search

Ouranosinc / xclim

Library of derived climate variables, ie climate indicators, based on xarray.
https://xclim.readthedocs.io/en/stable/
Apache License 2.0
333 stars 59 forks source link

Additional Indicator: Record-breaking days #1503

Open SarahG-579462 opened 1 year ago

SarahG-579462 commented 1 year ago

Addressing a Problem?

This new article by Hyde and Albarracín in Nature shows that "each record heat day in a county in 2022 increases perceptions that excessive heat is getting worse, even when controlling for average temperatures, the number of extreme heat days, and demographic factors. Our estimates suggest that exposure to sixteen record heat days predicts roughly the same difference in excessive heat perceptions as between the average Democrat respondent and a political independent. This effect is stronger for populations that are more skeptical of climate science, including Republicans, as well as respondents with weaker beliefs in climate change and more frequent consumption of conservative media."

It would be a useful communication tool to be able to get these statistics for future climates, especially if there are observable trends in the data. Could maybe be interesting for Portraits Climatiques as well.

Potential Solution

Given a series, find the number of days in a given period (year, season, month, etc) where the record temperature is broken (bonus: within the top N temperatures). Doing this for precipitation would be cool too.

Additional context

The baseline record temperature could be given by the user, or assumed to be the first year of the series (in which case NaNing this value might be worthwhile?).

Contribution

Code of Conduct

huard commented 1 year ago

Interesting ! I'm wondering if that function should return more info than just the number of broken records. I'd want to know on which days it happened, what was the year of the previous record, and by how much it got beaten.

Zeitsperre commented 1 year ago

Interesting ! I'm wondering if that function should return more info than just the number of broken records. I'd want to know on which days it happened, what was the year of the previous record, and by how much it got beaten.

Would we want to store that information as a handful of variables/coordinates added to the Dataset, or would it make more sense to have individual functions for each of those statistics? Neat idea.