Different Spatial Sampling Models in Geographical Analysis

Taking the estimation of area-percent about different land use in certain city as an example, we evaluated the effect of simple random sampling, stratified sampling based on administrator region, stratified sampling based on knowledge and stratified-systematic sampling based on spatial autocorrelation. Sample size and precision (standard deviation) were made decision criteria to estimate the effect. The results showed that: 1.) Similar estimations were acquired. 2.) Compared with simple random, sampling stratified sampling possessed higher precision and small sample size. 3.) Stratified sampling based on knowledge had higher precision than stratified sampling based on administrator region, through both had the same sample size. 4.) Stratified-systematic sampling based on spatial autocorrelation could reduce data redundancy evidently and the cost of investigate. Though the precision was lower than other models, the difference was not significant. Therefore, stratified-systematic sampling based on spatial autocorrelation was deserved to be recommended.