Abstract:

Adaptive cluster sampling (ACS) appears to be an effective method for sampling rare and clustering population. The forest vegetation is most rare and clustering in west China. Based on the density of Hedysarum scoparium in Ulanbuh desert edge, four kinds of adaptive cluster sampling methods and two simple rand sampling methods had been carried out, there were simple random sampling with primary units selected with replacement, simple random sampling with primary units selected without replacement, adaptive cluster sampling based on Hansen-Hurwitz estimator with primary units selected with replacement, adaptive cluster sampling based on Hansen-Hurwitz estimator without primary units selected with replacement, adaptive cluster sampling based on Horvitz-Thompson estimator with primary units selected with replacement, adaptive cluster sampling based on Horvitz-Thompson estimator with primary units selected without replacement, and simulation resampling of six methods had also been conducted, the result of which had been compared. The result showed that the design of adaptive cluster sampling using Horvitz-Thompson estimator with initial sample was selected without replacement was more effective than the others, the mean estimator relative error of which was 0.037%, and the mean variance estimator was 0.03571. These results were propitious to increase the precision and efficiency for forest inventory.