marsh

@mrshll
307 Followers
390 Following
115 Posts

high-tech skills, low-tech ambitions.
---
For work, I run https://upstream.tech, an org that forecasts the future of rivers and monitors natural and conserved spaces. I also started dynamical.org to make weather data more fun to work with.

For not work, I'm running, gardening, making music, and searching for the intersection of mysticism, art, and technology.

webhttps://mrshll.com
pronounshe/they
marsh3d ago

A 1993 paper treats Homer's Odyssey as a meteorological field report. The storms "check out." Dactylic hexameter: history's most overlooked data format.

New Weathering episode: https://dynamical.org/podcast/007/

marshApr 1

We did it, chat. ECMWF AIFS Single forecast is now on dynamical.org

https://dynamical.org/catalog/ecmwf-aifs-single-forecast/

Spatial domain: Global
Spatial resolution: 0.25 degrees (~20km)
Time domain: Forecasts initialized 2024-04-01 00:00:00 UTC to Present
Time resolution: Forecasts initialized every 6 hours
Forecast domain: Forecast lead time 0-360 hours (0-15 days) ahead
Forecast resolution: 6 hourly

Thank you to European Centre for Medium-Range Weather Forecasts - ECMWF, AWS Open Data, and Source Cooperative ❤️

marshMar 25

📡 NOAA's Multi-Radar/Multi-Sensor System (MRMS) is now on @dynamical_org !

Radar-based precipitation analysis for the entire continental US, 1km resolution, every hour, back to 2014.

Song: Flashes From Everywhere · Stereolab · Instant Holograms On Metal Film

marshMar 23

obsessed with this little thing I added to the @dynamical_org website. Pulls latest GFS analysis - landmass colored by tmp, shows 10m and 100m wind and cloud cover. Globe is lit by a time-approximating sun angle.

Heavily inspired by Lucas Pope's work on Return of the Obra Dinn. Ditherpunks unite.

marshFeb 17

update to https://dynamical.org/scorecard

- added ECMWF IFS ENS and NOAA HRRR
- utilizes our asos-parquet for observations
- calculates lots of different metrics for temperature/precip. statistics are read on-demand from a statistics.parquet

it's still just for fun and possibly wrong (lol). figuring out the "right" metric for precipitation is hard! especially when forecast product spatial resolutions differ (don't want to reward smooth "safe" forecasts)

Show threadmarshFeb 11

Thankfully, ISU's Iowa Environmental Mesonet has a great ASOS archive. It's not the same as ISD but it's filling in.

I've created a big experimental, auto-updating GeoParquet version that you can test out here: https://dynamical.org/catalog/asos-parquet/

marshFeb 11

Late last year, NOAA deprecated ISD -- which I used for all of my analyses needing up-to-date weather observations.

I asked the community for alternatives and ... it was unsurprisingly fragmented. I sent paragraphs of feedback to GHCNh, NOAA's intended successor, but it was clearly created for a different use case.

marshFeb 4

weathering podcast 006 is out: zero-shot forecasting

https://dynamical.org/podcast/006/

dynamical.org

Advancing humanity's ability to access, understand, and act on accurate weather and climate data.

dynamical.org
marshFeb 2

We were without power for 5 days in Nashville due to an ice storm. Was one of the stranger experiences in recent memory. We are back now, my parents only got power back yesterday. I jotted down some in situ thoughts

https://mrshll.com/2026_01_27.html

marshDec 18
Cumberland River Forecast - Chronos-2 vs TiRex