time-split contour files

[AugustinMortier]

Context

Recent updates in the maps visualization page have made the regional/station time series always being used in synergy with the contour or pixel maps. Those time series include the time steps for which are computed the pixel maps, while the current contour plots already contain all time steps as keys and did not require time series to be loaded beforehand. However, considering increasing aggregated "geojson"in some projects (80MB uncompressed for CAMS2-82) and the not so straightforward caching implementation of compressed files, we could consider splitting the geojson files per time step, and rely on the preloaded time series to load relevant layers.

at least, for development/testing purposes, AeroVal would need:

Wish

• the contour entry point to accept a time-step argument and load the proper file.
• if the file does not exist for the requested time-step, just return an error. The front end will use this information and make another query without time-step information. It needs to be handled by the font-end somehow because the content of the file and callback functions will have to adjust.

Structures

current

• entry point: ${window.API_ROOT}/contour/${aeroval.project}/${aeroval.experiment.name}/${obsVar}/${mod}
• returned file: contour/{variable}_{model/obs}.geojson

suggested

• entry point: ${window.API_ROOT}/contour/${aeroval.project}/${aeroval.experiment.name}/${obsVar}/${mod}/${timeStep}
• returned file: contour/{variable}_{model/obs}_{timestep}.geojson

Note

The number of files in the contour directory might be very high. We might think about introducing sub-directories?

[AugustinMortier]

AeroVal now supports contour-split files. An implementation can be seen here, where contour files have been split with a post-script.

In the current impementation, the files are organised in a new sub-directory, as following:
{project}/{experiment}/contour/{var}_{model/obs}/{var}_{model/obs}_{timestamp}.geojson

[thorbjoernl]

So then is the idea that we run that script on all existing files? If so it is just a question of how to coordinate it and only a file path template change would be required on aerovaldb side.

If not I can probably do something like the filtering I did for glob_stats, maybe we can discuss tomorrow?