The water bodies on this map are a model assumption, not a feature of the sea. They are administrative polygons drawn for the Water Framework Directive. Every figure computed inside one — including the colour — is a statement about that partition as much as about the water, and whether the sea changes where these lines run is an open question rather than a settled one. Even a name as obviously singular as “Roskilde Fjord” is a claim about the water and not a fact about it.
Colour is coverage by two named layers, and nothing wider. “Fitted model” means DCE (2015) published a statistical relation between nutrient load and an indicator for this area, fitted on 1990–2012 data. “Repeated observation” means at least one bathing station with a multi-year record. Those two layers are counted because they are what the water-plan assessment leans on — they are not a census of what Denmark measures, and an earlier version of this caption said they were.
Two cautions, because this map is easy to over-read. First, the requirement does not rest only on what is coloured here. Alongside the fitted relations sit 3D mechanistic models that resolve stratification and oxygen at depth, an eelgrass transect programme, and an annual oxygen-deficit survey — reconstructed from the primary method documents in Grundlaget. A map that showed only these two layers and called itself the state of knowledge would be arguing against a thinner case than the one that exists.
Second, a pale area is a statement about these layers in this extract, never about the sea. ODA’s own topic extracts carry light casts back to 1980 and CTD profiles back to 1970 at stations this colouring does not count, and what a coverage figure is missing is itemised in There is no Denmark. The fitting window also ends in 2012, which is fourteen years before the policy it is used to support — that gap is the subject rather than a footnote.