Matt Onderlinde

Meteorologist/Programmer, PhD. Creating a web platform for weather data visualization.

[Matt Onderlinde @Weathernerds]

Added a new section to Weathernerds to host demo videos and limited case studies:
https://t.co/wns8sgTUJ5

The "Severe Storms" video discusses tomorrow's forecast:
https://t.co/sZhfHZQVJk

I'll build out the video page over the coming weeks.

Be safe tomorrow!
#Severe #wxtwitter

[Matt Onderlinde @Weathernerds]

Another impressive forecast verification from the GDM team.

[Matt Onderlinde @Weathernerds]

Pretty impressive short term spread in the ensembles for Sinlaku. Big implications for Guam. AI-based GDM-FNV3 much farther east… Not making a forecast; just making note so we can return to this afterwards.

[Matt Onderlinde @Weathernerds]

Pretty impressive short term spread in the ensembles for Sinlaku. Big implications for Guam. AI-based GDM-FNV3 much farther east… Not making a forecast; just making note so we can return to this afterwards.

[Matt Onderlinde @Weathernerds]

@daily_msu They certainly have made it look easy 😏

[Matt Onderlinde @Weathernerds]

@CC_StormWatch Jackson NWS forecast reasonably in line with the HRRR FRAM guidance 😬

[Matt Onderlinde @Weathernerds]

Indeed.
https://t.co/vudqJ0eFT9

[Aaron Rigsby @AaronRigsbyOSC]

Brutal driving conditions on the Indiana turnpike near South Bend #INwx

[Matt Onderlinde @Weathernerds]

I-94 (or 80/90) around the bottom of Lake Michigan will be a poor choice Sunday night...

[Matt Onderlinde @Weathernerds]

My heart goes out to Jamaica. May God protect them and get them through this. 🙏

[Matt Onderlinde @Weathernerds]

@JaydonShawwx the HFBI tracker jumps from Octave to the disturbance NHC currently marks as 60%, which also develops in the outer domain of the Octave run. Why it rockets all the way to cat5 in the HFBI output is still odd tho... 🤔

[Matt Onderlinde @Weathernerds]

The layer steering plots on Weathernerds show the evolution of the flow around Imelda nicely. If it hadn’t been for Humberto carving out a path through the upstream ridge, the flow likely would’ve stayed SSErly over Imelda and led to poorer outcomes for the US East Coast.

[Matt Onderlinde @Weathernerds]

It’s fixed. Thanks!

[Matt Onderlinde @Weathernerds]

@ECMWF I'm getting "Access denied: 530" for wmo:essential requests at ftp://diss.ecmwf.int/ ( or in the browser at https://t.co/MDrrrEHrzm )

Has the login info changed?

[Matt Onderlinde @Weathernerds]

@tropicalupdate 🙂

[Matt Onderlinde @Weathernerds]

@stormchaserjs @AndyHazelton A couple of points for clarity:

GenCast does a bias correction on the TC intensity forecasts.

FNV3 actually computes its TC intensity forecasts with AI in a separate step from the model fields. ie, it’s not just a tracker being run on the output fields.

[Matt Onderlinde @Weathernerds]

Compelling evidence of GDM’s utility continues to mount. Their ensembles recently added to Weathernerds:

https://t.co/sqDDVg5w9L

[Michael Lowry @MichaelRLowry]

Former NHC Branch Chief James Franklin shared these verification plots tonight from all publicly available models for Hurricane Erin. @GoogleDeepMind (GDMI) crushed it, even besting the consensus aids. We'll see if this holds the rest of the season, but color me impressed.

[Matt Onderlinde @Weathernerds]

@AndyHazelton Good question. I've started investigating that a time or two but never dove deep enough to figure it out.

[Matt Onderlinde @Weathernerds]

Alright, Google Deep Mind tracks are live. Ingest code is still pretty raw so bear with me if there are issues. Also, I switched to an "Ensembles" drop menu now. Thanks to the GDM team for their phenomenal work!!!

[Matt Onderlinde @Weathernerds]

@AndyHazelton FNV3 is their latest experimental version that includes the intensity "correction"

[Matt Onderlinde @Weathernerds]

@tom_r_andersson @FerranAlet My Pleasure!!! The work you guys are doing is absolutely phenomenal! Excited to see where it goes from here...

[Matt Onderlinde @Weathernerds]

Coming soon (probably tomorrow) to Weathernerds: Google Deep Mind AI track forecasts (FNV3 & GenCast). These are proving to be pretty useful for genesis, track, and even intensity forecasting.

[Matt Onderlinde @Weathernerds]

@stormchaserjs Not yet 🙂 Once they start providing the raw tracker output it will be added. I don’t have the compute resources to run the tracker myself on that ensemble set 🙁