overshooting tops, though in the absence of a radar to see beneath the anvils it’s hard to say for sure.

The CMORPH2 product is a satellite-based assessment of rain accumulation. SSEC’s Real Earth can display CMORPH precipitation estimates for a one hour, one day, or seven day accumulation period. Here’s the one day total satellite-estimated rainfall accumulation for 20 May 2026. Tutuila is somewhere between 50 and 75 mm for the day, which is basically 2-3 inches of rain.

This storm featured two characteristics that made it particularly impactful. The first was its high TPW value, which translated into significant rainfall on the ground. This could have been even worse if the storm hadn’t been so quickly moving to the southeast. The second is that it had a high DCAPE, which was responsible for strong wind gusts at the surface. Recall that video at the top of this post, with the rapid and direct shift in the wind direction? That’s a tell-tale sign of a microburst, which was likely given the strong downward instability. It’s fortunate that at this time of day there were no aircraft attempting to land at the Pago Pago International Airport, because a microburst like this is a very hazardous situation for aviation. Thanks, Jane and Brayden, for sharing your pictures and your stories!

View only this post Read Less

The GOES-16 satellite was temporarily brought out of storage for annual testing (details) — which allowed for a 3-GOES (GOES-18/GOES-West positioned at 137.0 W longitude, GOES-16 at 104.7 W longitude and GOES-19/GOES-East at 75.2 W longitude) Water Vapor image comparison of the signature of a leeside cold frontal gravity wave (reference) that propagated southward across the High Plains during the 18 May – 19 May 2026 period (above). The images are displayed in the native projection of each satellite.

In the immediate wake of the cold frontal passage there were northerly/northeasterly wind gusts of 40-42 kts in the Panhandles of Oklahoma and Texas, as well as in northeastern New Mexico — and the highest peak wind gust at post-frontal METAR sites was 44 kts at Tucumcari, New Mexico (KTCC), which occurred at 0537 UTC (below).

Surface observations showed that the cold front moved through Dodge City in southwest Kansas just before 2100 UTC on 18 May — and data from the Dodge City rawinsonde launch a few hours later at 0000 UTC on 19 May (below) depicted the shallow post-frontal cold air that extended from the surface to the 850 hPa pressure level, as well as dry air throughout much of the middle/upper troposphere.

A plot of the GOES-19 Band 9 weighting function — calculated using rawinsonde data from Dodge City at 0000 UTC on 19 May (below) showed that the peak of upwelling 6.9 µm radiation occurred at the 500 hPa pressure level, with no contribution from the surface. This helped to underscore the vertically-propagating nature of the leeside cold frontal gravity wave, which allowed its unambiguous mid-tropospheric signature to be evident in 6.9 µm Band 9 water vapor imagery.

View only this post Read Less

During a period when American Samoa had been under a Flood Watch, the Weather Service Office at Pago Pago requested that a GOES-18 (GOES-West) Mesoscale Domain Sector be positioned over the region — due to their lack of radar, more frequent satellite imagery can be a critical tool for monitoring the development of deep convection. 1-minute GOES-18 Infrared imagery (above) showed clusters of deep convection that moved across the main island of Tutuila during a 7-hour period on 18-19 May 2026. At the Pago Pago METAR site on Tutuila (NSTU), most of their calendar day 24-hour precipitation for 18 May (5.21 inches) occurred in a 3-hour period during the GOES-18 animation shown above — and several Flash Flood Warnings were issued amid a variety of Local Storm Reports of flooding.

After a small cluster of convection rapidly developed just west of Tutuila shortly before 0000 UTC on 19 May, an overshooting top exhibited an infrared brightness temperature of -80.4ºC at 0018 UTC (below). Convective clusters then produced a thunderstorm with heavy rain showers at Pago Pago during the 0100-0200 UTC time period (METAR observations), contributing to much of the 4.15″ of rainfall that occurred during the 3-hour period ending at 0300 UTC.

In a comparison of 1-minute GOES-18 Visible and Infrared images (below), the overshooting tops of rapidly-developing convective clusters that moved over and near Tutuila were very apparent. The violet pixel in the 0020 UTC infrared image highlighted an overshooting top brightness temperature of -81.0°C.

View only this post Read Less

1-minute Mesoscale Domain Sector GOES-19 (GOES-East) Visible images (with/without an overlay of the FDCA Fire Mask derived product) and Shortwave Infrared images (above) showed the thermal anomaly associated with a fast-moving grass fire (the Sharpe Fire) that moved northeastward across the Oklahoma border into Baca County in far southeastern Colorado on 17 May 2026 — prompting the issuance of an Evacuation Immediate order for southern Baca County residents in the vicinity of Campo at 2045 UTC. RAWS sites upstream of and near the fire recorded southwesterly wind gusts as high as 45-47 mph.

This grass fire burned very hot, and at 2017 UTC it first exhibited a 3.9 µm shortwave infrared brightness temperature of 137.77ºC (below) — which is the saturation temperature of GOES-19 ABI Band 7 detectors.

1-minute GOES-19 GeoColor RGB images with Next Generation Fire System (NGFS) Fire Detection polygons (below) also showed the rapid northeastward spread of the wind-driven fire’s thermal signature. The surface observation site near Campo, Colorado recorded relative humidity (green) values as low as 4%, and southwesterly wind gusts (red) as high as 45 mph. Note how the wind direction at area observation sites fluctuated from a more northerly-northeasterly direction early in the day to southwesterly later in the day, as a segment of a quasi-stationary frontal boundary briefly surged northward (surface analyses).

View only this post Read Less