Quelle: Forstarchiv 87; 1, 24-30 (2016)
Autor(en): NOTHDURFT A, RITTER T
Abstract: Multi-purpose forest inventories are focused on a broad set of target variables. However, many of these target variables are derived from estimates of a single important attribute, namely the number of trees per area unit. Obtaining an accurate estimate of the tree density can therefore be considered as a crucial prerequisite in various forest inventory settings. Counting the number of trees in a forest seems to be trivial at first glance, but it can be challenging in fact (i) if an improper sampling approach is used, (ii) when trees are sparse, and (iii) if trees are masked by obstacles or limited sighting conditions. We will show different approaches (i) to correct the design-based biased k-tree sampling estimator, (ii) to efficiently sample rare trees, and (iii) to correct for the bias introduced by missing trees due to limited sighting conditions. The paper demonstrates statistical methodology how to answer the central question: How many trees are in the forest?
© DLV München