Home > analysis, Datasets, temperature > Time maps

Time maps

I am told the name for these is Hovmoller maps.

RSS TLT map

Click for full size.

UAH TLT

Click for full size

Both these datasets suffer residual annual cycle. I’ve done a lot on this and for the post 2003 period I can remove it from RSS or either of UAH 5.2 or 5.2 to very similar results, identical for UAH. Not shown here.

The maps themselves show a critical thing, the heat flow from equator to poles.

Getting there was fun.

After a good deal of trouble and by chance learning why so many people have  perplexing problems with gnuplot I’ve managed to get sane results with a common time axis. I so wish people would say what they mean and not omit.

Turns out that gnuplot plot (2D data) and splot (3D data) may not be validly mixed in a multiplot, more than one graph done on a single sheet. Unfortunately that kind of works so what you are doing wrong never surfaces.

Answer is trivial, usually is. Don’t mix plot and splot. The secret trick is supply a dummy third dimension for 2D data.

‘splot “-” using 1:2:(0)

The dummy is the (0) instead of plot “-”

File dash means the data follows on the command line and that was yet another thing which took months to figure out.

So what is the difference between “-” and –

A huge amount but the two are never joined in the documents. Former means text data will follow and latter binary data will follow. Documentation for binary is even worse, more or less isn’t anything sane, just hand waving to people use it.

Text data turns out to be trivial. Just a list, with blank lines to separate rows and finally the letter e to say end. Result is I can avoid intermediate files and just send to data from Lua to gnuplot via a pipe.

Advertisements
Categories: analysis, Datasets, temperature
  1. April 7, 2014 at 08:26

    The most noticeable feature I see is full circles not curves. Diam 4 years.

    BTW if you find problems with gnuplot (doc or otherwise) it’s worth forwarding sugestions to the devel list. (It’s a low volume list if you subscribe.If you don’t subscribe the comment takes a day or two to be vetted manually).

    gnuplot-beta@lists.sourceforge.net

    The project is quite responsive, sometimes amazingly fast. I’ve made suggestions that have been added to CVS source code in 24h !

    I’m glad to see you’re getting to grips with it. It took me a while too but it’s amazingly powerful once you do.

  2. April 7, 2014 at 13:36

    Circles? That’s an interesting observation.

    I’ve seen these in all manner of ways and sizes. These come from gridded dataset which are in a mess. This tends to show.

    I think there are novel things which could be done if anyone has the time and wish.

    Here is a fragment already on file

    There are a couple of thump / hole there. If this could be resolved high res in all domains I would be interested in whether it is a damped decay oscillation.

    Crutem2 and so on start to show just how poor these data really are.
    Satellite stuff is all very well but it is not surface.

  1. No trackbacks yet.

Leave a reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: