Using the \citet{Griffith2005} method of ESS estimating, this study analyzes the effect of random and purposeful sampling on auto-correlated population data for gridded population modeling. Mainly these sub-sampling methods depend on a combination of random, systematic, and geographically stratified sampling, and balance sampling methods . Finding of this study may provide feedback on the error produced by sampling for large data sets and effective sampling technique for gridded population modeling.