The data from all model grid points will be used where the model grid is exactly equal to half the interpolation grid length (Fig3.4.2-1).  


Fig3.4.2-1:  For a given interpolation grid (red circles) the proportion of model information taken into account depends on the model grid resolution (black crosses). When the interpolated grid is twice the model grid length (or less), all model grid values will be used in the interpolation.  Interpolation grid (red circles). Model grid resolution (black crosses).


When the model grid is less than half the interpolation grid length (i.e. the interpolated points are spaced out by more than twice the grid length), then the proportion of used grid points decreases.

Fig3.4.2-2:  An example of sub-grid selection where only every 3rd grid point is used (red circles).  In this case, if grid point data availability is at 0.25° resolution then the data is interpolated at 0.75° resolution.  Similarly, if data is available at 0.25º resolution and the requested interpolation grid length is 1.0º, 2.0º or 4.0º, only every 4th, 8th or 16th interpolated value is selected and disseminated.  Interpolation grid (red circles). Model grid resolution (black crosses).


The sub-grid scale selection and interpolation has the undesired effects that model grid point values which may essentially represent small scales, may by chance appear to represent much larger scales and interpolation will not conserve area totals.  In particular sub-grid scale sampling is  not optimal for use for surface fields, such as precipitation (Fig3.4.2-3).



Fig3.4.2-3:  Example of the effect of inappropriate interpolation of precipitation fields. To the left the forecast is interpolated in a 0.25 x 0.25° grid, to the right in a 3 x 3° grid where clearly there is a significant loss of useful detail, particularly near southwestern Norway.