The next speaker is Chris Daly, the renowned developer of the PRISM spatial climate modelling system. His work has earned him the 2004 Applied Meteorology Award from the American Meteorological Society.
We will discuss his 2006 paper: "Guidelines for assessing the suitability of spatial climate data sets", which can be found at the link below:
http://onlinelibrary.wiley.com/doi/10.1002/joc.1322/abstract
Feel free to post any questions or comments in advance of or during the discussion at UNSW at 2pm on Friday!
we were impressed by the importance of ground-truthing climate maps and seeking expert advice on their accuracy. We also recognise that this takes a lot of resources so there is no easy win here...
ReplyDeleteThe CV data was potentially worrying too... that when you validate on data kept aside from the original model you can get very different answers. Especially if predicting to them involves extrapolation... need weather stations on those mountain tops!
ReplyDeleteI'm a little worried about the error estimation - if you can't predict well then at least you want to have decent estimates of the error so you are aware of and can account for the uncertainty. But there is the suggestion here that we can't come up with reliable estimates of error, which I'm worried about - I hope statistical modelling of climate maps has improved to address this over the ensuing years!
ReplyDeleteLemmings with thermometers has been mentioned but I missed the reference...
ReplyDeleteibuttons have been mentioned a bit because a few ecologists here have been using them in the field and using these to map climate in a (hopefully) more ecologically realistic way.. mostly with Mick Ashcroft at the Australian Museum.
ReplyDeletediscussion about which resolution to use... some suggestion that maybe we should use coarse resolutions because these can be estimated more accurately, but that doesn't help if what is happening at a fine resolution is more relevant to your qn!
ReplyDelete