The difference between observation TYPE and QUANTITY

Since DART is designed to assimilate data from any data source into any model, the assimilation algorithms need a way to define how observational data sources relate to model state variables.

DART does this by defining a single generic observation QUANTITY, such as zonal wind, and mapping many specific observation TYPEs, corresponding to source observations, to the single QUANTITY.

For example, QuikSCAT and radiosondes are both capable of measuring zonal wind. DART defines two observation TYPEs:

  • QKSWND_U_WIND_COMPONENT for the QuikSCAT observations of zonal wind

  • RADIOSONDE_U_WIND_COMPONENT for the radiosonde observations of zonal wind

and relates both of these TYPES to a single QUANTITY: QTY_U_WIND_COMPONENT.

Thus TYPE and QUANTITY have a many-to-one relationship. This distinction enables you to assimilate or evaluate observation platforms independently of one another with a single observation sequence file; reducing the possibility of error.

The forward observation operators are implemented based on observation QUANTITY. When requested, the model generates a QTY_U_WIND_COMPONENT, it doesn’t need to know that it will be compared to an observation from QuikSCAT or one from a radiosonde.


It is usually scientifically very interesting to be able to compare the assimilations one TYPE of observation verus another. An observation sequence file can have many types of observations. DART has the capability to assimilate (or evaluate) any combination of observation types without getting bogged down in dataset management. The same observation sequence can be used for experiments that include or exclude certain observation types. This procedure can ensure that you are actually performing the experiment that you think you are performing.