Scope

Main requirements

Gridviz is a visualisation tool for gridded statistics on the web based on vector web-mapping approach (dynamic client-side on-the-fly rendering).
High level of genericity: The tool should be reusable for any kind of gridded statistics. It should comply with grids defined in any CRS, for any resolution (rounded) and any geographical extent (European or national).
Only squared grid cells.
Fast visualisation (reduced loading and rendering time).
Pan and zooming functionality. Possibly, a zoom threshold could be defined in order to shift from raster to vector visualisation mode. This threshold should depend on performance considerations and grid resolution.
When a grid cell is passed-over, its value should be shown in a tooltip text.
For developer users: A developer reusing the tool should be able to specify which data source is used for each zoom level. These datasources could be entire CSV files or tiled CSVs, depending on the zoom level. Different styles should be proposed (dee below some proposals). Ad-hoc specifications for the input data consumed could be defined and should be documented (here). It should certainly follow tiling principles. - Tools/services to support production of these data should be provided.

The maps will be shown as a coverage. No background will be used to ensure no visual bias is introduced by the changing darkness of the background. Information on the geographic context will be limited to foreground layers such as geographical names (main cities) and administrative borders.
The tool should be based on open-source technologies to ensure wide sharing and reuse. The intention is to share it among the ESS to encourage the production and dissemination of gridded statistics.
For performance reasons, the visualisation should be performed in the CRS of the grid (no reprojection on client side). It should especially not be constrained to the web mercator projection.
The tool could be designed as a new Leaflet layer type and published as a plugin.
First test with existing gridded statistics: Geostat population grid (2006 and 2011), accessibility grids, tourism attractivity grids. National datasets at sub-km resolutions could be used. The longer term objective is to be ready for the publication of the census 2021.
Rendering based on SVG elements added on the fly to the DOM should be forgotten - strong performance limits are expected for large number of cells. Possible rendering techniques: Canvas, webGL. ...