On the use of auxiliary information to define the sampling design for large-scale geospatial data

In many fields of application it’s common to be interested in spatially-related phenomena and in particular to deal with attributes that, being defined on continuous spatial domains, are observed on a fine grid. It is well known that in order to investigate these phenomena through a sample survey it is more accurate to spread the sample over space and to exploit the rich auxiliary information available from various types of large-scale observations. Such information is often used in the estimation stage of a survey but it may be interesting to use it in the design stage as well. Therefore, we propose a two-step sampling design which investigates the relation between the auxiliary and study variables in order to identify when and how it is useful to exploit this information, in addition to the units’ spatial location, in the the second-step of the sampling selection process.