Bob

@fluxcapdoodah

Investigating ideas, arguments, evidence, strategies and stories. Likes, follows and retweets are bookmarks, sometimes quizzical, not endorsements.

Vrijeme pridruživanja: lipanj 2018.

Tweetovi

Blokirali ste korisnika/cu @fluxcapdoodah

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @fluxcapdoodah

  1. proslijedio/la je Tweet

    Of course, as I urged in 2003 in "Heat storage within the Earth system" the robust metric wrt global warming is to monitor changes in ocean heat content.

    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet

    Informative by J. Hansen. Incl "no good excuse for why we are not monitoring the aerosol direct and indirect climate forcings" Also, since warm bias (& other uncertainties) in land sfc T, if insist on sfc info being used, just focus on SST trend.

    Prikaži ovu nit
    Poništi
  3. proslijedio/la je Tweet
    prije 6 sati
    Odgovor korisnicima

    And, more broadly, climate models are't "fit" to observational data. They are spun up in 1850 conditions and fed our best estimates of CO2 concentrations and other components of radiative forcing to estimate how the climate changes from that point.

    Poništi
  4. proslijedio/la je Tweet
    prije 6 sati
    Odgovor korisnicima

    Out of sample testing such as paleoclimate, kicking the tires of parameter choices in highly idealized experiments to compare with theory, looking for emergent constraints based on observed physical processes that weren't tuned during model development, etc.

    Poništi
  5. proslijedio/la je Tweet
    prije 6 sati
    Odgovor korisniku/ci

    How does one prevent overfitting in climate modelling?

    Poništi
  6. proslijedio/la je Tweet
    prije 5 sati
    Odgovor korisnicima

    True, though if anything I'd expect tuning to result in poorer correlation between hindcasts trends and ECS. In CMIP6 we see largely the opposite, which suggests to me that the effects of higher model sensitivity swamped any model tuning. Could be wrong though!

    Poništi
  7. proslijedio/la je Tweet
    prije 5 sati
    Odgovor korisnicima

    I don't trust the ECS from a model with an accurate hindcast trend much more than one with a less accurate hindcast trend. Could just be tuned more or have strongly compensating forcing errors / responses to forcing. 2/2

    Poništi
  8. proslijedio/la je Tweet
    prije 5 sati
    Odgovor korisnicima

    Relationship between hindcast GMST and forecast GMST (what you're talking about) is different than relationship between hindcast GMST bias and forecast GMST bias (what I'm talking about). I'd argue we care much more about the latter than the former. 1/2

    Poništi
  9. proslijedio/la je Tweet
    prije 5 sati
    Odgovor korisnicima

    It's a notably better constraint in CMIP6 than CMIP5 (where ECS and TCR were lately uncorrelated with hindcasts performance):

    Poništi
  10. proslijedio/la je Tweet
    prije 5 sati
    Odgovor korisnicima

    Yes, but hindcast GMST is a poor constraint on forecast GMST ().

    Poništi
  11. proslijedio/la je Tweet
    prije 6 sati

    To clarify a tad: hindcasts from CMIP6 perform fairly well on average, but those models with lower sensitivity (< 4C) tend to do a notably better job than those with high sensitivity.

    Prikaži ovu nit
    Poništi
  12. proslijedio/la je Tweet
    prije 6 sati
    Prikaži ovu nit
    Poništi
  13. proslijedio/la je Tweet
    prije 6 sati

    Climate modeling is one of many lines of evidence that we use to determine climate sensitivity. Across all those lines of evidence our best estimate is still ~3C per doubling CO2, with a range of 1.5C to 5.5C (accounting for these new models).

    Prikaži ovu nit
    Poništi
  14. proslijedio/la je Tweet
    prije 6 sati

    Lets not get ahead of ourselves; climate models in general have been good at projecting future warming, but the latest generation models with very high climate sensitivity actually don't do a particularly good job "hindcasting" past warming:

    Prikaži ovu nit
    Poništi
  15. proslijedio/la je Tweet
    prije 8 sati

    Important to note: models with lower climate sensitivity also correctly simulate the historical record.

    Poništi
  16. proslijedio/la je Tweet

    The same climate models that have successfully predicted actual warming to date are now running "red hot," projecting nightmare warming scenarios of 5C and more. Scientists aren't sure if the models are wrong or if our goose is cooked. ()

    Poništi
  17. proslijedio/la je Tweet

    Oxford official to students demanding divestment: "I am not able to arrange any divestment at short notice. But I can arrange for the gas central heating in college to be switched off w/immediate effect. Please let me know if you support this proposal.”

    Poništi
  18. proslijedio/la je Tweet
    prije 11 sati

    While the embarrassing RCP8.5 climate scenario is gradually dispensed with, climate alarmists are replacing it with a new model scare via

    Poništi
  19. proslijedio/la je Tweet

    RCP8.5👀⤵️ BBC News - Dry, hot summers could become 'common' in Scotland

    Poništi
  20. proslijedio/la je Tweet
    prije 6 sati

    I like the informational version of I am working on. The maths has, as usual, shown me the way, & I'm trying to fill in the physics of why it works. I derived with (Unruh) photons, then from Heisenberg. Info- is a 3rd confirmation & feels fresher

    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·