Wild! But I am not sure the return periods are quite right. 2025 (+7°F above trend) and 1913 (−6°F) are nearly symmetric outliers in a 136-year record, suggesting the scale parameter may be underestimated. A non-stationary GEV anchored to the actual extremes might give more realistic return levels.
Aren't traditional cat models used for ratemaking legally tethered to historical claims? The newer 'climate risk' flavor (like FS) feels like physics-first/uncalibrated hazard modeling rather than actuarial loss modeling. Makes sense they’d be more unwieldy—but KCC being a 13x outlier is wild
I wonder how many of these complaints are actually material flood issues. People love to complain and so will call-in the smallest of puddles in the street as "flooding".
Nice! I’d also love to see some more local or regional examples, since those tend to be especially useful for adaptation decision-making.
I work in industry, and I see this all the time. e.g. professional services firms often prioritize soft skills in hiring, which means many “climate scientist” roles end up filled by people who don’t fully understand the data or methods they’re working with.
ESSOAR
Yeah. I never see evaluation of the alignment of observed local/ regional trends with near-term climate model projections. Dunn et al. 2020/2024 is helpful. Models are awesome but we should consider diverse pieces of evidence when making million dollar decisions. Not just do what NASA-NEX data says.
"There is not a lot of agreement on what best practices would even look like (even from domain experts deeply enmeshed in this stuff who really do know what they're talking about)."
Yes! Really comes down to specific use cases (e.g., "changes in 1% AEP rainfall" or "number of 'wet' days")
Yes. Using these data off the shelf will under-sample extremes.
An entirely different approach is needed for wet tails.
The training data don't work well (Livneh, Metsim, ERA5+WRF). Pierce et al. 2021 helped, but it needs to be obs-based, spatial extreme value (e.g. the Atlases).
I think govt involvement is also about liability. There are better private-sector flood maps out there, but what CEO is going to expose their company to litigation when people don't like the publicly shown scores? FEMA deals with LOMAs day and night, and they have sovereign immunity.
How much of this is new policies vs just crazy inflation since 2020? fred.stlouisfed.org/series/CAUCS...
Many ultra-high-net-worth individuals are drawn to the exclusivity and aesthetic of waterfront property, regardless of the climate risks. The ability to self-insure and pay in cash removes many of the typical financial constraints or regulatory hurdles that would otherwise deter such investments.
Yes. Could be representativeness issues (grid vs point) or other assimilation or model biases. Gridded products ≠ in situ observations. I developed a dataset to deal with these issues: essopenarchive.org/doi/full/10....
Rx burns may have public health benefits, but because those benefits are delayed, they’re often heavily discounted. That can make them harder to justify in the near term, despite potential long-term payoff
New online Wildfire Mitigation Plan Database from PNNL. Over 400 wildfire mitigation plans from 170 utilities across 19 U.S. states. Useful to understand, compare, and improve strategies for managing wildfire risks to power infrastructure.
www.pnnl.gov/publications...
Glad this issue is getting attention! But isn't this what trend preserving approaches like Cannon et al. (2015) are supposed to help with? LOCA's bias correction may not be trend preserving (not sure)
I’m an independent climate risk researcher working at the intersection of climate science, infrastructure, and decision-making. I focus on hazard modeling, risk quantification, and integrating social science into climate resilience. More on my work & publications: www.djrasmussen.co (12/12)
Our preliminary estimates are consistent with FEMA’s elevations—but are based on 70 years of observations, not modeled storms. They also avoid potentially problematic EV fitting assumptions (e.g., MLE) and include full uncertainty estimates (see: ascmo.copernicus.org/articles/11/...) (11/12)
More work is coming. We’re expanding this framework (again led by Joao Morim) to develop fully probabilistic estimates of storm surge extremes for the U.S. and select territories (10/12)
Regardless of attribution, our findings have implications for infrastructure design. Planners and engineers with long horizons should consider storminess w/ sea-level rise. The now-revoked Federal Flood Risk Management Standard (FFRMS) is still a useful reference (8/12) www.fema.gov/sites/defaul...
Though modest in magnitude, I’d argue these trends matter. It’s possible that internal climate variability and anthropogenic forcing are exerting opposing influences—partially masking the full signal. Disentangling these drivers requires large ensemble simulations (7/12)
We do not directly attribute these changes to anthropogenic climate forcing, but they are consistent with known shifts in sea surface temperatures (SSTs) and mean sea level, both of which are projected to continue rising. (6/12)
These trends have accelerated in several regions since 1975. In hotspots—especially the Gulf and Southeast Atlantic coasts—storm surge extremes are increasing at rates comparable to or exceeding MSL rise. (5/12)
The result: we find robust, spatially coherent trends in storm surge extremes along 70% of the U.S. coastline—contradicting the longstanding view that such regional trends in storminess don’t exist.
In some areas, the magnitude of these trends rivals contributors to regional sea-level rise. (4/12)
We use a spatio-temporal Bayesian hierarchical model to combine 70 years of tide gauge data across the U.S. This framework accounts for spatial dependence across sites and quantifies uncertainty in the estimated trends. (3/12)
Previous studies using tide gauges have identified storm surge trends at a few U.S. locations—but the results have been inconsistent. No robust evidence of regionally coherent trends have emerged. (2/12)
Rising mean sea levels are clearly driving more frequent coastal flooding. But what about changes in storminess? That’s been harder to assess, due to limitations in how storm surge is measured. Our new study in Nature Climate Change (led by Joao Morim) addresses this. 🧵 (1/12)
Been wondering about home value growth. JP Morgan mentioned it, but it’s often left out of the insurance gap convo. Home values have nearly tripled in some places since early 2010s. How could premiums possibly keep up? Lots of factors at play but feels like more attention is needed on this.
Interesting read, thanks! Home values have nearly tripled in some markets since 2012 (eg., Calif). How significant is the gap between rising asset values and the rate at which premiums have increased?
