A mid-level atmospheric wave with subtle warm-air advection forced some strong thunderstorms in central Texas this morning. The operational high-resolution convection-allowing NWP models did not handle these storms well at all.

ProbSevere LightningCast, an image-based AI model, picked up on the rapidly growing convection about 15-20 minutes before lightning initiation. LightningCast predicts the probability of lightning in the next 60 minutes at any location using GOES-R ABI reflectance and brightness temperature data.

ProbSevere v3 uses machine-learning models to predict next-hour probabilities of severe weather (large hail, damaging wind gusts, tornadoes), by incorporating storm-object tracking and extracting features from radar, satellite, lightning, and short-term NWP data. The animation below shows how the probabilities of any severe weather evolved for these storms as they approached the Dallas-Fort-Worth metro region. Hail up to 1.5″ in diameter was reported.

At 12:44 UTC, ProbSevere v3 (PSv3) probability of severe was 53%, vs. 23% for the operational ProbSevere v2 (PSv2), as seen in Figure 3. A post-mortem analysis revealed that, compared to PSv2, the PSv3 model was able to combine the sub-severe MESH (maximum expected size of hail), ENI flash lightning density, and moderate mid-level azimuthal shear in an environment with high effective shear (> 50 kt) to produce a stronger (and more accurate) probability of severe. PSv3 exceeded 50% four minutes before PSv2, and exceeded 40% eight minutes before PSv2 (see Figure 4). Having more accurate and timely probabilistic guidance prior to reported severe weather is critical in helping the NWS issue more accurate and timely severe weather warnings.

Both ProbSevere v3 and LightningCast will be evaluated by NWS forecasters at the 2023 Hazardous Weather Testbed, held in May and June in Norman, OK.

View only this post Read Less

JMA Himawari-9 Infrared Window (10.4 µm) images of Tropical Cyclone Judy in the South Pacific Ocean on 27 February 2023 (above) displayed some unusually cold cloud-top infrared brightness temperatures — with isolated vales of -100ºC and colder (red pixels embedded within interior darker black regions).

The Central Cold Cover feature was displaced to the NNW of the low-level storm center — and Judy was moving through an environment of low deep-layer wind shear (below).

===== 28 February Update =====

On the following day, 2.5-minute interval rapid scan Himawari-9 Infrared images (above) showed Judy as the storm was rapidly intensifying during its southward trek across the Vanuatu island chain. A ragged eye structure was intermittently observed.

A GMI Microwave (85 GHz) image at 1717 UTC (below) revealed that the eye of Judy was passing over the island of Epi, Vanuatu at that time.

===== 01 March Update =====

Judy then further intensified to a Category 3 storm at 0900 UTC on 01 March (SATCON) — however, not long after that time, 2.5-minute rapid scan JMA Himawari-9 Infrared images (above) showed that the ragged eye became obscured by high clouds from surrounding deep convection.

View only this post Read Less

What products are available in the Arctic in between SAR observations? Consider the composite image above, showing RADARSAT-2 SAR ice concentration at 0453 UTC on 22 February 2023 over/around Saint Matthew Island (direct link to footprint here at the STAR SAR Wind site*) and an RADARSAT Constellation Mission-3 (RCM3) image at at 0546 UTC on 22 February 2023 south of Russia (direct link to footprint here at the STAR SAR Wind site*). The background image is Level 2 AMSR-2 imagery with high spatial resolution: around 3-km pixel size!

The SAR orbits for 22 February that are shown below. There is a cluster of orbits over this part of the Gulf of Alaska from 0442 – 0546 UTC, but then few observations until after 1800 UTC. One could rely on Geostationary data to fill in that gap — but the Sun does not rise over this part of the Gulf of Alaska on 22 February until after 1800 UTC. This is when microwave data can be useful.

This image of GCOM-W orbits on 22 February 2023 (from this website) show a GCOM-W overpass over the Gulf of Alaska near 1430 UTC. GCOM-W AMSR-2 Ice Concentration product from that overpass are available at NASA Worldview (direct link) , as shown below. If this image information is included in the SAR image above (it is!), then ice forecasters can view ice edge information — albeit at resolutions that are grosser than SAR data provides. This Level 2 AMSR-2 imagery has 3-km resolution. This microwave resolution allows forecasters to track the sea ice edge in between SAR overpasses.

How do forecasters view this blending of information with different horizontal scales? Mike Lawson from the Sea Ice had this comment on the imagery at the top (and it also explains the colored arrows!)

.... In my opinion this is the best case scenario to use this imagery in there is a cloudy gap between SAR images. The red and yellow arrows are features that can clearly be tracked based on our previous analysis (yellow lines). The blue arrows highlight the boundary of the pack ice vs. marginal ice, which I think AMSR handles quite well in most cases.
Another situation I didn't get a good screenshot example of happened during this event. You can grab both the morning and evening images for the day then use them to estimate how fast the ice edge is advancing/retreating. In this case, it looks to be around 8 nm between images or an avg of 16nm of retreat per day. These estimates can really help us get a handle on what to forecast for certain wind regimes or what estimate to give customers when they call and ask for an update.

---

Thanks to Mike for sending along these images to Tom Greenwald, SSEC/CIMSS. Thanks also to NSIDC.

*Note: The retrieved winds at those links are invalid over ice, but the patterns in the retrieved winds will show where the ice sits.

View only this post Read Less

Sentinel-1A overflew the Hawai’ian islands twice, once on 26 February and again on 27 February. The animation above shows the AWIPS presentation of SAR winds at 1616 UTC on 26 February. Images from the NOAA/STAR website for those two footprints are shown below as toggles between the derived wind speed and the Normalized Radar Cross Section at 16:16:24 (top) and 16:16:55 (bottom), i.e., for about 1 second of imagery from Sentinel-1A! The orginal images can be found at the NOAA/STAR Sentinel-1A website link above for 26 February.

One feature stands out in the toggle from 16:16:55 directly above: the band of strong winds seemingly emanating from a point near 18.8^oN, 156.67^oW. That hourglass appearance in the derived winds indicates a small-scale error in the winds used at the start of the retrieval (as noted in Quick Guide for this product); note the feature is absent in the NRSC fields. The derived winds there are most likely not correct.

A zoomed-in view over the Alenuihaha channel between Hawai’i to the south and Maui to the north, below, shows the very strong winds characteristic of that body of water. Winds of 25-30 knots are widespread, with a small area of winds exceeding 40 knots just off the south coast of Maui! Winds are much weaker in the lee of the islands.

The slow animation below shows a region in the SAR winds where (highlighted) isolated pockets of strong winds are diagnosed. Those strong winds have the look of regions where ice contamination in the retrieval is possible. Note, however, that the GOES-13 Brightness Temperature is only just below 0^o C (in the -2^o to -4^oC range); Cloud-top phase shows only liquid cloud in that region. However, the Level 2 Cloud Top Temperatures shows values colder than -50^oC!! (Here’s an AWIPS-sampled example near the coldest cloud top with a large difference between the Band 13 Brightness Temperature and the Derived Cloud-Top temperature!). Consider that It’s possible that the ice features are not quite at the size range of a pixel.

Part of the zoomed-in feature of the downwind-from-Hawai’i NRCS image (available online here) has an anticyclonic swirl that is very reminiscent of a von Kármán vortex feature (such as here), but at much smaller scale!

---

Shortly before 0500 UTC on 27 February, Sentinel-1A again overflew the Hawai’ian island chain, but this time farther west. The toggle at bottom shows the wind distribution downwind of Kauai along with the GOES-18 Band 2 (Visible) imagery (that has been brightened considerably). A noticeable tail of relatively calm winds extends downwind of the island. At the time, based 0000 and 1200 UTC soundings from Lihue, below, from this site, a strong inversion was present at around 700 mb.

View only this post Read Less