[Proposal] Update Loss schema with more attributes

[matamadio]

What is the context or reason for the change?

From Samuel Roeslin, DRMKC - Risk Data Hub.
We are making progress and added the fields related to the RDLS Loss metadata in the Risk Data Hub (RDH) database

To capture the loss/impact, our [JRC's] system works as follow using the following RDLS fields after an asset code has been selected:

loss_type,
impact_type,
impact_metric,
metric_dimension,
impact_unit.

We have a comment/suggestion regarding impact_unit. We feel that this field could be reviewed/improved.

What is your proposed change?

We feel that:

• This field should be sufficient by itself to relate to a single type of unit. This comment relates mostly to “count”. For this category impact_unit has always to be interpreted in combination with impact_metric as impact_unit alone is not sufficient to understand to which loss it relates. From the screenshot I put below you can see that impact_unit = ‘count’ can be both related to economic losses related to damaged road/rail, a number of bridges affected, and population losses.
• We would suggest that this field should capture more fundamental concepts related to the unit i.e. count, time, surface (we would although add people, length, cost for example).
  The actual unit can be captured in a separate field like what is done for the currency.
• As possible unit we also would like to see length (not just area) as sometimes losses are captured with m or km (i.e. number of kilometers of road affected)

Can you provide an example?

Attached you’ll find an extract on how we use the RDLS in our system. This is a non-exhaustive list as more asset categories can always be added.

asset_metric_table_RDH_20250616_ToShareWithWB.xlsx

[stufraser1]

See the quantity_kind codelist which includes: area, count, currency, length as option to describe the quantity. We can add 'time' to this codelist to cover the use of disruption where impact unit is year, week, month, hour, day (already included in impact unit codelist), and 'unit of production' to cover agricultural crop losses.

I'm not sure what 'surface' refers to in the proposal, in "capture more fundamental concepts related to the unit i.e. count, time, surface." Clarification needed before suggesting a solution.

impact_unit can be improved by adding fields to describe:

• 'metre', 'kilometre', 'mile', 'yard' to give a unit to quantity kind = length.
• 'people' for using metric_dimension = population and quantity kind = count.
• 'livestock' for using metric_dimension = product and quantity kind = count. Note: dimension for types of product could be endless and could include cars, ships, pizzas and so on, but livestock is likely most used in risk analysis; fall back on open nature of codelist for others.
• 'kg per hectare' for using metric_dimension = product and quantity kind = unit of production.
• 'cost' when using quantity kind = currency.

Codelist: impact_metric: add 'production_loss'.

As the metric / dimensions become more clear, is there a case for tying impact unit, dimension these together in one codelist?

[samroeslin]

By surface I meant Surface area. Used for example to describe crop losses, one might refer to the affected surface area of farmland.

In the current list of codes linked to impact_unit I see hectares, m2 Square meters, km2 Square kilometers.

[stufraser1]

By surface I meant Surface area. Used for example to describe crop losses, one might refer to the affected surface area of farmland.

In the current list of codes linked to impact_unit I see hectares, m2 Square meters, km2 Square kilometers.

Thank you - then this is covered in the quantity_kind codelist which includes: area, count, currency, length as option to describe the quantity.

[matamadio]

A loss dataset could include both the total exposure value and the cost of loss.
Would be good to indicate both in the metadata; the loss attributes would be under "cost", while the total value under the exposure section of the same dataset; in fact, we can allow both "Exposure" and "Loss" schemas within the same metadata.

In agreement with #288, suggesting these changes:

• Turn metric dimension into an array (multiple values allowed for one dataset)
• Quantity kind (array) to be joined with "unit" codelist > rename from impact_unit, extended from loss schema to exposure
  • Add more options to the "unit" codelist
  • If quantity = currency, then join the "Currency" codelist (bypassing "cost" in the exposure subschema)

[stufraser1]

Proposed additions from #292 (comment) have been made into 0.2 dev branch directly.

[stufraser1]

Diagram at https://docs.riskdatalibrary.org/en/latest/reference/schema/#loss-metadata needs to make clear the role of quantity_kind in defining loss metric (currently not shown)

[matamadio]

Add "Exposure to hazard" within impact metrics.
Semantycs are not optimal but is the best way to communite the output of an assessment that measures impact in terms of exposure to hazard.