Study Of Adaptive Cluster Sampling Technique On Vegetation In Ulan Buh Desert

In this paper, a fitting study based on data of a fixed region was carried in adaptive cluster sampling. Nitraria tangtorum, which represents the typical vegetation in Ulan Buh Desert in Inner Mongolia, was studied. On the basis of adaptive cluster sampling study, a second-order adaptive cluster sampling method was proposed, combined adaptive cluster sampling with traditional sampling techniques. Adaptive standards, selection and determination of initial sample's proportion, were all studied by charts visual analyzing and mathematics theory verifying. And then, estimation accuracy and sampling efficiency were considered as the standard of determination of sampling design parameters, the interaction relation between sampling parameters were also discussed. Moreover, since Nitraria tangtorum in dune having special distribution pattern, a linear regression equation was made to describe the relationship between dune perimeter and area, based on fractal theory. The Nitraria tangtorum's area was estimated by using its perimeter, and adaptive cluster sampling was also analyzed. Study results show that:(1)By adjusting adaptive cluster sampling design parameters, sampling design parameters can be found, which in accordance with the estimation accuracy of sampling and sampling requirements of relative efficiency. When the adaptive standard was greater or equal to the average of pilot unit, adaptive cluster sampling is more effective than simple random sample; the relative efficiency and average estimation of overall sample increase with initial sample proportion. (2) The second-order adaptive cluster sampling used in this study area, can achieve corresponding estimation accuracy, and consistent with its estimation theoretical. (3)The estimation model between dune perimeter and area, which based on fractal theory, was y = 1.984 x - 1.157 .After independent sample testing, the model estimation accuracy is 89.8%. The estimation accuracy of adaptive cluster sampling is large than 80%, so its sampling efficiency is simple random sampling's five times.This article has provided a new approach, which was applied to vegetation inventory with uneven distribution in the Ulan Buh Desert in Inner Mongolia. This method can combine remote sensing technology and adaptive cluster sampling technique effectively, and it's feasible.