The use of small-scale process descriptions at larger scales has been identified as a major problem in hydrological modelling. An alternative is to use measurements taken at a larger scale, such as remote sensing data, and to develop process descriptions at that scale. A hypothetical catchment area, or virtual hydrological reality dataset has been generated to explore the value of different types of data in model structure and parameter identification. Though only an approximation to the natural responses of the catchment, this virtual reality can be used to explore the use of observational series that would not normally be available. A data-based mechanistic approach was used to model evapotranspiration from 1 km data, as though they were available from remote sensing. The resulting parameterization is of a similar form to an evaporative fraction model. It is an example of the type of parameterization that could be used to describe the land surface, without estimation of effective parameters from small-scale measurements, and avoiding overparameterization.