Use Case: Uncertainty visualization

Implementation: #7

Summary

Typical web portals for geospatial data pay little attention to communicating the uncertainty in data. Often this is because the source data simply does not contain this information, but it's also true that there is not much support in typical web portal tools (e.g. Web Map Services) for transmitting uncertainty information (note that a recent paper describes an approach for adapting WMS for doing this).

Uncertainty can be present in both continuous (numeric) and categorical data, with different communication methods being appropriate in different cases.

Detailed Description

1. User selects a dataset and variable for display in MELODIES portal
2. MELODIES portal detects that the variable comes with associated uncertainty information, and suggests approrpriate visualizations (e.g. contour-over-raster, bivariate colour maps etc)
3. The user can switch the display of uncertainty information on and off
4. An "info window" explains to the user what is going on

I envisage that the MELODIES demo portal would have some "intelligence" to detect uncertainty information in source data. Perhaps this can be transmitted in dataset or coverage descriptions.

Some issues

1. Should we render images on the server (as in WMS) or can we create these visualization on the client? (Is there a Javascript library for generating contours?)
2. How do we encode the uncertainty information in the dataset or coverage descriptions?

What's the Linked Data story?

The semantic relations between the different components of uncertain variables (e.g. mean and variance) could be encoded as RDF.