Observations from mesoscale networks are affected by the whole range of dynamic scales, including very small, fast-evolving scales and local events that cannot be adequately resolved even by high-resolution, convection-resolving models.
Such phenomena are real: they do not come from errors characterizing the instrumentation. They can be relevant or not, depending on the aims of the user.
Representativity errors should then be accounted for when comparing model fields with observations: they need to be considered as a part of the observational error when observations are compared with model fields for purposes such as analysis, assimilation, verification. The observational error has then two components: instrumental error and representativity error.
Conversely, it may be important to understand which real phenomena can be simulated by a model and which can not. In other words, if what matters are exactly those scales that can not be resolved by the model, it is necessary to change, improve or substitute, the model.
A particular and important component of representativity error arises from the difference between model orography and real orography.
A representativity study is in general necessary before addressing a study such as:

• data assimilation: objective analysis, definition of the initial condition, system trajectory reconstruction;
• forecast verification based on any kind of observations.

Works on representativity error:

• C. Lussana, F. Uboldi and M. R. Salvati, 2010. A spatial consistency test for surface observations from mesoscale meteorological networks. Quarterly Journal of the Royal Meteorological Society, 136, 1075-1088.   DOI: 10.1002/qj.622
• poster presentation at EMS 2013: F. Uboldi and C. Lussana, 2013. Representativity of a mesonet temperature observations with respect to model fields. Poster presentation at EMS 2013, 10-14 September, Reading UK.
• IN ITALIAN: Presentazione presso ARPA-Emilia Romagna, Bologna, "SimInAria", 5 Giugno 2012: F. Uboldi 2012. Confronto tra stime da campi di modello e osservazioni della rete regionale lombarda - classificazione delle stazioni per rappresentatività riferita a COSMO-I7 e ECMWF-0125.