The aggregated forecast series uses the history data from the underlying series to make its own forecast on the summarized history data.
The aggregated forecast series are treated like any other Forecast series but with the difference that the historic data comes from an aggregation instead of being imported as one historic series. An aggregated forecast needs to be connected to weather stations, in similarity with Forecast series, and the different stations are weighted for the Aggregated Forecast series.
The child nodes are in this case only used to provide history data and thus the forecast can be turned off for those series.
When historical outcome data ends at different times for the sub-series, the old models (Cilia and Aiolos) set the end of the aggregated historical data as the latest end of data for all sub-series. In order to avoid using too low aggregated historical outcome data due to missing values in some sub-series, the quarantine has to be set to a sufficiently high number.
The newer models (Aiolos 2.0, Cilia 2.0, Auxiliary, and Zephyros) set on the contrary the end of the aggregated historical data as the earliest end of data for all sub-series. Also, series without any data-values are ignored.
If we want to override this behaviour for a sub series that perhaps is relatively small and has problems with the data availability, just set the ‘Enddate’ for this series to 1980-01-01. The program will then ignore this series when calculating the end date of the outcome data for the aggregated forecast series. If the ‘Enddate’ is set to 1970-01-01 then the sub series will be ignored with when calculating the end date of data as well as when aggregating the values, i.e. the series is then completely switched off.