Automate data map

[stufraser1]

Automate process to update data map via geojson, on data harvest

[ErikKBethke]

See GFDRR/geonode#31

[stufraser1]

@fjacon What would be the level of effort, to build in a process to generate a GEOJSON file in the data harvesting process, that could be used to populate the TH data coverage map? This is currently done manually via GIS software (manually produce shapefile, convert to GEOJSON and host on github for implementation).

[arnaud-morvan]

As I understand, we should add in the web app a geojson rendered view of the coverages based on the decision tree results.

Note that it seems there is no real need for a real file. The geojson could be rendered at runtime, this would simplify the publication process. In that way we would be capable of displaying differents coverage maps for the admin and public databases. Note that the public data could be cached based on the publication date (like other pages as I remember).

The only thing I do not know precisely is how you differentiate the regional and national datasets (we only have a global / local differentiation in processing part for the moment).

[stufraser1]

OK, that sounds fine. There is no particular need for a GEOJSON file for export of this data coverage map.
In the present map, we manually assigned datasets labelled as 'local' but known to have an extent larger than one country as regional. There is no obvious and clean way to auto-identify these at the moment: Perhaps we can include a label in the metadata to identify regional datasets (e.g. as part of a string in supplemental information?), or perhaps cleaner - can you write into the code something that can read whether the Geonode region list is not global and not only a single country? - I.e. the region list would be populated with > 1 country name AND/OR > 0 region names (this list is hierarchical in Geonode)

[arnaud-morvan]

As I understand, we should add, in the web app, a geojson rendered view of the coverages based on the decision tree results.

Note that it seems there is no real need for a real file. The geojson could be rendered at runtime, this would simplify the publication process. In that way we would be capable of displaying differents coverage maps for the admin and public databases. Note that the public data could be cached based on the publication date (like other pages as I remember).

The only thing I do not know precisely is how you differentiate the regional and national datasets (we only have a global / local differentiation in processing part for the moment).

[fjacon]

Reminder:
It would take about 4 days (~30h) to deal with:

• Create SQL request and geojson view of coverage map data.
• Optimize perfomances, we may use some cache and/or PostgreSQL materialized view, refreshed after decision tree and/or publication.
• Use the new view in Coverage map dialog (we may use one request per hazardtype to reduce loading time and memory usage on first show).
• Deploy on production

[stufraser1]

Please also highlight in red/italic text in the pop-up of data layers for each country, which layer is rated most high quality and will be shown in the TH UI.