The DS-9640 data set is divided into 50 states and 9 regions. I was curious about regional variation. So here is a quick look at the 9 regional and the conus annual temps.
I found the spaghetti colors pretty difficult to follow. So a grey chart is below.
Tamino had an interesting post in Jan 2008, nearly three years ago, which I will refer to as Tamino’s Bet. It displayed one way to test if a trend has continued, stalled, or reversed. No doubt that it is not the most rigorous statistical treatment around – maybe not even close. But it sure is visual.
Anthony Watts has picked up and posted some investigation by Steve Mosher (here and here) into the effects of the 5 standard deviation filter used by CRUTEM. Steve was looking into the 5SD filter to see if that could explain the differences between his global averaging and CRUTEM. It doesn’t. To see why it doesn’t, lets take a brief look at the effects of the 5SD filter within CRUTEM.
Building my own gridded temperature anomaly code introduced me to the R “raster” package. One of the things that both I and Mosher realized pretty quickly is that much of the GHCN metadata work I had been doing this spring could be refactored in R. The original work was done with a mix of original and extended java classes developed from netcdf-java as well as a Linux version of the gdal toolkit. With Peter O’Neill quietly asking me some questions about those results, it’s well past time to take a another look.
There is a package for R called RGoogleMaps.
Just feed it coordinates and a zoom level and – voilà – you get back a Google Map image. They are not small – about 500K – 1 meg.
I presented a similar chart for GSOD data a few weeks back. Looks like it is time to present another for GHCN. These are the stations in v2.mean for the dates indicated.
Static images for every 10 years below the fold.