Seems a good time to make some decisions.
- Concluded that for the time being leave GISStemp alone, it is too different and liable to change.
- A lot has been achieved and so consolidating into something useful actually with users is the right move now.
- Adding fairly standard gridded datasets is easy, just added ghcn merged which took about 5 minutes, with one change needed, had to turn input scaling off. (only because I copied a template which had it on) Seems it might be one of the more interesting datasets.
What really needs doing?
- finish some documents which might be sufficient for someone reasonably computer literate
- see what happens when I try install on a victim W32 computer. Not expecting problems other than things missing which I had not considered.
- Might try it on a FreeBSD box here, if that starts working again… a spot of bother. If it works Linux ought not to be an issue. Mostly a matter of make Lua available and add any needed libraries. Make GNUplot available. The rest is any OS specific changes such as file specifiers.
I need to find some people to try this out.
What is working?
Most of it cosmetic things aside.
- Extract time series for a lat/long
- Extract time series for a list of lat/long supplied in a file
- Calculate weighted mean for global, southern and northern hemispheres
- Render image for a specified month to Hammer, Moll or Linear projection
- Render images given yearfrom and yearto
- A few ancillaries such as dataset aliases making the command line easier
- Conversion of published datasets into database format but this might be beyond some users.
- Lambert equal area with user specified central lat/long, this works but not properly.
- Rendering trips two warnings from GNUplot, but it works anyway
- Topping and tailing of datasets to the actual data extent could be better. (annoyance)
- Ditto with requested data (annoyance)
- How database creation works is a little awkward and needs user file moving
- Incorrect command line behaviour on Lambert but this is not fully working anyway.
- I am sure there is more
To be done
- Plots need a major workover and from this better sizing will appear.
- Improvement to filenaming
- Command line specifying of output type
- Command line invokation of numbered images for video creation using third party tools (video has been done)
- Continue trying to improve colour mapping
- Configurable contour line colour and possibly optional numeric labelling
- Cache overlays instead of create each time
- Speed optimisations, although speed is perfect reasonable, the usual, if it takes more than a couple of seconds some people get annoyed. Instant won’t happen even if I re-code in C.
Current databases here, all turnkey except for crutem2 which will not be normally used.
Earlier I took 5, had a trawl of a few web sites. Notice WUWT had a post on some Texan temperatures. Why not, extract a variety of time series for that lat/long and look. The end result is quite nice after I applied low pass to see what is happening. (I have filtering software with sensible end correction, another unpublished item)
I was surprised the datasets are so similar and eyeballing the WUWT post the local met records seem similar. Gridcells are not exact and in the case the satellite data is on a finer grid.