29 November 2014

Still living

Still living, just been away from the blogosphere on other things.  Some of them will find their way back here as posts. 

In the mean time, Kevin O'Neill, who has introduced an interesting idea in one of his comments at Multiple Working Hypotheses, has encountered one of the annoying things about to do science.  Namely, a data set he has been using was discontinued.  I sympathize.  At work, a satellite I was about to make use of in our operations died the week before our implementation.  A resource for looking in to climate (at least as long as you don't need to go before 1979) is the The NCEP Climate Forecast System Reanalysis.  The numerical results of which do include the outgoing longwave radiation.  There are slightly differing (in resolution and time spans) archives at NCDC and NCAR.  The NCDC archive mentions that it is 500 Terabytes.  That sounds about right.  It'll take a while to download.  Or you can use the NOMADS subsetting capabilities to extract just the fields and regions that you're interested in.

I'll also be getting back to my sea ice guesses for ARCUS, and evaluating them.  This time around, as I prepare at work to do some more substantial sea ice things, I'll do a general survey of how they all performed, in all years.  Then to focus on the remaining method.  A point related to the method of multiple working hypotheses is that you have to be active in weeding them down.  You'll generate more and better ones to take their place.  But you have to make room first.

Kicking around in the 'almost done' bin is a post on how long it takes to detect an acceleration in global mean temperature.  Acceleration being a change in the trend.  This was prompted by some todo at the Washington Post Capital Weather Gang, when someone claimed that he'd found a negative acceleration (i.e., a decrease in trend, which would, if continued, turn to a cooling trend).  I'll give away the answer here -- it takes about 40 years (at least 40 years) to define the acceleration. 

Another 'nearly done' is to revisit Does CO2 correlate with Temperature?.  It's almost 6 years since the original, and for all 6 years, there's been talk of 'hiatus', 'pause', and 'climate hasn't changed in N years'.  N varies a lot by who is talking.  Perhaps the additional data will break the correlation, since CO2 has certainly been rising.

I'm also going, at some point, to play on the blog with Bayesian statistics.  Readers who like Bayes, please do correct me as I (inevitably) make mistakes.

Plus in January, when I'm done with the meetings, holidays, and other things, of December, I'll hang back out the question place shingle.  Probably some minor notes before then.