[Proposal] Inclusion of observed/empirical/historical datasets

[matamadio]

What is the context or reason for the change?

It is not clear which kind of data is accepted by RDL in relation to empirical observations / historical state or events.

The hazard schema uses the calculation method attribute to define the type of event.

calc_method
Inferred
Simulated
Observed

So in case of Hazard, it can cover:

• layers produced with hazard models based on a set of observations or proxies, e.g. flood RP scenarios based on the distribution of extreme values from long-term historical data, or historical scenario modelled on the base of remote-sensing data collected during events ("Inferred")
• layers produced with hazard models based on general/global assumptions, and not on local data/observations ("Simulated")
• layer including the observations collected during an event, e.g. sampling of water depth at location points ("Observed")

What is your proposed change?

The same attribute could be (in concept) extended to:

• Exposure
  • layers derived from remote-sensing such as buildings, land use etc would be "Observed"
  • layers derived from remote-sensing proxies and statistics/indicators such as population_year maps would be "Inferred"
  • layers produced by modelling observervations and statistics within a theoretical framework such as future population scenarios or land use change would be "Simulated"
• Loss/impact/risk (related to GFDRR/rdl-standard#6 )
  • result of risk models (from any type of hazard/exposure data) would be "Simulated"
  • losses estimated from emprical data, e.g. empirical data provides number of houses affected, then the calculated economic damage based on costs assumption is "Inferred".
  • impacts measured in post-disaster phase would be "Observed" > here we overlap with EM-DAT and Desinventar, which is something we might avoid.

[pzwsk]

Moving to standard repo.

[matamadio]

Re-framing the issue for the whole schema; deleting previous comment

[matamadio]

This calc_method attribute has been included for H and V schema. For L schema, calc_method is used if probabilistic/simulated (as from V schema). Else if observed or inferred event, los_approach is used, as it refers specifically on how the losses are computed independently from V models.

los_approach: Overall approach on which the loss calculation is based (Empirical survey, Analytical, Judgement, Hybrid)

Included in v0.2, propose to review and close.