MET Norway¶

The MET Norway integration discovers MEPS forecast runs, acquires selected variables, optionally interpolates them onto an existing Shyft DTSS geo database, and optionally produces a concatenated forecast. A forecast reference time identifies a run; valid time identifies a time inside that run’s forecast.

The pipeline is:

MET Norway
    |
catalog discovery/cache
    |
forecast acquisition
    |
optional spatial interpolation
    |
fc DTSS/geo database
    |
optional forecast concatenation
    |
cc DTSS/geo database

Responsibilities¶

shyft-integration-met-no is the installed, long-running foreground service. Its manager restores active versions and runs an agent for each logical configuration name. Agents reconcile the desired forecast-reference-time period with discovered forecasts and the stored forecast archive.

shyft.integration.met_no.Client is the Python configuration and operational administration interface to that already-running service. It stores/reads configuration versions, selects or clears persistent active versions, inspects runtime and acquisition status, and requests reconciliation. Constructing a client does not start the service; it exposes no process start/stop/restart API. See Configuration for the Python store/activate, switch, and passive configuration workflows.

Authority and storage¶

fc: The authoritative acquired forecast archive, in a pre-existing DTSS geo database. Its configured geometry is authoritative for destination geometry; the integration does not create or extend that geometry definition. Verified forecast presence in fc proves acquisition completion. Persistent acquisition state is bookkeeping, not the source of truth for forecast data.
cc: An optional, pre-existing derived concatenated forecast database. It is generated from verified forecasts already stored in fc. It must have a valid base/anchor t0; the integration neither creates nor automatically rebases it. See The cc anchor precondition.
MET service RocksDB: Local service/configuration metadata: model/configuration versions, active versions, acquisition bookkeeping, and the shared catalog cache. It does not replace the DTSS forecast archive and is not wholly disposable, since it also holds the configurations.

Start with Service to run the process, Configuration to prepare acquisition tasks, and Operations for deployment, status, and recovery. Examples and Python API links executable regression examples and the live Python API reference.