An alternative plot design for adaptive cluster sampling

Trippstadt

2008

S. 81-87

3 Abb.,16 Lit. Ang.

Artikel aus einem BuchUnselbständiges Werk

200156514

Adaptive cluster sampling (ACS) is believed to be an efficient sampling technique for sampling rare and geographically clustered populations, but practical challenges do obviously prevent its wide application in forest inventories so far. A new and practical plot design, based upon "conditional plot extension", is thus proposed here as a practical simplification for ACS. Based on the Horwitz-Thompson estimator, its statistical performance was evaluated by simulation with both real data and artificial data of different spatial patterns. The first and very preliminary simulation results show that it exhibits some superiority only for geographically clustered populations. For the clustered populations, its performance depends not only on the spatial arrangements and sizes of clusters, but also the choice of appropriate design elements including plot extension factor and critical value. Further simulation studies are still ongoing for the assessment of this comprehensive statistical performances and design optimization.