Michael Tippett 🎃

@mktippett.bsky.social

Minor bug fixes and performance improvements

I think we should also measure our success by whether what we do "enables people to understand and think more clearly and effectively about the world around us." I think that climate science was *already* struggling to measure success in this way, and the advent of LLMs will make this harder!

ENSO forecasts from individual models can be “idiosyncratic” and their average tends to be better. Here is the NMME multimodal mean dressed with a synthetic, performance-based 100-member ensemble (with code) github.com/mktippett/nm...

Line chart showing a synthetic 100-member NMME forecast plume for Niño 3.4 seasonal sea-surface temperature anomalies from JAS through AMJ. Thin pale-blue lines represent individual ensemble members, dashed gray lines mark the 10th and 90th percentiles, and a thick black line shows the multi-model mean. The mean rises from about 1.7°C in JAS to a peak near 2.6°C in OND–NDJ, then declines steadily to about 0.6°C by AMJ. Ensemble spread is widest around boreal winter and remains substantial into spring.
Michael Tippett 🎃@mktippett.bsky.social · 4w ago

Here is one recipe for computing relative Nino-3.4 in models. Bonus content: plots of the latest NMME forecasts, skill scores, and yes, code github.com/mktippett/nm...

ENSO forecasts from individual models can be “idiosyncratic” and their average tends to be better. Here is the NMME multimodal mean dressed with a synthetic, performance-based 100-member ensemble (with code) github.com/mktippett/nm...

Line chart showing a synthetic 100-member NMME forecast plume for Niño 3.4 seasonal sea-surface temperature anomalies from JAS through AMJ. Thin pale-blue lines represent individual ensemble members, dashed gray lines mark the 10th and 90th percentiles, and a thick black line shows the multi-model mean. The mean rises from about 1.7°C in JAS to a peak near 2.6°C in OND–NDJ, then declines steadily to about 0.6°C by AMJ. Ensemble spread is widest around boreal winter and remains substantial into spring.
Michael Tippett 🎃@mktippett.bsky.social · 4w ago

Here is one recipe for computing relative Nino-3.4 in models. Bonus content: plots of the latest NMME forecasts, skill scores, and yes, code github.com/mktippett/nm...