Brian Matilla
@bxmatilla
Meteorologist (RA) | Python-smithy on some days? | short-term, high-impact weather w/ CAMs | M.S. (2018) at @miamirosenstiel.bsky.social | Husband | Latino | Racing aficionado. All opinions my own & not my employer’s.
Those are some possibly intense winds headed right for South Bend, IN with this bow echo… the eastern-most tornado warning in this group highlights hazards as “Tornadoes and wind gusts in excess of 100 mph.” Sheesh.
Just under 5 hours remaining until the abstract submission portal closes for the 107th AMS Annual Meeting (5:00 PM EDT). There is no deadline extension this year, so get your abstracts in before it is too late!
There is WoFS, a traditional physics-based NWP ensemble. There is WoFSCast, a deep-learning AI emulator of WoFS. And there is NowcastNet, a deep-learning model trained to emulate MRMS rainfall rates. Jorge Duarte’s poster at 3:00 PM today will look into each model’s respective forecast quality.
High-Shear, Low-CAPE (HSLC) events can be some of the trickiest to forecast, particularly like the tornadic QLCS that impacted parts of OK on 26 Feb 2023. Madeline Diedrichsen’s talk at 2:30 PM looks into whether or not downscaled WoFS forecasts combined with more frequent DA helps predictability.
The Texas Hill Country Flash Flood event from last year continues to be a prominent talking point. @drstratman.bsky.social looked into whether or not coupling WoFS forecasts with NSSL’s FLASH system could amplify lead times of potentially dangerous flash floods, so check out his talk at 2:30 PM.
We’re staying on the theme of the 2026 SFE here, and much work was done this year to get WoFSCast, an AI/ML emulator of the original VLAB-WoFS system, prepared for evaluation alongside the other WoFS systems in realtime. Check out Corey Potvin’s talk at 11:30 as he’ll dive into those results.
Speaking of the Spring Forecasting Experiment, many new and existing CAMs, both physics-based and AI-based such as NVIDIA’s StormScope, were also evaluated this year. Adam Clark’s talk at 10:45 AM gets into more of the details for what was learned.
Earlier this year in the Spring Forecasting Experiment, the experimental MPAS-WoFS (version 2) was formally evaluated alongside VLAB-WoFS and other CAMs. Nusrat Yussouf’s talk coming up at 8:45 AM highlights some of the preliminary results of the v2 system and paths forward.
If you've ever worked with MPAS and WRF data, you've probably noticed that the output data of the former is much larger. What if small changes in the resolution make minimal impacts to forecast quality yet drastically reduce data footprint? Lou Wicker's talk at 11:15 AM gets into this.
The combination of moving towards the MPAS dynamic core and utilizing the new JEDI data assimilation system pushes WoFS closer to the NOAA UFS framework. @ksquared812.bsky.social's talk at 11:00 AM dives into some of the early results utilizing both of these frameworks.
One may be used to verifying WoFS (or any CAM) forecasts based on radar observations such as products from MRMS or even local storm reports. Chris Kerr's poster examines WoFS object-based performance against NWS Warnings (such as tornado and severe thunderstorm warnings).
In recent years, WoFS has been equipped with products to give forecasters and end-users the ability to predict fire weather behavior and environments. Sarah King's poster examines how incorporating high-resolution NGFS data with WoFS cases from 2020-2023 may help to improve upon fire wx prediction.
The push to move towards the MPAS dynamical core continues to gain momentum. Development at the Lab has begun to explore what a variable-resolution mesh for WoFS would look like and perform at the 1-km resolution. @drstratman.bsky.social's poster will shed some light into this.
The 14 March 2025 Carter County, MO tornado event has become the most prolific success story for what WoFS was built to accomplish. Miranda Silcott's talk at 5:30 PM will jump into how Emergency Managers and NWS forecasters used the guidance to give 2+ hours of lead time on this destructive EF3 tor.
I've alluded many times to how WoFS Version 1 (WRF-based) is well on its way to operations in the near future. Patrick Burke's talk at 4:45 PM will dive into the extensive implementation and use of the cloud-based WoFS system in NWS operations as part of the R2O demonstration.
On August 18-19, 2024, substantial flash flooding impacted portions of Long Island, NY. @drstratman.bsky.social's poster at 3:00 PM will highlight the impacts of assimilating New York State Mesonet data into an experimental 1-km WoFS system to assess any performance improvements with the added data.
Pat Skinner will give his talk at 2:15 PM to discuss results from applying a Bounded Normal Rank Histogram Filter versus the Ensemble Adjustment Kalman Filter to verify performance in WoFS forecast analyses using the DART data assimilation system.
The Warn-on-Forecast System will be on display this afternoon (and all week) at the @ametsoc.org Madison Summit! Here's what to look for today if you're in Madison or listening along virtually:
The Fox Hollow, NC tornado occurred in an area where WoFS depicted no shaded contour but, crucially, non-zero CAPE. This prompted a change to a new scale with more emphasis on the tails. With the updated scale, we can better capture those marginal CAPE events as well as some of the higher-end events
One of the cool things about this collaboration was giving the ability of underlaying WoFS data to the SPC Mesoanalysts for MDs. An example is shown from 14 March 2024 where the MD contained contours of WoFS 90th percentile 2-5 km updraft helicity from the 20 UTC forecast initialization.