EGU: Thursday

A view of the cathedral, which looms up over the streets on my way to the U-bahn.

Start off with a pile of global radiation / global dimming / global brightening, and indeed moisture trends (the best dataset for soil moisture comes from the FSU cos they cared about their wheat fields). A model works better to simulate soil moisture trends (increasing) if dimming taken into account.

Put up my wildly exciting sea-ice and AR4 models poster and browse amongst the others. Bit of a misc morning.

Crucifix: looking at climate sensitivity. LGM climate sens in models do not seem to be related to their CS for 2xCO2. Or for mid-holocene, or for the seasonal cycle. Because the forcing has a different pattern. What wasn’t clear was whether this affects the idea of constraining future CS by LGM bounds. Much ask Julia (or James, if you’re out there).

Krinner: why was Asia ice free in the LGM? 10/15 of PMIP models build up permament snow cover there. Suggestion: because of dust deposition there affecting the albedo. It was much dustier during the LGM, and there is a nice Gobi desert for a source, so this fits, and in his GCM it works.

Quirky poster of the day: “Do geomagnetic variations affect the foliar spiral direction of coconut palms?”. Coconut palms, apparently, grow either left or right hand spirals and this is not genetically determined; it is related to latitude, but better to geomagnetic latitude; and may well relate to earth currents. Curious. Nice to see some science-with-no-obvious-consequences still being done.

Bunk off to the Natural History museum for the afternoon, soon. First…

To Julia Hargreaves talk in the PMIP session. Its intersting stuff, or at least I think it is, because in 12 mins there was not enough time to understand it all. Or evem I fear, the main point; though it still connects to the estimating climate sensitivity, this time using proxies for LGM tropical ocean T and the MIROC gcm. I ask a “rude” question :-) (is your model good enough to be saying anything useful?) to which an ambiguous answer is returned.


  1. #1 Julia Hargreaves

    the main point is that …
    … you ought to use out of sample data that says something meaningful about the thing you are trying to predict in order to validate your estimate.

    [OK, I can go with that – W]

    In answer to your rude question – yes we showed that our prediction for the model isn’t very good, but at least we know that – most other people trying to predict climate don’t do any quantitative validation at all, so they don’t know their predictions are bunk!!!!!



New comments have been temporarily disabled. Please check back soon.