The grid values should not be considered as representing the weather conditions at the exact location of the grid point, but as a time-space average within a two- or three-dimensional grid box.  The discrepancy between the forecast grid-point value and the verifying observed average value can be both systematic and non-systematic:

• systematic errors reflect the limitations of the NWP model’s ability to simulate the physical and dynamic properties of the system;
• non-systematic errors reflect synoptic phase and intensity errors (as indicated by the left hand green arrow in Fig3.2.1).

Comparing NWP model output with point observations (as commonly happens in verification), systematic and non-systematic errors are introduced.  This is due to location of the model NWP output not being representative of the location, height and aspect of the observation, and also to sub-grid scale variability.

Fig3.2.1:  Ideally there should be a two step procedure to compare NWP model output and observations.

Firstly, compare grid point average to observation area average:

• systematic errors are due to model shortcomings.
• non-systematic errors stem from:
  • differences between NWP synoptic time and time of observation.
  • intensity errors.

Secondly, compare observation average and point observations: 

• systematic errors are due to how representative the station is for the area (e.g. location, height and aspect of the observation).
• non-systematic errors are due to sub-grid scale variability.

Fig3.2.2:  In reality, the comparison between NWP and observations must for simplicity bypass the area average stage.  This results in the systematic and non-systematic errors arising from distinctly different sources.  The effects related to the two green arrows in Fig3.2.1 are here combined into one.

Systematic errors due to model deficiencies and/or observational representativeness can be partly corrected by statistical means (e.g. Model output statistics MOS).  A series of forecasts will also help in dealing with uncertainty.

Non-systematic synoptic errors can be dampened by different ensemble approaches (e.g. ENS alone, ENS and HRES, probability considerations, forecast error growth).  However, sub-grid variability (notably for rainfall but other parameters too) can be addressed through downscaling.  Downscaling converts the grid box area average probability density functions from the raw ENS into "point rainfall probability density functions" for points within each grid box.

New downscaling techniques are being developed accordingly (see for example the Point Rainfall product).