Providing predefined spatial boundaries

[KowatschFab]

One might want to use predefined regions as a spatial parameter to be able to generate values over e.g. different countries to compare. Such spatial boundaries could look like:

• hexgrid
• rectangular grid
• country grid

Another potential addition coult be the ISEA DGGs.

[bonaparten]

IMO, I think in case of hex grids and rectangular grids would be difficult to say how big a cell should be. The risk is that too big or too small cells are useless for "specific" requests (in this case it would be still better just to define bboxes). If the feature will not be used so much, I don't think the implementation effort will be worth it.
I think country grids is a good idea and more useful. It should be possible to use the name of the country, maybe as bbox value (bboxes=germany). The same could be possible with - at least important or capital - cities.
It could be also useful to use predefined physical regions i.e. bboxes=amazon_rainforest, although it is difficult to limit them (for example, at which latitude and longitude does it really end?).

[tyrasd]

with country/region/city boundaries there are some quirks to consider:

• which definition of boundaries should we use? think of disputed areas, but also the difference between land mass vs. the various maritime boundaries of influence of countries
• which data source for the boundaries to use? OSM might be incomplete or have broken geometries, other sources might be outdated.
• which spatial resolution is necessary? full resolution (~meters accuracy) might be available but make queries slow. Such accuracy is not necessary for all queries, so some might benefit from approximated/simplified boundaries.

[KowatschFab]

related to the new structure discussed in #160