[Proposed by Gerard Heuvelink]
Which soil class map is more informative? A map with a detailed legend and low purity or a map with a coarse legend and high purity? Can we characterise the information content of a soil map with the Shannon entropy or differential entropy and if yes, what does this tell us? Can the economic value of a soil map be assessed and is this then a proper measure of its information content? How about the fitness-for-use of soil maps? Can we use the concept of soil map information content to help guide and improve our soil mapping activities?
We cannot answer yet to these (good) questions because we do not have enough experience of real uses of our DSM products. We must not limit our effort to only delivering these products. We should be involved in projects with real users of soil data, as diverse as possible. Anyway, a prerequisite to this challenge is to produce accurate and communicable measures of uncertainty of the DSM products (see Challenge 2) so that all the information, i.e. the legend and the purity in the example of a soil map, is on the table.