[Proposal] Add guidance to docs to instruct users to include key non-metadata in datasets themselves

[odscrachel]

What is your proposed change?

Proposal to remove fields which are not consistent across the entire data file being described. This also includes fields which can be found within the data itself. Spreadsheet

Can you provide an example?

Fields suggested for removal:

• anyOf.exposure.model.occupancy
• anyOf.exposure.model.occupancy_time
• anyOf.exposure.model.taxonomy_code
• anyOf.exposure.model_add_attributes
• anyOf.vulnerability.model.taxonomy_code
• anyOf.vulnerability.specifics.par_names
• anyOf.vulnerability.specifics.im_name
• anyOf.vulnerability.specifics.im_units
• anyOf.vulnerability.specifics.n_events
• anyOf.vulnerability.specifics.n_assets
• anyOf.vulnerability.specifics.is_fit_good
• anyOf.vulnerability.specifics.is_edp_thre
• anyOf.vulnerability.specifics.is_dm_factor
• anyOf.vulnerability.specifics.is_casualties
• anyOf.vulnerability.specifics.is_downtime

[stufraser1]

Noting that we should instruct users to include these fields in the data files, not assume they will.

[odscjen]

Some of these are now being kept and incorporated into vulnerability.function see #83

[odscjen]

The fields listed in the initial description have all been removed. Renamed this issue to reflect the remaining task.

[duncandewhurst]

Noting that we should instruct users to include these fields in the data files, not assume they will.

@stufraser1 do the individual fields need to be listed somewhere? If so, we can add a new "what to include in risk datasets" page under how to publish risk datasets. Otherwise, the sentence added in #234 should be sufficient.

[duncandewhurst]

@stufraser1 @matamadio to draft content.

[matamadio]

@stufraser1 draft content for review

What to include in risk datasets (What is not included in metadata)?

There is an important distinction between data content, data structure and the metadata schema. The RDLS provides a metadata standard but does not impose any specific data structure.

Hazard

A hazard dataset contains, as a minimum, the intensity value at a particular location for the event or scenario represented. Aside from the hazard value and location, most information about the data is metadata, therefore part of the RDLS metadata schema.

Exposure

The RDLS metadata schema can describe the main category of exposure, associated quantities and cost units, and taxonomy standard. Other attributes such as the taxonomy code or the quantity variable associated to each feature (e.g. replacement cost or number of people for a goven location) are part of the data itself.

Vulnerability

The RDLS metadata schema covers a fairly large amount of attributes and parameters to describe vulnerability datasets (models). The actual parameter values from a vulnerability relationship would be part of the data.

Loss

The RDLS metadata schema can accomodate attributes describing both geospatial and tabular loss data. Key metatada include the type of hazard and exposure considered, the type and unit of impact considered and related economic cost units. Impact and loss values are part of the data content.