OPTIMAL GROUND-BASED SAMPLING FOR REMOTE-SENSING INVESTIGATIONS - ESTIMATING THE REGIONAL MEAN

Remote sensing investigations often involve sampling on the ground to estimate the mean of some property within ground resolution elements. Investigators have used classical statistics to determine the size of sample required to produce a desired precision. However, classical statistics is based on assumptions that do not hold when the target population is spatially dependent. Remotely sensed data and ground cover are usually spatially correlated, and in these circumstances the size of sample required will be less when sampling is done on a regular grid. This is demonstrated for several variables measured at the ground.