A cool job for the climate-data-finance nerds

Looks like Apache Sedona has been democratised for smaller datasets and local analyses. Looking forward to making my #rstats code go brrrr 🔥 sedona.apache.org/latest/blog/...

When the world's biggest insurer says climate change has pushed traditional home insurance to its limits, it's already too late to start preparing. And we are in no way ready for a post-insurance world. Gift link to my column for @opinion.bloomberg.com www.bloomberg.com/opinion/arti...

Oh my word. There is a r package for everything.

At the Minnesota state fair today, what’s this years must eat?

I am in Milwaukee for a wedding- no one told me it’s giving Williamsburg Brooklyn pre-2010

I will unfortunately not be able to make this years @climateproof.news happy hour due to pre-existing commitments but you should if you are a #climaterisk professional! luma.com/p1w9klj2

Very cool paper! Happy to see stuff like this make it to the academic journals

New York climate week is on the horizon. Does anyone have any good #climaterisk sessions they can recommend?

This chat with @michaelwara.bsky.social was really great, very informative

I’m a big fan of #rstats but I am jealous of all the tools that are built python first.

I seem to be unable to open csvs or parquets using positron directly from file explorer with the “open with” function. Anyone else having this issue #rstats

#rstats has anyone added a new provider to ellmer? I see that it’s doable but I’m hoping to see if someone has blazed that trail for me and I can follow in their footsteps steps.

The “orange line is a climate risk scenario, indicating a recession which would be roughly twice as bad as the euro crisis, in line with the peak-to-trough in 2008-9 and only exceeded by the COVID-19 shock”. #climaterisk www.ft.com/content/79fc...

That’s enough posts on how to speed up r for now. More nat cat musings to come!

Another #rstats gis hack: I’ve been trying to take a 2.5 mil spatial points, buffer them, and then union the buffers into one shape. Turns out if you just order those points by lat, lon you can cut down run time required by st_union() from 6.5 hours to just 5 minutes!! 🏎️🏎️

For the #rstats peeps: are there any packages/paradigms for multi agent systems? Python has things like langchain, do we have anything similar?

It took ~1800 secs to convert my sf to raster using the raster and terra packages in #rstats. It only took 3 secs using fasterise! 🔥🔥 #gischat

Updated all-time costliest National Flood Insurance Program (NFIP) events on record. Note: Helene now sits at #4 ($6.18B), though nearly all of those payouts have come in Florida ($5.8B). North Carolina only at ~$0.2B. Most Americans living inland from the coast have no flood insurance coverage.

Spent 20 minutes cutting down the run time of a line of #rstats code from 5 secs to 1.6 secs. This was satisfying, but honestly a really bad use of my time

Not a function that I need to use very often but a good’un have in your back pocket. Turns delimited values in a cell into rows! #rstats tidyr.tidyverse.org/reference/se...

I’m going to pack up my catastrophe models and go home. Why didn’t I think of this? Problem solved.

Near surface smoke forecast for the CONUS area. Model run: 20250531T13:00 UTC Data from NOAA's HRRR CONUS (https://rapidrefresh.noaa.gov/hrrr/) Note that this product comes with absolutely no guarantee of availability, accuracy or completeness. #Wildfires #Smoke #Manitoba #AirQuality

Very cool way to give context to your LLM

In other #rstats musings, I thought I had cracked the data.table syntax but I am now back to the belief that it is a form of punishment. Only send me tidy code .😤

Used o3 mini to write some #rstats code to turn data into a raster and then spatial average the raster. This is something I’ve never done before and genai made it a lot easier. LLMs aren’t perfect but if you know a bit and can prompt well, it’s really good at filling gaps in your knowledge.

If you're interested in hard applied problems in climate/natural hazards, come have a look! We value a creative and problem-solving mindset with demonstrated experience communicating to science and non-science audiences. Hurricanes, floods, wildfire, thunderstorms, earthquakes, we do it all.

“The company is developing a digital twin of its supply chain and factory processes that will analyze the vulnerability of its supplier network in real time” www.wired.com/story/manufa....

#climaterisk www.ft.com/content/506f...

Important bit from the article to kill the hype: "Turner ... concedes that AI models like his also rely entirely on those physics-based models for training. 'It absolutely doesn’t work if you take their training data away and just use the observational data to train off.', he says."

The worlds first cat bond etf launched this week- via @kgreifeld.bsky.social We’ll have to see how it works but it’s a great development for the ILS market place www.bloomberg.com/news/article...

The CEOs that cozied up to national leadership really didn’t get much in exchange

An interesting #climaterisk cross-peril correlation I wasn’t aware of www.wired.com/story/how-hu...

“The excel of geospatial analysis” 😭😭😭

The things you can do with the latest LLM models, especially helping me with coding, is mind blowing

We’ve got to fix the insurance market place in the US for consumers. This is exactly when the industry should be showing its societal value but it’s hands are tied.

Maybe a NFIP for wildfire needs to be created? I would love to see a small universal levee paid by everyone to pool our risk together

I really hope we can somehow figure out hot to communicate physical risks to homeowners before these events happen

This is the new standard for what a truly catastrophic wildfire event can look like.

I am a simple man. I see developments in LLMs and I get excited l. o3 looks promising!

This wasn’t listed as a potential career path after my masters!???

Beautiful colour on this Hungarian red tonight.

I just used the o1 preview models- and I know I am late-but holy hell it did a great job at turning a complex sql script into a #rstats script.

Stuff like this has been taking up about 80% of my thinking time outside of work lately :)

can I code fast? no. but can I code well? also no. but does my code work? alas, no

If you’re going to write a news article about a research report and its findings, please - for the love of god - link to the original report

Can one of the rust wizards rewrite the dplyr backend for this r muggle? Pls

Really been pondering the nexus of cat risk and non-insurance markets as of late. Anyone know of any deep dives on these intersections?

I have ambitions to make a 'Climate Scientists in Industry' Starter Pack but my current list is looking a little light. Who's out there working in insurance, climate services, or other new sectors that I should be following?