Location data for roads

[duncandewhurst]

CoST Ukraine capture the following location data for infrastructure projects relating to roads:

• The identifier for the road, drawn from either the state roads index, or a regional administration's road index, e.g. 'T03' or 'Т-16-10'.
• (TBC) The index the identifier is drawn from.
• The start and end locations for the works, specified by a kilometer marker plus a number of meters. One project can impact multiple sections of the same road, so the start and end location pairs are captured as a semi-colon separated list in single field, e.g. '53+200 - 57+100; 67+600 - 70+100`
• The region the project is in

Currently this data is modelled using the locations field in OC4IDS with one location object per road, e.g.

{
  "id": "3926",
  "description": "516+200-646+200",
  "gazetteer": {
    "scheme": "М-03"
  },
  "address": {
    "region": "Харківська"
  }
}

I've recommended that the mapping of the road identifier to .gazetteer is updated as follows, based on the index the road identifier is drawn from being known. E.g. for a road in the state roads index:

  "gazetteer": {
    "scheme": "UA-SRI",
    "id": "M03"
  },

This issue is to consider modelling for the list of start/end locations, since .description isn't a good mapping.

Terminology used in linear referencing systems for start and end locations includes 'from-measure' and 'to-measure' (from arcgis) and 'fromToLocation' (from ISO 19148).

[jpmckinney]

We can maybe have:

{
  "id": "3926",
  "descriptors": [
    {
      "scheme": "kilometer-range",
      "value": [53.2, 57.1]
    },
    {
      "scheme": "kilometer-range",
      "value": [67.6, 70.1]
    }
  ],
  "gazetteer": {
    "scheme": "UA-SRI",
    "id": "М-03"
  },
  "address": {
    "region": "Харківська"
  }
}

1. There are many, many ways to describe a location. We have fields for three very common means (geometry, gazetteer, physical address), but there are others like kilometer marker ranges, an item's warehouse location, etc. I don't think we eventually want a dozen miscellaneous descriptors, so an array of user-definable descriptors seems better.
2. I don't understand why the notation is kilometer + meter instead of kilometer . (decimal point) meter. The latter is just using regular numeric notation... Also, with JSON, we can use an array to express an interval ([min, max]) instead of a string.

Note that the array would be wholeListMerge, which is fine.