Mastodawn

RadicalEcologist Jan 22, 2025

Hey, so, we all know weather is going to get more dangerous over the next few years. Also, the USA weather services are probably not going to be operating at their best (to be optimistic). I know there are other sources, but the whole of the world society feels like it is on the edge of darkness. So...
Proposal/question: using the plethora of easily accessible data, compute, and brilliant people; how could we form a federated and decentralized weather service?

I imagine it would require some standardized protocols and modeling teams. Similar things do exist for other areas of research. However, I am an ecological modeler and know very little relevant information beyond using climate models. I am not completely sure what is needed for weather. Thoughts or ideas? Or is anyone even interested in this?

#wx #weather #federation #decentralized #opensource #foss #modeling #resilience #anarchism

Wulfy—Speaker to the machines

@RadicalEcologist

A back of the napkin model for distributed, independent #weather system.

1. Sensor net.
Using simple IoT Raspberry pi style "weather stations".
Or maybe even phones to provide secondary (temperature only) data.
Old school weather forecasting relied on a network of sensor remote weather stations read and telephoned in by ordinary folks.

2. Simplified, distributed computing forecasting models. Initially limited to short range (less computationally intensive) models.

3. $Weathercoin crypto. WAIT!
HOLD THE PITCHFORKS!
A Blockchain accounting for folks setting up resources to compute the models.
You "mine" the coin by using clock-cycles to compute the models.

AI6YR Ben Jan 22, 2025

@n_dimension @RadicalEcologist

The first one exists.
aprs.fi

Dragon-sided D Jan 22, 2025

@n_dimension @RadicalEcologist Any time you think you need a token, what you really want is a tip jar or to sell merch. Much more efficient, much more socially and morally just.

Wulfy—Speaker to the machines Jan 22, 2025

@dragonsidedd @RadicalEcologist

I like the idea of a "honesty jar".

It would be interesting to see some real numbers from #mastodon instance operators, not the $ but the % breakdown.
As that could be a good service analog.

I expect it would be something like 2% are regular supporter$, 5% are casual and the rest are freeloaders.

I am not sure if we can get many folks to pay for "open weather" data.

I was once trying to map data uplink performance degradation to rainfall pattern, but when I reached out to the Government weather office, they wanted a lot of money for that data, even though we were a not for profit. That killed that line of inquiry.

Wulfy—Speaker to the machines Jan 23, 2025

@RadicalEcologist

Ok... I'm using the other side of the Napkin. 1/2
Federated #weather #meterology system.

Numerical Weather Prediction (NWP)
is the least resource intensive model.
- Divides atmosphere into 3D grid cells
- Solves fluid dynamics equations for each cell

Resolution/Scale trade-off:
- Global models: ~9-13km grid spacing
- Regional models: ~1-4km grid spacing
- Each halving of grid spacing requires ~8x more computing power 😑

For continental USA you need about 1 station for 100km/sq or 53.51 hectopascals per imperial Troy tonne in freedom units.
For meaningful resolution.

You need to abandon satellite data (unless it's free). There may be possibility to gather maritime weather data if this takes off.

Wulfy—Speaker to the machines Jan 23, 2025

@RadicalEcologist

Ok... another napkin. 2/2
Federated #weather #meterology system.

Per Station Data Points:
- Temperature, humidity, pressure: 3 points
- Wind speed/direction: 2 points
- Precipitation: 1 point
- Cloud camera (low-res sky photos): ~1MB per image
- Sampling rate: Every 5 minutes
- Daily data per station: ~300MB (mostly images)
- Total daily data: ~30PB (sic!)

NWP Distributed Computing required;
- Grid resolution: 4km spacing (practical for regional)
- Vertical layers: 50
- Total grid points: ~6 million grid cells
- Computations per grid point: ~1000 operations
- Update frequency: Every 15 minutes

Distributed Computing Model:
- Comparable to Folding@Home which peaks at ~2.4 exaFLOPS
- For 24hr forecast:
* Each client needs ~4GB RAM
* ~1 hour computation time per client per grid section
* Minimum active clients needed: ~50,000
* Network bandwidth per client: ~1GB/day
* Central coordination server: ~100TB storage, 40Gbps network
(That's beefy...but not "let's drain the rivers beefy") I just priced the NAS the other day. And the storage would cost maybe $8000. Which is a pittance.

@n_dimension @RadicalEcologist Following up on Raspberry Pi IoT station, Weewx exists but I feel like it’s targeted at nerdy folks who know what they’re doing and have the listed supported weather stations.

But I do like your ideas 🤙🏾

RadicalEcologist Jan 23, 2025

@n_dimension I like everything you said except the crypto bit. Every attempt to use crypto for useful things, like distributed compute or networking, has failed pretty badly. To avoid the additional ethical concerns I think such a thing could work without crypto or any broadscale financing. For expensive parts of this Mutual Aid, standard donations, or merch (like I saw in another reply) are much better solutions.

Wulfy—Speaker to the machines Jan 23, 2025

@RadicalEcologist
I think all the crypto battles have been already fought so I don't want to rehash it, but to say EVERY use case for a token has failed us inaccurate.

VISA springs to mind, their settlements are being done in Etherium since 2021 and stable coins process billions DAILY just to mention a couple.

Not to mention pretty much EVERY (edit: "advanced") Government is still developing their own crypto (CBDC) which inarguably are totally evil.

RadicalEcologist Jan 23, 2025

@n_dimension Sorry, I meant useful thing other than act as an upgraded (somehow more dystopian) fiat currency or a tool in forex (things I am obviously biased against). Certainly, they can function in those spaces. I was mostly referring to attempts at using crypto to incentivize or be backed by work that has direct utility. For some of the more promising on paper examples; mesh networking, distributed computing, software development, etc... Those have all failed.
Don't get me wrong, I want people to be paid for their labours. I just don't see any place for crypto in that other than simple payments (i.e. donations). Of course there is the additional irony of using a climate change accelerant to reward people for trying to help survive a climate catastrophe.

Wulfy—Speaker to the machines Jan 23, 2025

@RadicalEcologist

This whole thing (waves arms awkwardly about) is a clown show.

Might as well juggle balls and make sure to paint a big smile on our faces as we awkwardly stumble in our oversized clown shoes over the precipice.

RadicalEcologist Jan 23, 2025

@n_dimension HA, perhaps! But, I would rather cross said precipice with community all equipped with a parachute and a helmet and, for me, what is left of my dignity.

Besides, I look trying or not in "this whole thing" like I look at *quitting*. Sure, it is always an option and all things do eventually come to an end, but you don't get to find out what else there is if you don't keep trying. Could get better or worse or, most likely, waves of both, so why not try now and you can always *quit* tomorrow.

Wulfy—Speaker to the machines Jan 23, 2025

@RadicalEcologist

Thank you so much.
Your kind words embody the spirit of community.
With folks like you around, some of us will make it.
Keep being you
/Hug

RadicalEcologist Jan 23, 2025

@n_dimension Wow! I don't believe there are greater compliments! Thank you kindly!!! I hope you'll stick around for whatever this might or might not turn into.
Be well!!

Starraven Feb 7, 2025

@RadicalEcologist @n_dimension

You don't need to use crypto for distributed computing. Use BOINC.